TRY Innsyn

Official Norway, the EU and world data in one MCP

400 tools for everything you need to ask public sources about, in Norway, the EU and worldwide, gathered in one MCP you can buy. 67 sources, and the AI chooses which ones to query.

See which tools and sources the AI would use

100
use cases, all run live
19
roles and sectors
67
sources in Norway, the EU and worldwide
400
tools the AI can use
See the scenarios15 with charts, 100 in the catalogue

The top 15

Scenarios from work and everyday life

Each use case is a scenario: a role, a specific question and an answer you see as charts. The limits on each page say what the data does not cover.

Tools and sources

What is behind the answers?

TRY Innsyn looks things up in 67 open sources through 400 tools. See what each source provides and which licence applies.

about 45,000
statistics tables
about 670
geographic layers
6
sources that need a key
59 of 67
sources where a licence was found

Open tools and sources →

The full catalogue

All 100 use cases

Filter by role, source or type. Open a use case to see the question, the steps and the limits. You can copy the questions.

Showing 100 of 100
UC-001The ageing wave in figures: 80+ growth and use of services in SteinkjerOwn pageStatistics Norwaycommunity planner in a municipality, Mid-sized municipality in Trøndelag (Steinkjer) · Municipalities and counties
WorkflowVerified 23 Oct 20263 tools
Open the scenario →

Sources Statistics Norway

Situation

The planner needs to back up the social element of the municipal master plan and the financial plan with the number of older people the municipality will have and what current use of services implies. The figures are in two different Statistics Norway (SSB) tables (population projections and KOSTRA) that have to be found and combined by hand.

The question

“How many residents over 80 will Steinkjer have in 2040 compared with today, what share of the over-80s use home-based services and institutional care today, and what does that mean for the number of users if use stays the same?”

How the AI does it

  1. Statistics Norway

    ssb_search_tables “framskrevet folkemengde” returns table 14746 (regional projections, municipal level, 2026–2050, 9 alternatives).

  2. Statistics Norway

    ssb_get_table_info 14746 returns the municipality code 5006 for Steinkjer, the statistics variable “Personer” (the main alternative MMMM) and the age grouping agg_Funksjonell5 (0-24, 25-34, 35-44, 45-69, 70-79, 80+).

  3. Statistics Norway

    ssb_get_data 14746 with Region 5006, Alder (code list agg_Funksjonell5), ContentsCode Personer and Tid 2026 and 2040.

  4. Statistics Norway

    ssb_search_tables “KOSTRA omsorg” returns table 12209 (selected key figures for health and care, municipality, 2015–2025). ssb_get_data 12209 with the region variable KOKkommuneregion0000 = 5006 and Tid set to the latest year.

  5. The AI multiplies today’s shares by 80+ in 2040 and shows the difference from today.

Result

Steinkjer goes from 1,666 to 2,362 residents over 80 (up 42%) between 2026 and 2040, while the 0–24-year-olds fall from 6,631 to 5,773 (down 13%). Today 32.6% of the over-80s use home-based services and 7.3% have an institutional stay; unchanged use means about 770 users of home-based services in 2040 (compared with about 540) and about 170 with an institutional stay (compared with about 120).

Why it matters

A sourced set of figures for the financial plan and the social element in minutes instead of days, with a clear line between the Statistics Norway projection and the municipality’s own assumptions.

Limits

The projection is the main alternative; Statistics Norway has eight other alternatives that show the uncertainty. An unchanged rate of use is a simplification, and the KOSTRA share applies to 2025. The figures say nothing about capacity, staffing or health status. The KOSTRA table uses a different region variable from the other Statistics Norway tables.

Evidence from the run

Run 2026-10-23: table 14746 returned Steinkjer 80+ = 1,666 (2026) and 2,362 (2040); table 12209 returned 32.6% of the over-80s with home-based services and 7.3% with an institutional stay (2025).

UC-002How did other municipalities run the same procurement? Weighting and bidsDoffinprocurement adviser in a municipality, Mid-sized municipality in a procurement partnership · Municipalities and counties
WorkflowVerified 23 Oct 20262 tools

Sources Doffin

Situation

The municipality is about to tender a framework agreement for software licences and wants to see how others have set it up: cooperation, weighting of price against quality, number of bids and time taken. There is no single place to compare.

The question

“Find the largest recently awarded IT framework agreements from municipalities in Doffin. For the licence agreement for Indigo IKT, Digitale Gardermoen and Bodø: how many buyers, how many bids, how was it weighted and how long did it take?”

How the AI does it

  1. Doffin

    doffin_search_notices with sok konsulent, type resultat and sortering hoyest_verdi finds the large agreements. The licence agreement (2026-106826) is number three.

  2. Doffin

    doffin_get_notice 2026-106826 returns buyers, bids, winner, weighting and timeline.

  3. The same call on a central government agreement (the Norwegian Environment Agency, 2026-115663) returns a weighting to compare with.

  4. The AI makes a comparison table and a proposed weighting and cooperation model.

Result

The licence agreement had 23 buyers (municipalities, inter-municipal companies and a municipal joint body), a ceiling of NOK 1 billion, only two bids and Crayon AS as the winner. The weighting was 65% price, 30% quality and 5% compliance, and it took 69 days from the notice (6 Feb) to the award (16 Apr). The Norwegian Environment Agency weighted price at 25%.

Why it matters

The municipality can choose weighting and form of cooperation based on experience, and sees that a large procurement partnership got few bidders, which is a signal about the market and about how the contract should be divided.

Limits

The notice does not give bid prices, evaluation notes or complaints. The ceiling is not what is spent, and the weighting of the criteria can be missing from some notices. Only advertised procurements are included, not direct purchases below the thresholds for advertising.

Evidence from the run

Run 2026-10-23: doffin_get_notice 2026-106826 returned 23 buyers, antallTilbud 2, winner CRAYON AS, weights 65/30/5, ceiling 1,000,000,000 NOK and timeline 2026-02-06 to 2026-04-16.

UC-003Control committee: the municipality’s companies, their debt and purchasesEnhetsregisteretRegnskapsregisteretsecretary to a control committee, Large city municipality · Municipalities and counties
WorkflowVerified 23 Oct 20262 tools

Sources Enhetsregisteret, Regnskapsregisteret

Situation

The control committee wants to look at the municipality’s inter-municipal companies: how large they are and how much debt they have. The companies have to be found one by one, and the accounts are in a separate register.

The question

“Find inter-municipal companies (IKS) and municipal enterprises (KF) in Bergen. What do the 2025 accounts of Grieghallen IKS show: revenue, debt and equity?”

How the AI does it

  1. Enhetsregisteret

    brreg_search_enheter with organisasjonsform IKS,KF and kommunenummer 4601 (Bergen): 3 hits, all IKS.

  2. Regnskapsregisteret

    regnskap_get_accounts for Grieghallen IKS (912157288): revenue, debt and equity for two years.

  3. Regnskapsregisteret

    regnskap_get_accounts for the municipality itself returns NOT_FOUND (tested on Oslo kommune, 958935420): municipalities do not report to this register.

Result

Grieghallen IKS had NOK 21.8 million in operating revenue in 2025, NOK 188.6 million in long-term debt and only 1.1% equity (NOK 2.2 million of NOK 192.5 million). The net result is 0 because interest takes the whole operating profit.

Why it matters

The committee gets a first overview of the companies and quickly sees which ones are heavily debt-financed, before it asks for documents and notes.

Limits

The municipality itself returns NOT_FOUND, and KFs are part of the municipality (the municipal accounts), so they are not expected here. Ownership shares, guarantee liabilities and notes are not available. The municipality filter matches on address, not on owner municipality, so companies with an address in another municipality are missing. The municipality’s own purchases are no longer covered by this use case.

Evidence from the run

Run 2026-10-23: brreg_search_enheter IKS,KF kommunenummer 4601 returned totalElements 3 (Bergen Havn Farvannsforvaltning IKS, Grieghallen IKS, Interkommunalt Arkiv i Hordaland IKS). regnskap_get_accounts 912157288 returned operating revenue 21,801,428 NOK, sumEiendeler 192,467,755 and equity ratio 1.1%; the same tool for 958935420 (Oslo kommune) returned NOT_FOUND.

UC-004Cycle traffic: hourly peaks and weekdays versus weekends in TrondheimTrafikkdataadviser for cycling and walking, County authority (roads and transport) · Municipalities and counties
WorkflowVerified 23 Oct 20263 tools

Sources Trafikkdata

Situation

The county authority is considering improving a cycle route at Hangarbrua and wants to see when during the day and week people cycle there, and which direction the peaks go in. The figures are in the traffic data of the Norwegian Public Roads Administration, but have to be fetched station by station.

The question

“Which cycle counters do we have in Trøndelag, and how is cycling spread over the day and between weekdays and weekends at Hangarbrua sør in Trondheim?”

How the AI does it

  1. Trafikkdata

    trafikkdata_find_stations(fylkenummer [50], type “sykkel”) returns 17 cycle counters in Trøndelag (8 shown), at least three in Trondheim, among them “Hangarbrua sør - sykkel” (id 74148B4018506).

  2. Trafikkdata

    trafikkdata_get_station(74148B4018506) shows the direction (Lade/Leangen), that data exists from 2026-05-20 and that annual average daily traffic (AADT) does not exist yet.

  3. Trafikkdata

    trafikkdata_get_volume(intervall “time”, 2026-10-21) returns hourly counts per direction.

  4. Trafikkdata

    trafikkdata_get_volume(intervall “dogn”, 2026-10-10 to 2026-10-22) returns daily counts per direction for 13 days.

Result

Hourly profile and weekday/weekend profile for the counting point: on Wednesday 21 October, 240 bicycles with peaks at 07–08 (28, most towards Lade) and at 16–17 (30, most towards Leangen); weekdays average about 234 and weekend days about 89 per day (2.6 times).

Why it matters

A basis for priorities and sizing (winter maintenance, lighting, width) without manual counting. The same query can be repeated for all 17 counters.

Limits

One counter per point, with no link to weather and no speed. The station is new (first data 2026-05-20), so AADT and a seasonal profile do not exist yet. There is no search by position, only by name and county. The last hour and day may be incomplete.

Evidence from the run

2026-10-23: 17 cycle counters in Trøndelag; Hangarbrua sør: 240 bicycles 2026-10-21 (28 at 07–08 and 30 at 16–17), 2,458 bicycles 10–22 October with coverage of 99–100%, the AADT list empty.

UC-005Farmland protection and heritage sites: check a plot before rezoningGeonorgeRiksantikvarenplanner and case officer in a municipality, Municipality on Jæren (Klepp used as an example) · Municipalities and counties
WorkflowVerified 23 Oct 20263 tools

Sources Geonorge, Riksantikvaren

Situation

A developer wants to convert cultivated land to housing. The planner wants soil quality and automatically protected heritage sites nearby gathered for the case report. The sources are held by NIBIO and the Directorate for Cultural Heritage (Riksantikvaren).

The question

“At 58.768 N, 5.634 E in Klepp someone wants to convert cultivated land. What does the soil map say about soil quality, and which automatically protected heritage sites are within one kilometre?”

How the AI does it

  1. Geonorge

    geonorge_search_datasets(sok “jordsmonn”) returns NIBIO’s Jordsmonn (0b960bb3-d9af-499e-a204-72833a4bcaaa, WMS), Jordressurser (3690c283-a1f2-4e95-8cd4-2d6115eb28f7, WMS, not published yet) and Erosjonsrisiko (0cb28e90-7ae9-4201-9303-ac6db0edfc6f, with no queryable service).

  2. Geonorge

    geonorge_query_dataset(Jordsmonn, punkt) returns the properties of the polygon at the point. Use `lag` to choose the theme.

  3. Riksantikvaren

    riksantikvaren_search_heritage_sites(lat, lon, radiusKm 1, vernetype “Automatisk fredet”) returns automatically protected heritage sites, nearest first.

Result

The point is in a polygon of organic soil, mapped in 1988 (detailed methodology), 182 daa, class 1 “best egnet klima og velegnet tekstur” for carrots. There are 10 automatically protected heritage sites within 1 km, the nearest a burial mound at 723 m and Klepp middelalderske kirkegård at 826 m.

Why it matters

A quick factual basis for the case report on farmland protection and heritage considerations, and it shows early on whether the heritage authority needs to be brought in.

Limits

The Jordsmonn WMS responds with many crop layers (suitability for carrots, parsley root and so on), so use `lag`. The mapping is from 1988 and says nothing about current farming. Riksantikvaren gives a centre point, not an outline, and only known sites: unknown ones can turn up during archaeological surveys. No zoning plans.

Evidence from the run

2026-10-23: Jordsmonn in Klepp: organic soil, class 1, 182 daa, mapped 1988; Riksantikvaren: 10 automatically protected heritage sites within 1 km, nearest 723 m (burial mound).

UC-006Coastal zone plan for Frøya: sites, warnings and sea conditionsDirectorate of FisheriesNorwegian Coastal AdministrationMET Norwayland-use planner, Coastal municipality in Trøndelag · Municipalities and counties
WorkflowVerified 23 Oct 20263 tools

Sources Directorate of Fisheries, Norwegian Coastal Administration, MET Norway

Situation

The planner is revising the coastal zone plan and wants to see the aquaculture sites with their traffic light status together with navigation warnings and sea conditions in the area. The sources are held by three different agencies.

The question

“Which aquaculture sites are in Frøya, what traffic light colour does the production area have, and which coastal warnings and wave conditions apply at Mannbruholmen?”

How the AI does it

  1. Directorate of Fisheries

    fiskeridir_search_aquaculture_sites with kommunenummer 5014: site, capacity (TN), species, position and the production area’s traffic light (prodAreaStatus).

  2. Norwegian Coastal Administration

    kystverket_get_coastal_warnings with lat 63.847, lon 8.537 and radiusKm 100: nearby warnings sorted by distance, plus warnings without a position.

  3. MET Norway

    met_get_ocean_forecast at the site: waves, sea temperature and current.

  4. Ask the AI to prepare material for a thematic map with sites, warnings and conditions.

Result

A list of sites in the municipality with capacity and traffic light, the nearest coastal warnings with distance, and today’s sea conditions.

Why it matters

The plan gets one shared factual basis instead of three separate requests.

Limits

The register does not give the total number of hits (page through), and the site boundaries only come with medGrenser. The traffic light applies to the whole production area, not the site. The coastal warnings are not official. No land-use plan decisions, fishing areas or protected areas.

Evidence from the run

Run 2026-10-23: Mannbruholmen (10232, Frøya, 7,020 TN salmon) is in production area 6 Nordmøre og Sør-Trøndelag with a yellow traffic light. Coastal warning 371/2026 (“Båeflua lanterne havarert”) is 76.5 km away, one warning has no position, and the waves were 1.4–1.8 m with the sea at 11.7 °C.

UC-007The Pupil Survey for all the municipality’s schools versus the countryUdirhead of education / school owner, Mid-sized municipality (Bodø) · Municipalities and counties
WorkflowVerified 23 Oct 20264 tools

Sources Udir

Situation

The head of education is taking grade 9 bullying figures to the politicians and must compare each school with the municipality and country. The figures are in Statistikkbanken at the Norwegian Directorate for Education and Training (Udir), where schools are looked up by organisation number and small schools have hidden figures.

The question

“What share of grade 9 pupils in Bodø say they are bullied in the Pupil Survey (Elevundersøkelsen) 2025-26? Compare the schools with the municipality and the country, and tell me which schools have hidden figures.”

How the AI does it

  1. Udir

    udir_stat_list_tables shows eight open tables, all from the Pupil Survey. Table 150 (EUG – Mobbing) covers primary and lower secondary school, 153 upper secondary school.

  2. Udir

    udir_stat_get_table_info (table 150) shows the dimensions Skoleår (TidID), Enhet (national, county, municipality, school), Trinn, Eierform, Kjønn and Spørsmål; question -50 is the index “Mobbing på skolen”.

  3. Udir

    udir_sok_enheter (register nsr, kommunenummer 1804, preferably kategoriId 1 for primary and lower secondary school) returns the schools with organisation numbers.

  4. Udir

    udir_stat_get_data (table 150) with TidID 202512, Kommunekode 1804 and TrinnKode 9 returns the municipality and all the schools in it; narrow it down with SpoersmaalID -50, KjoennID -10 and EierformID -10, and fetch the national figure with EnhetID -12.

  5. The AI makes a ranked table and points out schools where the value is null (not published).

IDs used: Statistikkbanken table 150, TidID 202512, municipality 1804.

Result

A table of all the schools with the share bullied, the municipality’s and the country’s figures, and a list of schools where the figure is hidden.

Why it matters

The material for the political case is ready in minutes, and the school owner sees which schools are above the national average.

Limits

Only the Pupil Survey is open in Statistikkbanken: no lower secondary grade points, national tests, grades or absence. Figures for schools with few responses are hidden (null), and the survey is self-reported. According to Udir, Statistikkbanken is not meant for external use and can change without notice. Without a category filter, the school register also returns private “schools” such as gyms.

Evidence from the run

Run 2026-10-23: grade 9, 2025-26, the index “Mobbing på skolen”: Whole country 10.2%, Bodø 8.8%, and three schools with 7.7%, 9.2% and 11.5%; small schools have null (hidden). udir_sok_enheter returned 53 units in municipality 1804.

UC-008Norway versus the Nordics and the EU: share of 30–34-year-olds with higher educationEurostatsenior adviser in a ministry (analysis and EU/EEA benchmarking), Ministry or directorate · Central government and agencies
WorkflowVerified 23 Oct 20262 tools

Sources Eurostat

Situation

The department is writing a memo on higher education and needs a comparison with the Nordic countries and the EU. Eurostat has tens of thousands of tables, and it is cumbersome to find the right dataset and the right codes for country and sex.

The question

“What share of 30–34-year-olds have higher education in Norway, Sweden, Denmark, Finland and the EU, and what is the gap between women and men in Norway?”

How the AI does it

  1. Eurostat

    eurostat_search_datasets “tertiary educational attainment” returns, among others, yth_educ_020 (30–34 years by sex, 2000–2025), tgs00105 (25–34 years, NUTS 1) and tgs00109 (25–64 years, NUTS 2).

  2. Eurostat

    eurostat_get_data yth_educ_020 with geo NO, SE, DK, FI and EU27_2020 and sisteTidsperioder 3. All values for sex (T, M, F) are returned because the dimension is not specified, and “forklaring” in the response confirms the codes.

  3. Eurostat

    eurostat_get_dataset_info can be used first to see the dimensions; here the codes were given directly.

Result

In 2025, 54.7% of 30–34-year-olds in Norway had higher education (down from 56.6% in 2023), compared with 56.4% in Sweden, 56.3% in Denmark, 41.6% in Finland and 45.6% in the EU27 (up from 43.8%). Norwegian women are at 63.3% and men at 46.4%, a gap of 16.9 percentage points.

Why it matters

The basis for comparison, with source and a common definition (ISCED 5–8), is ready in minutes, and the trend is easy to miss: Norway is falling while the EU is rising and Denmark is getting closer.

Limits

The age group is 30–34 years; the EU’s own education targets often use 25–34 years, so check the target’s definition. The figures are based on sample surveys and have uncertainty. Field of study, quality and effect on employment are not included.

Evidence from the run

Run 2026-10-23: Eurostat yth_educ_020 (updated 2026-09-10) 2025: Norway 54.7%, Sweden 56.4, Denmark 56.3, Finland 41.6, EU27 45.6; Norway women 63.3, men 46.4.

UC-009Follow up request resolutions: has the white paper been announced?Stortingetadviser in the political leadership’s staff, Ministry · Central government and agencies
WorkflowVerified 23 Oct 20263 tools

Sources Stortinget

Situation

The ministry needs an overview of what Stortinget, the Norwegian parliament, has asked the government to do and whether the follow-up has been announced. The request resolutions and the government’s list of cases are in different lists.

The question

“Has Stortinget passed any request resolutions on nuclear power in 2025-2026, what do they say, and is a related white paper on the list of announced cases?”

How the AI does it

  1. Stortinget

    stortinget_hent_vedtak with sesjonid 2025-2026, type ANMOD and sok “kjernekraft” returns resolution no. 783 of 2026-05-26 with text and link.

  2. Stortinget

    stortinget_hent_sak with sakid 200121 shows the proceedings and the short resolution.

  3. Stortinget

    stortinget_hent_ventede_saker (all 33) shows whether a white paper has been announced. No case on nuclear power was on the list.

IDs used: resolution 40036570 (no. 783), case 200121.

Result

A list of request resolutions with date and case, and an answer on whether the follow-up has been announced.

Why it matters

The political leadership gets early warning that a resolution lacks visible follow-up and can see which ministry owns it.

Limits

The list of expected cases only covers the coming months, so a case can come later without being on it now. The resolution data has no deadline and does not say which ministry has been given the task.

Evidence from the run

Run 2026-10-23: request resolution no. 783 (2026-05-26, case 200121): “Stortinget ber regjeringen legge frem en stortingsmelding om kjernekraft.” The 33 announced cases (October–December 2026) contained no white paper on nuclear power; Energidepartementet had only “Endringer i statsbudsjettet 2026”.

UC-010Protected area: regulations, management authority and restrictions before a dispensationNorwegian Environment Agencycase officer at the County Governor (Statsforvalteren), County Governor (environmental department) · Central government and agencies
WorkflowVerified 23 Oct 20262 tools

Sources Norwegian Environment Agency

Situation

An application for a dispensation concerns a measure in a protected area. The case officer wants to see quickly the form of protection, date of protection, regulations, municipalities and who the management authority is, and also restrictions such as landing bans and low flying.

The question

“What is Hardangervidda nasjonalpark protected as, since when and under which regulations, which municipalities and management authorities apply, and which access restrictions apply in the protected areas around Orrevatnet on Jæren?”

How the AI does it

  1. Norwegian Environment Agency

    miljodir_query_layer(lag “vern/0”, sok “hardangervidda”) returns the form of protection, date of protection, link to the regulations, municipalities, management authority and fact sheet.

  2. Norwegian Environment Agency

    miljodir_check_location(lat 58.72, lon 5.54, radius 2000) shows protected areas and restriction areas (landing ban, low flying below 300 m) for Orrevatnet and Jærstrendene.

Result

Hardangervidda nasjonalpark: protected 1981-04-10, regulations on Lovdata (1981-04-10-4830), six municipalities (Vinje, Tinn, Hol, Eidfjord, Nore og Uvdal, Ullensvang), IUCN II, managed by several County Governor offices and supervisory committees. On Jæren there is a landing ban and low flying is prohibited (below 300 m) in Orrevatnet naturreservat and Jærstrendene, with an exception for approaches to Sola airport.

Why it matters

The same basic data that the case officer would otherwise have to look up in Naturbase and on Lovdata, gathered with links to fact sheets.

Limits

The map layer is generalised (about 100 m), so boundary cases must be checked against the map in the regulations. The text of the regulations is read on Lovdata (the link is included). The answer does not decide the dispensation, and other land-use rules, such as the building ban in the shore zone, are not layers here.

Evidence from the run

2026-10-23: vern/0 “Hardangervidda nasjonalpark”, date of protection 1981-04-10, 6 municipalities, IUCN II; miljodir_check_location at Orre: 6 protected areas and 5 restriction hits (landing ban/low flying).

UC-011Weekly change log for kindergartens and schools in the countyUdirEnhetsregisteretadviser in the supervision department at the County Governor (Statsforvalteren), County Governor (public office) · Central government and agencies
MonitoringVerified 23 Oct 20263 tools

Sources Udir, Enhetsregisteret

Situation

The County Governor supervises kindergartens and schools and wants to know about new, changed and closed units in the county without searching registers. The changes are in registers at the Norwegian Directorate for Education and Training (Udir), but there are no alerts.

The question

“Which kindergartens and schools in Nordland have changed in the registers since Monday? Tell me if any have closed, and who owns them.”

How the AI does it

  1. Udir

    Step 1 (every Monday): udir_enheter_endret_etter with register nbr and the date of the previous Monday, then register nsr. The fields totaltAntall and avkortet say whether the list is complete.

  2. Filter on Fylkesnummer 18 and ErAktiv false for closed units; ErBarnehage and ErBarnehageeier separate a unit from an owner.

  3. UdirEnhetsregisteret

    For schools, udir_get_school_profile returns pupil numbers, ownership structure, supervision structure and OppstartsEllerStiftelsesdato, which separates new units from changed ones. udir_get_registre_for_enhet shows which registers an organisation number is in. For kindergartens, udir_get_enhet is used with register nbr (not run in this verification).

  4. The AI sends a short digest by email or Slack. Runs as a scheduled agent.

IDs used: the registers nsr and nbr, county number 18, the organisation numbers of a school and of Oslo kommune.

Result

A weekly list of new, changed and closed kindergartens and schools in the county, with owner and municipality.

Why it matters

The supervisors pick up closures and new establishments faster and avoid going through the registers by hand.

Limits

The change list does not distinguish new from changed, so the start date has to be checked for each unit. Bulk updates create noise (many schools in the Bodø extract have the change date 13 October). The school register also contains private businesses such as fitness centres. The lists have neither address nor coordinates. Use only at organisation level; owners that are sole proprietorships must not be mapped.

Evidence from the run

Run 2026-10-23: udir_enheter_endret_etter returned 33 changed kindergarten units since 2026-10-16 (among them Hoppensprett Nordland AS, kindergarten owner in municipality 1866, with ErAktiv false) and 652 changed school units since 2026-10-14. udir_get_school_profile (a school in Bodø) shows Statsforvalteren i Nordland as supervision structure and start date 1969-02-01.

UC-012Has the basic amount kept up with prices? G, CPI and the lump-sum grantOwn pageNAVStatistics Norwayadviser in an employee organisation (public affairs), Trade union that writes consultation responses on the state budget · Politics and public affairs
WorkflowVerified 23 Oct 20264 tools
Open the scenario →

Sources NAV, Statistics Norway

Situation

The organisation is writing input on benefits calculated in G, the basic amount (the 6G cap, sickness benefit, unemployment benefit), and on the lump-sum grant at birth. They need G amounts back in time, set against price growth, and do not want to type in figures from several websites.

The question

“What are 1G, 2G and 6G today, how much has G risen since 2015 compared with consumer prices, and what has happened to the lump-sum grant since 2023?”

How the AI does it

  1. NAV

    nav_get_basic_amount returns G today (valid from 1 May 2026).

  2. NAV

    nav_get_basic_amount_history with fra=2015-01-01 returns G for every year since 1 May 2015 (12 values with conversion factor).

  3. NAV

    nav_get_one_off_grant_history with fra=2016-01-01 returns the rates for the lump-sum grant.

  4. Statistics Norway

    ssb_get_data table 14710 (CPI, 2025=100, monthly back to 1920) with Tid 2015M05, 2016M05, 2025M05 and 2026M05.

  5. The AI calculates 2G and 6G and the percentage growth.

Result

G is NOK 136,549 from 1 May 2026, so 2G is NOK 273,098 and 6G is NOK 819,294. Since May 2015 (NOK 90,068) G is up 51.6% while CPI is up 42.1% (72.5 → 103.0): G has grown about 7% more than prices. In the last G year, G rose 4.9% against 3.1% for CPI. The lump-sum grant has been unchanged at NOK 92,648 since 1 January 2023, while G has risen 15.1% since May 2023 (118,620 → 136,549).

Why it matters

A calculation with official figures and sources for the consultation response in minutes, and a concrete point (the lump-sum grant stands still while G rises) that journalists and politicians can check.

Limits

The service gives the amounts and rates from NAV, the Norwegian Labour and Welfare Administration, not the rules: which benefits are calculated from how many G (6G, 2G etc.) must be checked against the act and regulations. The lump-sum grant does not follow G, so the comparison is an illustration, not a rule. CPI is an average for all households.

Evidence from the run

Run 2026-10-23: G = NOK 136,549 from 2026-05-01 (NOK 90,068 from 2015-05-01); lump-sum grant NOK 92,648 from 2023-01-01 (tilOgMed null); Statistics Norway 14710 CPI 72.5 (2015M05) and 103.0 (2026M05).

UC-013How did each party vote? Nuclear power from topic to vote countOwn pageStortingetpublic affairs adviser in an industry association, Industry association in the energy sector · Politics and public affairs
WorkflowVerified 23 Oct 20264 tools
Open the scenario →

Sources Stortinget

Situation

The adviser has to tell the board who actually voted for and against nuclear power in Stortinget, the Norwegian parliament. Today that means searching case documents and voting records on stortinget.no, and the voting is spread over ten votes in the same case.

The question

“Find Stortinget’s handling of the representative proposal on nuclear power in spring 2026. How did each party vote in the votes, and what did Stortinget end up adopting? Also show how each Høyre member voted on the proposal from FrP, H and V.”

How the AI does it

  1. Stortinget

    stortinget_hent_saker with sesjonid 2025-2026 and sok “kjernekraft” returns case 200121 (Dokument 8:191 S, status considered, Energi- og miljøkomiteen).

  2. Stortinget

    stortinget_hent_sak with sakid 200121 returns the proceedings (submitted 2026-03-19, committee recommendation 2026-05-12, debate 2026-05-21, vote 2026-05-26) and the short resolution.

  3. Stortinget

    stortinget_hent_voteringsresultat_for_sak with sakid 200121 and parti H lists all ten votes and each Høyre member’s vote on the first.

  4. Stortinget

    stortinget_hent_voteringsresultat with voteringid 27722 and medRepresentanter false returns the vote count per party for proposals no. 1 and 2 from FrP, H and V.

IDs used: case 200121, votes 27714–27723.

Result

An overview of the ten votes with vote counts per party, named votes for one party and what Stortinget adopted.

Why it matters

The board gets the correct case history and correct vote counts with case and vote IDs to refer to, without the adviser having to put them together by hand.

Limits

Only roll-call votes have votes per member. Around 69 members were absent at each vote, so “not present” says nothing about position. Stortinget does not publish reasons for the votes; they have to be taken from the debate. Voting data exists from the 2011-2012 session.

Evidence from the run

Run 2026-10-23: case 200121 had 10 votes on 2026-05-26. Proposals no. 1 and 2 from FrP, H and V fell 49–51 (for: FrP 29, H 14, KrF 4, V 2; against: A 31, MDG 5, R 5, Sp 5, SV 5). The committee’s recommendation was adopted 67–33 and resulted in the resolution “Stortinget ber regjeringen legge frem en stortingsmelding om kjernekraft”.

UC-014What is coming from the government? Weekly overview of announced casesStortingetsenior public affairs adviser, Public affairs agency with clients in several industries · Politics and public affairs
MonitoringVerified 23 Oct 20262 tools

Sources Stortinget

Situation

The agency needs to know which propositions and white papers are coming before the clients face consultation deadlines and consideration in Stortinget, the Norwegian parliament. The list of announced cases is spread across the ministries, and Stortinget publishes only today’s list.

The question

“Which cases has the government announced it will put before Stortinget this autumn? Group them by ministry and month, and point out the ones that concern housing, tech and defence.”

How the AI does it

  1. Stortinget

    stortinget_hent_ventede_saker with limit 40 returns all announced cases with title, type, ministry and expected month (33 in total).

  2. Stortinget

    stortinget_hent_regjering links each ministry to its minister, so that letters can be addressed correctly.

  3. Let the AI group the cases by the client’s topics. Run weekly and compare with the previous week, because Stortinget does not store history.

Result

A case list for each client with month and responsible minister, and a marker for what is new since the last run.

Why it matters

Early warning gives time to contact the ministry and prepare input before the case is a proposition with deadlines.

Limits

The list gives title, ministry and expected month, not content. Stortinget only publishes today’s list, so you have to store the history yourself. The month field is sometimes a date (one case had 08.12.26), and the government can move the cases.

Evidence from the run

Run 2026-10-23: 33 announced cases for October–December 2026, among them “Lov om aldersgrense for bruk av sosiale medier” (Barne- og familiedepartementet, December), “Endringer i husleieloven og ny boligleielov” (Kommunal- og distriktsdepartementet, December) and “Lov om beskyttelse av norsk forsvarsteknologi” (Forsvarsdepartementet, December). The government has 20 members, all from Ap.

UC-015Who has submitted input? Map the hearing on Helsepersonellplan 2040Stortingetpolicy adviser, Patient or user organisation · Politics and public affairs
WorkflowVerified 23 Oct 20263 tools

Sources Stortinget

Situation

The organisation wants to see who else has submitted input and who got to speak at the open hearing, to find allies and opposing voices. The input and the programme are on different pages on stortinget.no.

The question

“Which organisations have submitted written input to Stortinget’s health and care committee hearing on Helsepersonellplan 2040, and who got to appear at the open hearing on 28 September 2026?”

How the AI does it

  1. Stortinget

    stortinget_hent_horinger with sesjonid 2025-2026 and komite “Helse” finds hearing 10005918 (Meld. St. 11 (2025–2026) Helsepersonellplan 2040, input deadline 2026-09-23).

  2. Stortinget

    stortinget_hent_horingsinnspill with horingid 10005918 lists submitter, date and title (62 submissions in total). With medTekst true the content can be read.

  3. Stortinget

    stortinget_hent_horingsprogram with the same id returns the schedule for the open hearing with the order of organisations and question rounds.

IDs used: hearing 10005918, case 200350.

Result

A list of submitters and hearing participants grouped by type (trade unions, industry associations, patient organisations), and the text of the most relevant submissions.

Why it matters

Mapping of competitors and allies in minutes. The organisation sees which arguments have already been made and can apply for a place at the next hearing.

Limits

Only approved submissions are published, and submitters appear as Stortinget publishes them. The programme also contains names of individuals who represent organisations; use the organisations, not the people. The ministries’ own consultations are not here.

Evidence from the run

Run 2026-10-23: hearing 10005918 had 62 approved submissions, the eight newest from, among others, Melanor, Standard Norge, Fagforbundet, Vernepleierforbundet, Norges Blindeforbund, Autismeforeningen i Norge, Virke and Farmasiforbundet. The programme for 2026-09-28 had 112 items from 08.30, including Kreftforeningen, LHL and Diabetesforbundet.

UC-016The client’s field in Stortinget: follow the questions to the ministerStortingetcommunications adviser (crisis management), Communications agency that supports health trusts and organisations · Politics and public affairs
MonitoringVerified 23 Oct 20263 tools

Sources Stortinget

Situation

When Stortinget, the Norwegian parliament, asks the minister about the client’s field, the client needs messages and background ready before the answer comes. Today the agency follows stortinget.no by hand, and answers are often first discovered in the media.

The question

“Which written questions has the health minister received since 1 October, who asked them, and which are unanswered? Show the minister’s answer to a recently answered question, so we can predict the line of response.”

How the AI does it

  1. Stortinget

    stortinget_hent_regjering: the minister of health and care services has id JANVES.

  2. Stortinget

    stortinget_hent_skriftlige_sporsmal with sesjonid 2026-2027, til “helse” and status “alle” lists questions with asker, party, status and dates (51 so far in the session).

  3. Stortinget

    stortinget_hent_enkeltsporsmal with an id from the list returns the question and the answer in full text.

  4. Run daily, filter with sok on the client’s topic and let the AI summarise the line of response.

IDs used: question 127799 (answered), 129879, 129880 and 129856 (unanswered).

Result

A daily list of new and answered questions to the minister on the client’s topic, with asker, party and dates, and a short summary of how the minister usually answers.

Why it matters

The client gets a head start in crisis situations and can prepare talking points before the answer is public. The dates also show how long the minister takes to answer.

Limits

The answer is only in the single-question lookup, not in the lists. The tools cover written questions, question time and interpellations, not enquiries directly to the ministry. The question title in the lists is only the start of the question text.

Evidence from the run

Run 2026-10-23: 51 written questions to the minister of health and care services in 2026-2027, three of the four newest unanswered. Question 127799 (on drinking water in the municipalities’ emergency plans, asked by Høyre) was sent 2026-10-19 and answered 2026-10-22 with reference to drikkevannsforskriften and NOK 6 million extra for the Norwegian Food Safety Authority in 2026.

UC-017Follow an EU case: the AI Act from proposal to the Official JournalEuropean ParliamentEUR-LexEU adviser, Brussels office of a Norwegian business organisation · Politics and public affairs
WorkflowVerified 23 Oct 20264 tools

Sources European Parliament, EUR-Lex

Situation

The office has to explain to members where an EU case stands, who led it in the European Parliament and whether it has been amended since. The information is in the Legislative Observatory, the MEP pages and EUR-Lex, and is hard to put together.

The question

“Show the timeline for the AI Act in the European Parliament (procedure 2021/0106(COD)): who were the rapporteurs, when were the plenary votes, and when was it published? Has it been amended since it was adopted?”

How the AI does it

  1. European Parliament

    europarl_get_procedure with prosedyre 2021/0106(COD) returns the title, lead committee (IMCO), rapporteurs as MEP IDs and the timeline from referral to publication.

  2. European Parliament

    europarl_get_mep with mepId 124867 returns name, political group, national party and committee seats for one of the lead rapporteurs.

  3. European Parliament

    europarl_search_adopted_texts with tittel “artificial intelligence” and aar 2025 lists Parliament’s adopted texts. Use tittel and not sok (see limits).

  4. EUR-Lex

    eurlex_get_act with celex 32026R1744 (Digital Omnibus on AI) shows that the AI Act was amended in 2026.

IDs used: procedure 2021/0106(COD), MEP 124867 and 197665, CELEX 32024R1689 and 32026R1744.

Result

A chronology with rapporteurs and political groups, adopted texts and links to the current legal text.

Why it matters

Saves research on each case and gives members a source to refer to. The same recipe works for other procedures.

Limits

Content search (sok) in europarl_search_adopted_texts returned UPSTREAM_ERROR on 2026-10-23, while title search works. The API does not give the Council’s position, details from trilogue negotiations or vote counts per group. Norway has no MEPs, and how an act is incorporated into the EEA Agreement is not shown here.

Evidence from the run

Run 2026-10-23: procedure 2021/0106(COD) returned 25 events (referral 2021-06-07, plenary votes 2023-06-14 and 2024-03-13, signature 2024-06-13, publication 2024-07-12) and two lead rapporteurs from S&D and Renew. MEP 124867 (S&D) is a rapporteur. Title search in 2025 returned two adopted texts, among them T10-0286/2025 on AI in the financial sector.

UC-018Election campaign 2027: municipalities where the party is strongest and weakestNorwegian Directorate of Electionselection campaign strategist, County party or party secretariat · Politics and public affairs
WorkflowVerified 23 Oct 20262 tools

Sources Norwegian Directorate of Elections

Situation

The county party has to decide where to focus the campaign before the 2027 local elections. On valgresultat.no the figures are shown one municipality at a time, so it is cumbersome to see a whole county together.

The question

“Show Senterpartiet’s result in every municipality in Trøndelag in the 2023 municipal council election: share, change from 2019 and seats, sorted by share. Where did the party grow, and where did it fall most?”

How the AI does it

  1. Norwegian Directorate of Elections

    valg_get_party_results with valgaar 2023, valgtype ko, partikode SP and fylke 50 (Trøndelag) returns all 38 municipalities with percentage, change in percentage points, votes and seats.

  2. Norwegian Directorate of Elections

    valg_get_result with fylke, kommune and optionally krets goes down to polling district level where needed.

  3. Let the AI split the municipalities into defence, growth and decline, and ideally combine with Statistics Norway (SSB) figures to choose target groups.

Result

A ranked list of municipalities with the change since the last election, split into priority groups.

Why it matters

A faster basis for decisions on allocating resources in the campaign. You see where the party is growing and where it is losing.

Limits

Local lists have their own codes, so when you follow one party’s code you do not see voters who went to local lists. No candidate lists or personal votes. The Norwegian Directorate of Elections has no open licence (none found), so use the credit line “Kilde: Valgdirektoratet (valgresultat.no)” and clarify the terms for commercial use.

Evidence from the run

Run 2026-10-23: Sp got 12.9% in Trøndelag in 2023 (−9.1 percentage points, 31,002 votes, 220 seats, −109) across 38 municipalities. Highest in Tydal (58.5%) and Leka (52.8%, +9.4), lowest in Trondheim (2.0%). Largest falls in Verdal (−32.8) and Steinkjer (−23.2).

UC-019Which reindeer grazing districts does the project affect? (public affairs mapping)Geonorgepublic affairs adviser, Consultancy that assists a wind power or mineral developer · Politics and public affairs
WorkflowVerified 23 Oct 20262 tools

Sources Geonorge

Situation

The client is planning a project on Fosen and is looking at another near Karasjok. Before the first meeting with the authorities and people in the industry, the adviser needs to know which reindeer grazing districts cover the areas, and what they are called. Today that means searching in the Landbruksdirektoratet map viewer.

The question

“Our client wants to build wind power on Fosen and is looking at setting up near Karasjok. Which reindeer grazing districts are in the two areas, what are they called in Sami and Norwegian, and how large are they?”

How the AI does it

  1. Geonorge

    geonorge_search_datasets(sok “reindrift”) returns 20 hits. The first four have WMS and WFS: Reinbeitedistrikt (6bfec384-92cf-44d3-863b-0187afa06658), Flyttlei (f9c1e228-892f-4f1a-9e4e-b6d6149f373c), Restriksjonsområde (05a9cb73-639e-48ab-8533-bd80521a84bf) and Reindriftsanlegg (8dfa67c5-3099-4353-9ce0-72f9ebd44a2c).

  2. Geonorge

    geonorge_query_dataset on Reinbeitedistrikt with a box over the Fosen peninsula (63.6–64.3 N, 9.2–10.4 E) returns one district: “Fovsen-Njaarke / Fosen”.

  3. The same query with a point in Karasjok (69.4719 N, 25.5107 E) returns “Kárašjoga nuotabealli / Karasjok østre distrikt”.

  4. Next step, not run here: the same query against Flyttlei and Restriksjonsområde for migration routes and restrictions.

Result

District name (Sami and Norwegian), district code and area for each location, ready to go into the stakeholder map and the consultation plan.

Why it matters

The right district means the right dialogue partner and consultation body from day one, and the names are spelt the way Landbruksdirektoratet writes them.

Limits

Only boundaries and attributes, without geometry. Says nothing about seasonal pastures, migration routes or actual use until the other layers have been queried (not run here). The district polygons have the update date 2019 in the attributes. Not a legal clarification of reindeer husbandry interests and no substitute for consultation with the districts.

Evidence from the run

2026-10-23: a box over Fosen returns one district, “Fovsen-Njaarke / Fosen” (code VR, 4,345,591 daa); a point in Karasjok returns “Kárašjoga nuotabealli / Karasjok østre distrikt” (code ZQA, 2,435,140 daa); the catalogue has 20 reindeer husbandry datasets.

UC-020How much less electricity is Norway exporting in 2026?Own pageStatnettNVEbusiness journalist, National news desk · Journalism and fact-checking
WorkflowVerified 23 Oct 20263 tools
Open the scenario →

Sources Statnett, NVE

Situation

The journalist has heard that Norway “hardly exports electricity any more” and wants to check it against figures. Statnett’s exchange data is in an operational API, not in a ready-made table.

The question

“What was Norway’s net power exchange with other countries in the first to third quarter of 2026 compared with 2025, and what does the reservoir filling say?”

How the AI does it

  1. Statnett

    statnett_get_physical_flow with land ALL, frekvens kvartal, fra 2025-01-01 til 2025-09-30. Repeat for 2026 (the tool takes at most 400 days per call, so two calls are needed).

  2. Read nettoEksportMWh per quarter and statnettTotaler for export and import separately. Positive numbers are export.

  3. NVE

    nve_get_reservoir_levels to explain the change: reservoir level per price area compared with the 20-year normal.

  4. Statnett

    statnett_get_price_area_borders to explain which grid sections separate the price areas NO1–NO5.

Result

Two comparable figures with source: net exports Jan–Sep by quarter in 2025 and 2026, imports and exports separately, and the reservoir filling as the explanation.

Why it matters

A claim can be fact-checked in minutes, with quarterly figures others can repeat, without manual downloads from Statnett’s website.

Limits

Statnett has exchange data only from 1 January 2025, so longer time series do not exist. Physical flow is not trade or price, and Statnett does not explain the causes. The data lags several hours behind real time. No electricity prices.

Evidence from the run

Run 2026-10-23: net exports Jan–Sep 2025 were 17.06 TWh (Q1 7.36; Q2 4.43; Q3 5.28) and Jan–Sep 2026 were 3.67 TWh (Q1 0.19; Q2 1.49; Q3 1.99); imports 13.18 TWh in 2026 against 7.48 TWh in 2025. A call of 660 days returned INVALID_QUERY (max 400 days).

UC-021Fact check: “Norway has lower unemployment and inflation than the EU”Statistics NorwayEurostatfact-checker at an online newspaper, Newsroom at a national media outlet · Journalism and fact-checking
Combined toolVerified 23 Oct 20261 tools

Sources Statistics Norway, Eurostat

Situation

A politician claims that Norway does better than the EU on unemployment and inflation. The journalist has to find comparable series at Statistics Norway (SSB) and Eurostat and understand that the definitions are not quite the same before anything can be published.

The question

“Is it true that Norway has lower unemployment and lower inflation than the EU? Give me the figures for the last few years, with sources, and tell me where the definitions differ.”

How the AI does it

  1. Statistics NorwayEurostat

    compare_with_eu with indikator “arbeidsledighet”: Statistics Norway table 13760 (the Labour Force Survey, seasonally adjusted, monthly) and Eurostat une_rt_a (annual average) for Norway and EU27_2020.

  2. Statistics NorwayEurostat

    compare_with_eu with indikator “inflasjon”: Statistics Norway table 14711 (CPI) and Eurostat prc_hicp_aind (HICP).

  3. The text in “merknad” and the lines in “sources” are used as methodological caveats and source references.

Result

Unemployment in Norway was 4.5% in 2025 against 6.0% in the EU, but Statistics Norway’s monthly figure has risen to 5.0% in September 2026 (4.1% in July). For inflation the picture is mixed: HICP for Norway was lower than the EU in 2023 (5.8 against 6.4%), but higher in 2024 (2.8 against 2.6%) and 2025 (2.8 against 2.5%), and CPI for 2025 was 3.0%.

Why it matters

The fact check takes minutes and is based on two official sources with licence and link in the answer, instead of searching for comparable tables on two websites.

Limits

Only four curated indicators (population, unemployment, inflation, GDP per capita). The periods are not the same: Statistics Norway is monthly, Eurostat annual without 2026, and CPI and HICP do not measure the same thing (HICP leaves out owner-occupied housing). Other indicators require eurostat_search_datasets and ssb_search_tables.

Evidence from the run

Run 2026-10-23: compare_with_eu(arbeidsledighet) returned Statistics Norway 5.0% (2026M09) and Eurostat NO 4.5% against EU27 6.0% (2025); compare_with_eu(inflasjon) returned CPI 3.0%, HICP NO 2.8% and EU27 2.5% (2025).

UC-022Fact check: how many people use weight-loss medicines?FHIhealth journalist, National newspaper · Journalism and fact-checking
WorkflowVerified 23 Oct 20264 tools

Sources FHI

Situation

A story claims that “over ten per cent of the population” use weight-loss injections. The journalist has to find official figures in Legemiddelregisteret, which is in a technical statistics bank with its own code names.

The question

“How many people were dispensed weight-loss medicines on prescription in 2025, how many of them GLP-1 analogues, and how has the number changed since 2021? Tell me what the table actually measures.”

How the AI does it

  1. FHI

    fhi_list_sources shows Legemiddelregisteret (source lmr).

  2. FHI

    fhi_search_tables kilde=lmr returns 7 tables, among them 823 “Legemiddelgruppe – Vektreduserende. 2004–2025” (also ADHD medicines 819 and cardiovascular 820).

  3. FHI

    fhi_get_table_info 823 shows medicine groups (among them GLP-1 analogues, Saxenda, Wegovy, tirzepatide), sex, age groups 12–79, year and measures (number of individuals, per 1,000, DDD).

  4. FHI

    fhi_get_data 823 with Legemiddelgruppe_Verdi LegemidlerForVektreduksjon and GLP1AnalogerForVektreduksjon, Kjonn_Verdi TOTALT, Aldersgruppe_Verdi TOTALT, Utlevering_Ar siste(5) and MEASURE_TYPE AntallBrukere and Brukere_Per1000_Innbyggere.

Result

In 2025, 189,544 people (40.8 per 1,000 inhabitants aged 12–79, about one in 25) were dispensed medicines for weight loss, 170,860 of them GLP-1 analogues (36.8 per 1,000). From 2021 to 2025 the number of users rose 4.9-fold (38,818 → 189,544) and GLP-1 users 11-fold (15,595 → 170,860); from 2024 to 2025 growth was 19% for GLP-1.

Why it matters

The journalist gets an official figure with definition and year in a few minutes and can measure claims against it instead of quoting other media.

Limits

The table covers people aged 12–79 who have been dispensed a prescription in the weight-loss group; the groups for Saxenda and Wegovy are labelled “kun” (only those products), so diabetes medicines with the same active substance are not included. The register does not show why the medicine is used or whether use is continuous. The figures are national (the table has no geography) and updated 16 February 2026.

Evidence from the run

Run 2026-10-23: lmr/823 returned medicines for weight loss 38,818 (2021) → 189,544 (2025) people and GLP-1 analogues 15,595 → 170,860; 36.79 per 1,000 inhabitants in 2025.

UC-023Who got the largest framework agreements? From Doffin to the accountsDoffinRegnskapsregisteretEnhetsregisteretbusiness and public administration journalist, Newsroom · Journalism and fact-checking
WorkflowVerified 23 Oct 20264 tools

Sources Doffin, Regnskapsregisteret, Enhetsregisteret

Situation

The journalist wants to scrutinise the largest consultancy agreements in central government, see who got them and whether the winners are large enough to deliver. The notices are long, and the winners’ size is in another register.

The question

“Find the three largest awarded consultancy agreements in Doffin in 2026. Who won the largest, how many bids came in, how were the bids weighted, and how large are the winners compared with the agreement?”

How the AI does it

  1. Doffin

    doffin_search_notices with sok konsulent, type resultat, fra 2026-01-01 and sortering hoyest_verdi: 405 hits, largest first.

  2. Doffin

    doffin_get_notice on the largest (2026-115663, the Norwegian Environment Agency): lots, bids, winners and weighting.

  3. RegnskapsregisteretEnhetsregisteretDoffin

    regnskap_get_accounts for one winner (Kantega AS, 985815534) and get_company_procurements to see how many public contracts it has.

  4. The AI compares the ceiling with the winner’s revenue and writes a fact sheet for the story.

Result

The largest is the Norwegian Environment Agency’s framework agreement for digitalisation and technology: a ceiling of NOK 1,323 million split across four lots (NOK 909, 159, 170 and 85 million) and 54 bids. Nine suppliers won, among them PwC, Sopra Steria and Kantega. Price was weighted at 25%, competence and experience at 40–50%. Kantega has 13 awards in Doffin and NOK 316 million in group revenue.

Why it matters

The journalist goes from a search to a documented fact sheet in a short time, and can show that the figure in the headline is a ceiling shared between several suppliers, not an amount paid out.

Limits

Estimated value is a ceiling, not what is spent. Value per winner is not given (null), and actual use is decided in call-offs that are not advertised. Actual payments are not in these sources, and the free text in notices can in some cases mention contact persons who must not be reproduced.

Evidence from the run

Run 2026-10-23: sok konsulent, resultat, fra 2026-01-01, hoyest_verdi returned 405 in total, at the top 1,323,000,000 NOK (the Norwegian Environment Agency, 2026-10-08), then Skatteetaten 1,200,000,000 and a licence agreement for municipalities of 1,000,000,000. doffin_get_notice 2026-115663 returned antallTilbud 54, nine different winners and weights 35/40/25 on lot 1. get_company_procurements 985815534 returned 13 in total.

UC-024Find the quotes: what did the members say in the nuclear power debate?Stortingetpolitical journalist, News desk · Journalism and fact-checking
WorkflowVerified 23 Oct 20262 tools

Sources Stortinget

Situation

The journalist wants to quote what FrP and Høyre actually said in the debate, not what the party manifesto says. The transcripts are long (531 speeches in one day), and it is cumbersome to search them across days.

The question

“Find the speeches that mention nuclear power in Stortinget’s debate on 21 May 2026. Who said what, for which party, and at what time of day?”

How the AI does it

  1. Stortinget

    stortinget_hent_sak with sakid 200121 returns the transcript IDs in publikasjoner: refs-202526-05-21 is the day of the debate and refs-202526-05-26 is the vote.

  2. Stortinget

    stortinget_sok_i_referat with query “kjernekraft” and publikasjonid refs-202526-05-21 returns speeches with speaker, party, time, the case and an excerpt.

  3. Note: without publikasjonid the tool only searches the newest transcripts (default 3, max 10). The six newest returned zero hits, so find the right transcript ID first.

  4. Optional and not run here: stortinget_hent_publikasjon reads the whole transcript when the excerpt is too short to quote.

Result

A list of speeches with speaker, party, time and excerpt, ready to be quoted with source.

Why it matters

Precise quotes with time and source in minutes, and a lower risk of misquoting than when you read through the whole transcript.

Limits

The excerpts are short and truncated, so read the whole speech before you quote. The search requires all the words to be in the same speech. The field sak in voting sections points to the previous agenda item and not to the case being voted on. Only plenary transcripts are searched, not committee meetings.

Evidence from the run

Run 2026-10-23: the transcript refs-202526-05-21 (531 speeches) returned 36 hits on “kjernekraft”, among them a speech from FrP at 15:50 and one from H at 16:26. The six newest transcripts returned 0 hits. The voting transcript 2026-05-26 showed the text of the proposals, for example KrF’s proposal on small modular reactors by 2040.

UC-025Fact check: more doctorates, but fewer publication points?HK-dir (DBH)education journalist, National newspaper or online outlet · Journalism and fact-checking
WorkflowVerified 23 Oct 20262 tools

Sources HK-dir (DBH)

Situation

An opinion piece claims that Norway is educating more and more PhDs while research output is falling. The journalist has to find the official figures in DBH, where the tables have technical names and codes.

The question

“Is it true that the number of doctorates awarded has risen over the last ten years while the universities publish less? Give me figures from DBH for 2016–2025, split by sex, and publication points for UiO, UiB, NTNU and UiT.”

How the AI does it

  1. HK-dir (DBH)

    hkdir_search_tables with emne “Doktorgradsdata” returns table 100 (cooperation), 101 (doctorates awarded, aggregated, not suppressed) and 115 (people in doctoral programmes).

  2. HK-dir (DBH)

    hkdir_get_data on table 101 with Årstall between 2016 and 2025 and grupperPa Årstall returns totals, women and men per year.

  3. HK-dir (DBH)

    hkdir_get_data on table 373 (scientific publishing) for institution codes 1110, 1120, 1130 and 1150, 2021–2024.

  4. The AI calculates the changes and writes a short fact check with caveats and source credit.

IDs used: DBH tables 101 and 373.

Result

A time series for doctorates (1,410 in 2016, 1,878 in 2025) and the change in publication points per university, ready to use in a fact check.

Why it matters

The claim can be assessed against official figures in minutes, with exact years and table numbers the journalist can refer to.

Limits

The figures say nothing about the cause of the fall in points (changed channel levels, reporting rules and late registrations can play a part), and the latest year may be revised. Only four universities are included, not the whole sector. DBH requires a source reference and that AI-generated presentations are labelled.

Evidence from the run

Run 2026-10-23: table 101 returns 1,410 doctorates in 2016 and 1,878 in 2025 (998 women, 880 men; +33%). Table 373: publication points 2021→2024 fell 10% at UiO (14,284→12,819), 15% at UiB, 16% at NTNU and 16% at UiT.

UC-026Daily interest rate monitoring: policy rate, Nowa and the yield curveNorges Bankcredit analyst/treasury adviser, Regional savings bank · Finance, banking and insurance
MonitoringVerified 23 Oct 20261 tools

Sources Norges Bank

Situation

The bank follows the yield curve and changes in the policy rate for loan pricing and liquidity management. Today the figures are fetched by hand from Norges Bank’s website or pasted into a spreadsheet every morning.

The question

“Give me today’s policy rate, Nowa, the 3-, 6- and 12-month Treasury bill rates and the 3-, 5-, 7- and 10-year government bond yields. Calculate the spread against the policy rate and tell me if anything has changed since yesterday.”

How the AI does it

  1. Norges Bank

    norgesbank_get_interest_rates without arguments returns the latest value for all 11 rates in one response.

  2. Norges Bank

    norgesbank_get_interest_rates with rente=styringsrente and kunEndringer=true lists only the interest rate decisions.

  3. Schedule the run daily; the AI stores the previous day’s values and calculates spread and change.

Result

22 October 2026: policy rate 4.500%, Nowa 4.500%, Treasury bills 3/6/12 months 4.519/4.732/4.809%, government bonds 3/5/7/10 years 4.767/4.615/4.521/4.485%. Spread against the policy rate: +0.02/+0.23/+0.31 percentage points for the Treasury bills and +0.27/+0.12/+0.02/−0.02 percentage points for the bonds. The last interest rate decision was on 25 September 2026 (4.25 → 4.5%).

Why it matters

The same figures every day without manual work, with date and source credit, and a simple alert when the policy rate changes.

Limits

Only Norges Bank’s own series: not NIBOR, swap rates, the bank’s funding cost or forecasts. Treasury bills and government bonds are generic rates from Norges Bank, not traded prices. Nowa’s averages over 1, 3 and 6 months are not included.

Evidence from the run

Run 2026-10-23: one call returned all 11 series dated 2026-10-22 (policy rate 4.5; 10 years 4.485), and kunEndringer returned 18 interest rate decisions since 2021-09, the latest 2026-09-25.

UC-027Credit assessment: group and company accounts, three years of key figuresEnhetsregisteretRegnskapsregisteretcredit analyst in a regional bank, Bank · Finance, banking and insurance
WorkflowVerified 23 Oct 20262 tools

Sources Enhetsregisteret, Regnskapsregisteret

Situation

The bank is assessing a credit application from an IT group and needs margins, solvency and growth over several years. The accounts have to be fetched and calculated by hand, and group and parent company figures are easily confused.

The question

“Give me a credit assessment of Kantega AS (985815534) based on the last three years’ accounts. Use the group accounts. Calculate growth, operating margin, payroll costs as a percentage of revenue, equity ratio and liquidity, and tell me what is weak.”

How the AI does it

  1. Enhetsregisteret

    brreg_search_enheter with organisasjonsnummer 985815534 confirms that the company is active (konkurs and underAvvikling are false) and that the 2025 accounts have been filed.

  2. Regnskapsregisteret

    regnskap_get_accounts with organisasjonsnummer 985815534 and antall 6 (three years, group and company), optionally regnskapstype KONSERN.

  3. The AI calculates growth, operating margin, payroll share and trend. It uses KONSERN as the basis and shows the parent company’s figures as a check.

Result

The group’s operating revenue grew 2.7% to NOK 316 million in 2025, the operating margin rose from 9.7% to 10.6%, the equity ratio is 40.6% and the current ratio is 1.64. The parent company alone shows a margin of only 2.0%, which would have given a wrong picture. Payroll costs fell 1.8% while revenue rose.

Why it matters

The analyst gets a ready-calculated trend in a short time and avoids the classic mistake of reading the parent company’s figures as the group’s. That makes the first assessment of the case faster and safer.

Limits

Only figures as filed: no notes, cash flow, depreciation, pledges or debt maturities, and no credit score. Banks are not covered (the tool’s own description). The amounts are whole units in the currency of the accounts, which is not always NOK, and the figures are not checked against each other.

Evidence from the run

Run 2026-10-23: regnskap_get_accounts 985815534 returned KONSERN 2025 with operating revenue 315,621,068 NOK, operating profit 33,606,968 (10.6%) and equity ratio 40.6%; SELSKAP 2025 returned operating margin 2.0%; KONSERN 2024 returned 307,340,370 NOK. Hydro (914778271) shows the same pattern: group 6.8% against parent company −581.2%.

UC-028Risk screening of addresses for flood, quick clay and landslidesKartverketNVDB (Norwegian Public Roads Administration)Geonorge+2loss prevention officer / underwriter in non-life insurance, Insurance company with a housing and commercial property portfolio · Finance, banking and insurance
WorkflowVerified 23 Oct 20265 tools

Sources Kartverket, NVDB (Norwegian Public Roads Administration), Geonorge, NVE, Norwegian Environment Agency

Situation

Before a block of flats is insured or renewed, the company wants to know whether it is in areas the Norwegian Water Resources and Energy Directorate (NVE) has mapped for flood, quick clay or landslides. Today the addresses are looked up one by one in NVE Atlas, and the result is hard to document in the case.

The question

“We are going to insure a block of flats in Nedre Bakklandet in Trondheim and a holiday home in Flåm in Aurland. Are they in mapped flood, quick clay or landslide hazard zones, and what is today’s hazard level?”

How the AI does it

  1. KartverketNVDB

    describe_location returns coordinates and height (3.5 m a.s.l. in Bakklandet).

  2. Geonorge

    geonorge_search_datasets finds NVE’s Kvikkleire (a29b905c-6aaa-4283-ae2c-d167624c08a8), Flomsoner (e95008fc-0945-4d66-8bc9-e50ab3f50401) and Skredfaresoner (b2d5aaf8-79ac-40f3-9cd6-fdc30bc42ea1), all with WFS and WMS.

  3. Geonorge

    geonorge_query_dataset with a point: the quick clay area “Nedre Bakklandet” has hazard level “Middels”, consequence class 3 and risk class 4; the flood mapping for Nidelva covers 20-, 50-, 100-, 200- and 1,000-year floods.

  4. Geonorge

    geonorge_query_dataset with a box over Flåm (60.84–60.89 N, 7.08–7.16 E) returns 127 landslide hazard objects, among them 29 hazard zone polygons.

  5. NVEKartverket

    nve_get_hazard_summary returns today’s warning level for floods, landslides and avalanches as an extra.

Result

A risk note per address with zone name, hazard level, mapping source and links to NVE’s project and report page (given in the attributes).

Why it matters

Faster and more verifiable underwriting: the same query can be run against a whole portfolio of addresses, and each zone has source, date and report link.

Limits

The responses give attributes without geometry, so the zone edge and distance must be checked in NVE Atlas. No flood hit only means that the point is not in a mapped inundation area, and flood zones and hazard zones only cover priority areas. Storm surge, sea level, wind and surface water are not covered. Not a basis for pricing on its own.

Evidence from the run

2026-10-23: the quick clay area “Nedre Bakklandet” (read about 240 m from the address): hazard level Middels, risk class 4, and at the address itself 2 hazard zone boundaries + 2 release areas; flood: “Flom Nidelva” with mapped floods 20–1,000 years; the Flåm box: 127 landslide hazard objects (29 SkredFaresone, 91 SkredFaregrense).

UC-029Group overview of a seafood company from public registersEnhetsregisteretDirectorate of Fisheriesequity analyst, seafood, Brokerage · Finance, banking and insurance
WorkflowVerified 23 Oct 20263 tools

Sources Enhetsregisteret, Directorate of Fisheries

Situation

The analyst wants to see which licences, sites and applications a fish farming company actually holds, without reading the annual report. The licences are in subsidiaries, and the parent company shows nothing.

The question

“What fish farming licences, sites and applications do Mowi Seawater Norway AS and the Lerøy group have with the Directorate of Fisheries?”

How the AI does it

  1. Directorate of FisheriesEnhetsregisteret

    fiskeridir_get_company_overview on the parent company Lerøy Seafood Group ASA (975350940): 0 vessels, licences and sites, because the licences are in subsidiaries.

  2. Enhetsregisteret

    brreg_search_enheter with navn Lerøy and organisasjonsform AS lists 27 entities (Lerøy Aurora AS, Lerøy Midt AS, Lerøy Havfisk AS and others). Choose the relevant ones and watch out for unrelated hits such as Lerøy AS in the metal industry.

  3. Directorate of FisheriesEnhetsregisteret

    fiskeridir_get_company_overview per subsidiary, here Mowi Seawater Norway AS (921668236): licences with capacity in tonnes (TN), sites, 90 applications and 615 environmental reports.

  4. Directorate of Fisheries

    fiskeridir_search_environmental_reports on the organisation number: the newest sediment surveys with site condition 1–4.

Result

A table per legal entity with number of licences, capacity, sites, recent applications and condition from the latest environmental surveys.

Why it matters

Saves the analyst manual lookups in several registers, and quickly shows that a search on the parent company returns a misleading zero.

Limits

The overview is per organisation number. Group structure is not followed automatically, so subsidiaries have to be found by name search. It shows licences and permitted biomass, not production, harvest or finances. Owners who are private individuals have been redacted.

Evidence from the run

Run 2026-10-23: Lerøy Seafood Group ASA (975350940) has 249 employees and 0 vessels, licences and sites. Mowi Seawater Norway AS (921668236) shows 10 licences and 10 sites (more exist), 90 applications and 615 environmental reports; the name search Lerøy returned 27 AS companies.

UC-030The electricity bill in September: spot price versus Norgespris by price areaOwn pageElhubenergy adviser in a consumer organisation, Consumer organisation or electricity supplier · Energy and power
WorkflowVerified 23 Oct 20262 tools
Open the scenario →

Sources Elhub

Situation

Households ask whether Norgespris or spot price pays off, and the answer differs between the price areas. Elhub has a ready-made cost estimate per price area and month, but the dataset is little known and is in a catalogue of 17 datasets.

The question

“What would the electricity bill for September 2026 have been for a household using 15,000 kWh a year in each of the five price areas, with spot price, with Norgespris and with electricity support? In which areas did Norgespris pay off?”

How the AI does it

  1. Elhub

    elhub_list_datasets (entitet price-areas) shows the dataset METERING_VALUE_COMPARISON_EAC_PER_MONTH, with the filters estimatedAnnualConsumption (1–500,000 kWh) and consumptionGroup (household or cabin).

  2. Elhub

    elhub_get_energy_data with entitet price-areas, that dataset, fra 2026-09-01 and til 2026-10-01 returns one row per price area with kWh and estimated cost at spot price, Norgespris and support price. Without a filter, 15,000 kWh a year is used.

  3. Change the assumption with a filter (for example consumptionGroup cabin or a different annual consumption), and let the AI compare the areas and explain the differences.

Result

A table per price area with consumption and estimated monthly cost at spot, Norgespris and support price, and an assessment of where Norgespris paid off.

Why it matters

A dated and sourced answer to the question “does Norgespris pay off for us?” per price area, which can be used in a consumer story or a customer letter.

Limits

Elhub’s estimate for a modelled household, not invoices and not hourly prices: the MCP gives no hourly spot prices. The column description states NOK/kWh, but the figures read as NOK for the month (the cost divided by kWh gives NOK 0.40/kWh for Norgespris). Elhub publishes with a delay, and the terms of the schemes must be checked with the authorities.

Evidence from the run

Run 2026-10-23: September 2026, 15,000 kWh a year. Estimated monthly cost in NOK at spot, Norgespris and support price: NO1 1,111, 333 and 630; NO2 1,223, 345 and 650; NO3 1,012, 379 and 706; NO4 326, 359 and 300; NO5 1,270, 383 and 732. Only in NO4 was the spot cost lower than Norgespris.

UC-031Reservoir filling now versus the 20-year range, by price areaOwn pageStatnettNVEElhubpower analyst, Power producer or analysis firm · Energy and power
WorkflowVerified 23 Oct 20262 tools
Open the scenario →

Sources Statnett, NVE, Elhub

Situation

The analyst has to assess whether the winter could be tight and needs the reservoir level per price area compared with normal. The reservoir figures are held by the Norwegian Water Resources and Energy Directorate (NVE), and Statnett’s assessment is somewhere else.

The question

“How full are the reservoirs in NO1–NO5 now compared with the last twenty years, and what does Statnett say about the power situation?”

How the AI does it

  1. StatnettNVEElhub

    get_power_situation without arguments: Statnett’s level per price area (vurdering), reservoir filling for NO1–NO5 (magasin) and Elhub’s consumption and production for the latest published day.

  2. NVE

    nve_get_reservoir_levels without arguments: adds the country (NO) and three river basin regions (VASS), each with a normalt block (lowest, median and highest value for the same week over 20 years) and deviation from the median.

  3. Ask the AI to rank the areas by deviation from the median, point out those close to the 20-year minimum and set this against Statnett’s assessment.

Result

A ranked table of reservoir level, stored energy in TWh and deviation from the 20-year median per price area, with Statnett’s assessment alongside.

Why it matters

A quick basis for decisions instead of searching several websites. The change from the week before is in the response, so trends can be followed.

Limits

Weekly statistics (next publication 28 October at 13:00), per price area and not per reservoir. The normal block is history, not a forecast. No inflow, snowpack or spot prices, and Statnett’s level is a qualitative assessment.

Evidence from the run

Run 2026-10-23: week 42 (18 Oct) NO2 52.9% against the lowest week 42 value in 20 years of 53.0% (median 83.3%, deviation −30.4 percentage points), NO4 90.3% (+8.8) and the country 69.3% (60.6 of 87.4 TWh). At the same time Statnett assessed all five price areas as “Normal” (level 1).

UC-032Area check for offshore wind: petroleum, vessel traffic and waves at one pointNorwegian Offshore DirectorateNorwegian Coastal AdministrationMET Norwayproject developer, offshore wind, Offshore wind developer · Energy and power
WorkflowVerified 23 Oct 20263 tools

Sources Norwegian Offshore Directorate, Norwegian Coastal Administration, MET Norway

Situation

The developer is considering a point in the North Sea and needs to know quickly whether it conflicts with petroleum activities, regulated vessel traffic or rough wave conditions. The answers are held by the Norwegian Offshore Directorate, the Norwegian Coastal Administration and the Norwegian Meteorological Institute (MET Norway).

The question

“Are there licences, fields, wells or facilities near 59.35 N 4.50 E, is the point in a fairway or an area with a speed limit, and what are the waves like now?”

How the AI does it

  1. Norwegian Offshore Directorate

    sodir_check_location with lat 59.35, lon 4.5 and radius 10000 m: block, licences, fields, discoveries and the nearest wells and facilities.

  2. Norwegian Coastal Administration

    kystverket_check_sea_location with radiusMeter 2000: speed limits, fairways, compulsory pilotage, VTS and danger areas.

  3. MET Norway

    met_get_ocean_forecast at the same point: wave height, sea temperature and current for the next few hours.

  4. Repeat for several candidate points and ask the AI to put them together in a table.

Result

A simple area report per point: block 27/8, no petroleum activity within 10 km, no registered traffic regulation and wave height from MET Norway.

Why it matters

A first conflict check of many candidate points in minutes instead of days of map work.

Limits

Only a point check, not an area (polygon). No seismic data, CO2 storage, fishing interests or protected areas. The Norwegian Coastal Administration’s maps are not legally binding, and an empty response does not mean that no rules apply. Wave data is a forecast, not climate statistics for design.

Evidence from the run

Run 2026-10-23 for 59.35 N 4.50 E: the Norwegian Offshore Directorate found block 27/8 (North Sea) and no licences, fields, discoveries, wells or facilities within 10 km; the Norwegian Coastal Administration had all layers empty within 2 km; MET Norway gave a wave height of 2.3 m falling to 1.7 m (at 10–16) and a sea temperature of 12.8 °C.

UC-033Consumption, exchange and grid losses: explore the Elhub catalogueElhubenergy adviser for industrial customers, Energy consultancy · Energy and power
WorkflowVerified 23 Oct 20263 tools

Sources Elhub

Situation

The adviser is going to show an industrial customer how consumption in the price area is distributed and whether the grid loses a lot of electricity. Elhub has 17 open datasets at three levels, and it is not obvious which ones answer the question.

The question

“What share of electricity consumption in NO2 last week was industry, was the area a net exporter, and how large were the grid losses in Elvia’s grid area EIDSIVA-D?”

How the AI does it

  1. Elhub

    elhub_list_datasets without a filter: 17 open datasets (11 by price area, 3 by grid area, 3 by municipality) with resolution, maximum period, filters and columns.

  2. Elhub

    elhub_get_energy_data with entitet price-areas, datasett CONSUMPTION_PER_GROUP_MBA_HOUR, id NO2, 12–19 October and aggregering dag: consumption per group.

  3. Elhub

    elhub_get_energy_data with datasett EXCHANGE_PER_MBA_HOUR and the same week: energy into and out of the price area.

  4. Elhub

    elhub_find_grid_areas with sok elvia returns the EIC code for EIDSIVA-D. elhub_get_energy_data with entitet grid-areas and datasett LOSS_PER_MGA_HOUR returns grid losses.

  5. Ask the AI to calculate shares and grid losses in per cent.

Result

The secondary sector (industry) accounted for 54.6% of consumption in NO2 in week 42, NO2 had net exports of 161 GWh, and grid losses in EIDSIVA-D were 3.9%.

Why it matters

Shows how a generic data catalogue answers three different questions without knowing the dataset names in advance.

Limits

Elhub is about two days behind. The groups (primary, secondary, tertiary) are Elhub’s own and must be checked against the definition. The exchange cannot be broken down by neighbour with daily aggregation, and grid losses are Elhub’s calculation, not a cost. No individual customers or metering points (requires Maskinporten).

Evidence from the run

Run 2026-10-23: 17 datasets. NO2 12–18 Oct: secondary sector 372.2 of 681.3 GWh (54.6%), households 174.3 GWh; energy in 377.0 and out 538.3 GWh (net +161.3 GWh). EIDSIVA-D (50YWO-6UUIJ4BF1D, Elvia AS, NO1): grid losses 3.24 of 82.23 GWh (3.9%).

UC-034Wind power in Norway: who owns the largest wind farms, and which turbines?NVEinvestment analyst, Power company or infrastructure fund · Energy and power
WorkflowVerified 23 Oct 20262 tools

Sources NVE

Situation

The analyst is mapping who owns the largest wind farms and which turbines have been used, to assess development models and suppliers. The wind power database of the Norwegian Water Resources and Energy Directorate (NVE) has the data, but not in a format you can calculate with.

The question

“How much wind power is in operation in Norway, who owns the largest wind farms, and which turbines are in NO4?”

How the AI does it

  1. NVE

    nve_search_wind_power_plants without a filter and with maxTreff 5: number, MW and GWh for all plants in operation, and the five largest with owner and organisation number.

  2. Repeat with prisomraade 4, medDetaljer true and maxTreff 2: turbine manufacturer, type and size per turbine group.

  3. NVE

    nve_search_hydropower_plants with maxTreff 5: the largest hydropower plants with main owner, as a comparison.

  4. Ask the AI to group by main owner and turbine manufacturer.

Result

An overview of 65 wind power plants in operation (5,079 MW), the five largest with owner, and turbine types in NO4 (Nordex N149 and Siemens Gamesa SG-4.2-130).

Why it matters

Quick mapping of competitors and suppliers instead of manual collection from licence lists.

Limits

The wind register only shows the main owner, often a project company, and no ultimate investors. Part-owners are only listed for hydropower. Only normal-year production, not actual production. medDetaljer for a whole price area returns a great many rows, so use a low maxTreff.

Evidence from the run

Run 2026-10-23: 65 wind power plants in operation, 5,078.8 MW and 15,207 GWh normal-year production; largest Øyfjellet (Vefsn, 400 MW, 72 turbines, main owner Øyfjellet Wind AS) and Storheia (288 MW, Fosen Vind DA). NO4 has 13 plants and 1,159.2 MW; Øyfjellet has Nordex N149 (4.95–5.7 MW) and Kvitfjell (197.4 MW) Siemens Gamesa SG-4.2-130.

UC-035Check the NYTEK certificate and environmental condition for your own sitesDirectorate of Fisheriescompliance and HSE manager, Fish farming company with several sites · Aquaculture, fisheries and maritime
WorkflowVerified 23 Oct 20263 tools

Sources Directorate of Fisheries

Situation

The company must document that each site has a valid installation certificate (NYTEK) and see which nonconformities have been found and closed. The information is in several registers at the Directorate of Fisheries.

The question

“Does the site Grøtnes N (45005) have a valid NYTEK certificate, and which nonconformities were found at the verification?”

How the AI does it

  1. Directorate of Fisheries

    fiskeridir_search_installation_certificates (default: status ISSUED): valid certificates with site, holder, accredited inspection body, date of issue and expiry date.

  2. Directorate of Fisheries

    fiskeridir_get_installation_certificate with lokalitetsnummer: verification with nonconformities (grade and deadline), number of mooring analyses, deployment reports and inspections, and equipment.

  3. Directorate of Fisheries

    fiskeridir_search_environmental_reports with lokalitetsnummer or organisasjonsnummer: B and C surveys and site condition 1–4.

  4. Ask the AI to make a checklist with expiry dates, open nonconformities and sites in poor condition.

Result

A checklist per site: certificate valid until (2031-07-24 for Grøtnes N), inspection body, nonconformities with deadlines and the latest environmental survey.

Why it matters

Self-check without opening several portals. Open nonconformities and expiry dates are caught before inspection.

Limits

The registers show the certificate’s status, not the inspection report itself (the nonconformity is described only as “Se inspeksjonsrapport”). Files and component details are not served, and the method texts in the environmental reports, with consultant names, have been removed. Does not give a legal assessment.

Evidence from the run

Run 2026-10-23: 607 valid certificates. Grøtnes N (45005, Cermaq Norway Salmon AS): issued 2026-07-24, expires 2031-07-24, verification 2026-08-13 with one nonconformity graded CLOSED_MAJOR_NONCONFORMITY (deadline 2026-08-25), 14 units maximum and 6 present.

UC-036Licence pools and new applications: basis for a meeting with politiciansDirectorate of FisheriesEnhetsregisteretpublic affairs adviser for the aquaculture industry, Industry organisation or consultancy · Aquaculture, fisheries and maritime
WorkflowVerified 23 Oct 20263 tools

Sources Directorate of Fisheries, Enhetsregisteret

Situation

The adviser is preparing a meeting with politicians on biomass regulation between production areas. The licence pool scheme is little known, and figures on who takes part are not gathered anywhere.

The question

“How many licence pools have been approved, who holds the first one and how much capacity do they have, and what are the newest aquaculture applications?”

How the AI does it

  1. Directorate of Fisheries

    fiskeridir_search_license_pools (default: approved): pools with type, production areas, holders (organisation number) and total and current capacity in tonnes.

  2. Directorate of FisheriesEnhetsregisteret

    fiskeridir_get_company_overview on the holder’s organisation number (921668236, Mowi Seawater Norway AS) to put a name to the pool.

  3. Directorate of Fisheries

    fiskeridir_search_aquaculture_applications with nyesteForst: number of applications and the newest with type and applicant.

  4. Ask the AI to make a fact box on the pools and the applications for the meeting.

Result

Fact box: 34 approved pools, two examples with capacity and holders, and the three newest applications.

Why it matters

The meeting is based on verified figures and named players instead of estimates.

Limits

Shows approved pools and application status, not case handling or political assessment. The application search has no filter on status or type. Pools are shown with organisation numbers, so names have to be looked up. Private applicants have been redacted.

Evidence from the run

Run 2026-10-23: 34 approved pools. P01-P03-0001 (areas 1–3, holder 921668236, 43,194 TN total and 41,640 TN current, decision 22/19856) and P02-P03-0001 (areas 2–3, four holders, 25,862 TN, decision 26/3307). 946 applications; newest SP-2026-002804 (land-based, 22 Oct) and SP-2026-002802 (Lerøy Midt AS, change of site Engdal, 20 Oct).

UC-037From product weight to round weight: conversion factors for codDirectorate of Fisherieslanding station manager at a fish buyer, Fish landing station · Aquaculture, fisheries and maritime
WorkflowVerified 23 Oct 20262 tools

Sources Directorate of Fisheries

Situation

The landing station has to convert the weight of processed cod to round weight and wants to be sure that the factor is the current one. The factors are in a register of about 1,950 rows.

The question

“What is the conversion factor to round weight for cod (code 1022) with product condition 211 for different preservation methods, and which approved buyers are there in Vardø?”

How the AI does it

  1. Directorate of Fisheries

    fiskeridir_get_conversion_factors with fiskeart 1022 and produkttilstand 211: factor per preservation code, validity period and time of change.

  2. Add konservering (two digits) or tidspunkt to see the factor that applied on a given date.

  3. Directorate of Fisheries

    fiskeridir_search_buyers with sok Vardø: approval number, buying points and sales organisation.

  4. Ask the AI to calculate round weight for a given product weight.

Result

Four factors for cod with product condition 211: 1.50 (preservation 99), 2.55 (13), 3.65 (14) and 6.41 (17), all unchanged since 2017, and three registered buying points in Vardø.

Why it matters

The right factor in seconds from the same source the directorate itself uses, and fewer corrections of sales notes.

Limits

The tool only gives codes and numbers, not the names of product condition and preservation, so the code texts have to be looked up at the Directorate of Fisheries. The buyer register hides address and position for car, boat and quayside sales.

Evidence from the run

Run 2026-10-23: fiskeart 1022 and produkttilstand 211 returned 4 factors (preservation 13: 2.55; 14: 3.65; 17: 6.41; 99: 1.5), valid from 1991 (17 from 2013) and last changed 2017-11-01. The buyer register returned 3 locations in Vardø, among them approval numbers F707 and F836.

UC-038Plan Drammen–Oslo: reference route, speed limit and fairwayNorwegian Coastal Administrationoperations manager, Small shipping company in the Oslofjord · Aquaculture, fisheries and maritime
WorkflowVerified 23 Oct 20263 tools

Sources Norwegian Coastal Administration

Situation

The operations manager is planning a trip Drammen–Oslo and wants to use the Norwegian Coastal Administration’s recommended route and know which speed limits, fairway and VTS apply at critical points. The information is in several different map layers.

The question

“Give me the Norwegian Coastal Administration’s reference route Drammen–Oslo with length and waypoints, and check which speed limit and fairway apply at Svelvikrennen.”

How the AI does it

  1. Norwegian Coastal Administration

    kystverket_find_reference_routes with sok Drammen Oslo: four routes (NO-130002 Drammen–Oslo, NO-130003 Oslo–Drammen and NO-130012 Oslo–Hurum–Drammen 7m for smaller vessels, among others).

  2. Norwegian Coastal Administration

    kystverket_get_reference_route with NO-130002: waypoints, length, validity and product information (checked for 150 m length and 9 m draught).

  3. Norwegian Coastal Administration

    kystverket_check_sea_location with waypoint no. 7 (Svelvikrennen, 59.60445 N 10.41594 E) and radiusMeter 300: speed limits, fairway and VTS.

  4. Ask the AI to mark the waypoints where a speed limit or VTS applies.

Result

A route with 25 waypoints and 84.9 km, and for Svelvikrennen: 8 knots for both commercial and leisure vessels, main fairway Guldholmen–Drammen no. 1010 and VTS Horten.

Why it matters

Route and rules in one place make a shareable planning template and reduce the risk of breaking speed limits.

Limits

Reference routes are planning material, not nautical charts and not binding, and the service does not give units for deviation and depths (look them up in the RTZ standard). The map layers are not legally binding; the linked regulations apply. No AIS or ship positions. An empty layer does not mean that no rules apply.

Evidence from the run

Run 2026-10-23: NO-130002 Drammen–Oslo is 84.9 km (45.8 nautical miles) with 25 waypoints, valid 2026-06-10 to 2027-07-13. At Svelvikrennen: 8 knots (Drammen, Asker), main fairway Guldholmen–Drammen no. 1010 and VTS Horten.

UC-039Follow customers’ permit cases in an agency’s public journaleInnsynbusiness developer in an environmental technology company, Supplier to the oil and gas industry · Oil, gas and supply industry
MonitoringVerified 23 Oct 20262 tools

Sources eInnsyn

Situation

A supplier of treatment technology wants to pick up when the oil companies apply for new discharge requirements, since that is a sales signal. The Norwegian Environment Agency’s journal has them, but it is cumbersome to follow by hand.

The question

“What has the Norwegian Environment Agency registered in its journal about Equinor recently? Are there applications for discharges, and what do they concern? Give me a weekly list.”

How the AI does it

  1. eInnsyn

    einnsyn_search_journal with sok Equinor and orgnummer 999601391 (the Norwegian Environment Agency), and fra = last week.

  2. eInnsyn

    einnsyn_get_journalpost on the interesting entries: sender, case and documents.

  3. Run weekly, and let the AI mark new applications for discharges, monitoring and consents.

Result

Three new entries from 16 October: a monitoring programme for Mongstad, consent for a fire training ground at Sture-terminalen and an application to extend requirements for discharges of produced water and jetting on Oseberg. The last one is registered as 2026/88 with Equinor Energy AS as sender.

Why it matters

The supplier gets early signals about what the customers are applying for, and can make contact with a concrete offer while the case is open.

Limits

Only titles and metadata: the documents were “ikke tilgjengelig” without download links, and document text is not read. The journal shows letters, not decisions. Exempted names are hidden, and the search term must be in the public title or the case title.

Evidence from the run

Run 2026-10-23: einnsyn_search_journal sok Equinor, orgnummer 999601391 returned three entries dated 2026-10-16 and harFlere true. einnsyn_get_journalpost jp_01m5jvvgz3eccb112yaaqgn4za returned journal number 2026/88, sender Equinor Energy AS and two documents without links.

UC-040Potential customers on Johan Sverdrup: facilities, wells and operatorsNorwegian Offshore Directoratesales manager at a supplier company, Supplier to the oil and gas industry · Oil, gas and supply industry
WorkflowVerified 23 Oct 20262 tools

Sources Norwegian Offshore Directorate

Situation

The sales manager wants to understand the scope of a field (facilities, subsea templates, wells) and which operators have the most activity, to set priorities for the sales work. The data is spread across the Norwegian Offshore Directorate’s fact pages.

The question

“Which facilities and how many wells belong to Johan Sverdrup, is anything being installed, and which operators have the most active licences?”

How the AI does it

  1. Norwegian Offshore Directorate

    sodir_get_field_profile with navn JOHAN SVERDRUP: the field’s status, operator, discovery year, supply base, facilities with phase and start-up date, and wells by purpose.

  2. Look for facilities in the phase INSTALLATION, which points to upcoming deliveries.

  3. Norwegian Offshore Directorate

    sodir_search_licences with status ACTIVE and grupperPaa operator: number of active licences per operator.

  4. Ask the AI to put together a customer list with links to the Norwegian Offshore Directorate’s fact pages.

Result

A field profile with facilities and number of wells, and a ranking of operators by active licences, with links to the fact pages.

Why it matters

Quick customer mapping without browsing the fact pages, and the same query can be repeated for other fields.

Limits

No production volumes, reserves or contracts (they are in other Norwegian Offshore Directorate services). The operators have no organisation number in the data, so linking to the Brønnøysund Register Centre requires a name search. The field name must be written exactly.

Evidence from the run

Run 2026-10-23: Johan Sverdrup is in production, oil, operator Equinor Energy AS, discovered 2010, supply base Dusavik; 19 facilities (the multi-well templates J and N in the phase INSTALLATION) and 137 wells (61 production, 36 appraisal, 31 injection). 540 active licences: Equinor Energy AS 222, Aker BP ASA 130, Vår Energi ASA 59.

UC-041Exploration since 2006: discoveries and exploration wells per yearNorwegian Offshore Directorateexploration analyst, Oil company or consultancy · Oil, gas and supply industry
WorkflowVerified 23 Oct 20262 tools

Sources Norwegian Offshore Directorate

Situation

The analyst wants to show whether exploration on the Norwegian shelf is falling or holding up, but the figures have to be put together from two registers.

The question

“How many discoveries and how many exploration wells have there been per year since 2006, and which were the peak years?”

How the AI does it

  1. Norwegian Offshore Directorate

    sodir_search_discoveries with funnaarFra 2006 and grupperPaa funnaar: number of discoveries per year.

  2. Norwegian Offshore Directorate

    sodir_search_wellbores with formaal WILDCAT, paabegyntFraAar 2006 and grupperPaa paabegyntAar: exploration wells per year.

  3. Ask the AI to put the years side by side, find the peak years and calculate an average for 2006–2025 (2026 is not complete).

Result

Annual series 2006–2026 with discoveries and exploration wells, ready for a chart.

Why it matters

A fact-based description of the level of exploration without counting by hand in the fact pages.

Limits

Discovery year and drilling year are not the same, so the ratio of discoveries to wells is not a success rate. 2026 is a part year. No volumes (recoverable resources) and no exploration costs.

Evidence from the run

Run 2026-10-23: 350 discoveries 2006–2026, most in 2009 (29), 2008 (26) and 2014 and 2025 (22 each); 661 exploration wells, most in 2009 (44), 2019 (43) and 2013 (42). 2025 had 40 wells, 2026 so far 14 wells and 11 discoveries.

UC-042CPI adjustment of rent: the change over the last 12 monthsStatistics Norwayproperty manager, Management company for flats and commercial premises · Property and construction
LookupVerified 23 Oct 20262 tools

Sources Statistics Norway

Situation

Many leases are adjusted every year in line with the consumer price index (CPI). The manager has to find the right index month, calculate the percentage change and update many contracts on the same day.

The question

“What is the change in CPI from September 2025 to September 2026, and what will the new monthly rent be for a flat at NOK 15,000?”

How the AI does it

  1. Statistics Norway

    ssb_search_tables “konsumprisindeks” returns table 14710 (CPI, 2025=100, monthly 1920M03–2026M09, updated 9 October 2026).

  2. Statistics Norway

    ssb_get_data 14710 with Tid 2025M09 and 2026M09.

  3. The AI calculates the change and the new rent and shows the table number and month as an audit trail.

Result

CPI went from 100.6 (September 2025) to 104.0 (September 2026), an increase of 3.4%. A rent of NOK 15,000 becomes NOK 15,507 (up NOK 507).

Why it matters

The right index, the right month and a calculation that can be checked, instead of manual lookups and calculation errors on many contracts. Can be run every month after Statistics Norway (SSB) has published.

Limits

The contract’s own terms decide which month and which index apply, and the rules on adjustment in husleieloven must be checked; the tool only gives the index. The index is given to one decimal place, so rounding can cause small differences in the percentage.

Evidence from the run

Run 2026-10-23: ssb_get_data 14710 returned 2025M09 = 100.6 and 2026M09 = 104.0 (the table updated 2026-10-09).

UC-043Contractor: what is advertised in Vestland now, and what is pre-announced?Doffintender manager at a contracting firm, Mid-sized contractor in the Bergen region · Property and construction
LookupVerified 23 Oct 20261 tools

Sources Doffin

Situation

The contractor wants to see all new building and civil engineering tenders in Vestland, including pre-announced projects and market dialogues, to plan capacity. The county authority, the municipalities and central government clients advertise separately.

The question

“Which building and civil engineering procurements in Vestland have been advertised or pre-announced since 1 September? Include the client, deadline and value, and tell me which ones suit a contractor with 60 employees.”

How the AI does it

  1. Doffin

    doffin_search_notices with kontraktstype bygg_og_anlegg, nuts NO0A2 (Vestland), type planlegging and fra 2026-09-01: 44 hits.

  2. Read the typer field: ADVISORY_NOTICE is an advance notice or market dialogue, ANNOUNCEMENT_OF_COMPETITION is an advertised tender.

  3. Repeat with type konkurranse and status aktiv, and save the list for next week.

Result

44 hits. Among them Forsvarsbygg (roof rehabilitation, Borgen, deadline 5 Nov), Bergen kommune (charging facilities in five resident parking areas with 407 spaces, deadline 13 Nov), Vestland fylkeskommune (bridge maintenance on 51 bridges 2027–2029, deadline 2 Dec) and Bergen Vann (water main in Lars Hilles gate, deadline 18 Nov).

Why it matters

The contractor sees the whole region’s pipeline in one search and can plan capacity instead of following several clients’ portals.

Limits

The filter planlegging returns all notices marked PLANNING, including advertised tenders, so typer has to be read for each hit. Many notices lack an estimated value. Tender documents and drawings are held by the client.

Evidence from the run

Run 2026-10-23: doffin_search_notices kontraktstype bygg_og_anlegg, nuts NO0A2, type planlegging, fra 2026-09-01 returned 44 in total, among them 2026-116429 (Vestland fylkeskommune, 51 bridges, deadline 2026-12-02) and 2026-116529 (Forsvarsbygg, deadline 2026-11-05).

UC-044Protection and building history before converting an Art Nouveau blockRiksantikvarenarchitect and heritage adviser, Architects’ practice with restoration commissions · Property and construction
WorkflowVerified 23 Oct 20262 tools

Sources Riksantikvaren

Situation

Before the sketch design, the adviser wants to know whether the building and its neighbours are protected, which type of protection applies, and how the building is dated and described. The register is on kulturminnesok.no, but does not give an overview of a whole street.

The question

“We are converting an apartment block in Nedre Bakklandet in Trondheim. Are the building and its neighbours protected, what type of protection applies, how is it dated, and what does the register say about material and original use?”

How the AI does it

  1. Riksantikvaren

    riksantikvaren_search_heritage_sites(lat, lon, radiusKm 0.3) returns registrations nearest first, with distance, type of protection and link.

  2. Riksantikvaren

    riksantikvaren_get_heritage_site(id “255818-0”) returns dating, main material, original function, protection act and extent (box).

Result

The building is registered as an Art Nouveau apartment block from 1908 (dating “sannsynlig”), brick, originally a dwelling, type of protection “Vernet etter PBL” (municipal protection). A neighbouring building is “Kommunalt listeført”, and the protection act and section are empty, so it is not formally protected under kulturminneloven.

Why it matters

Shows early on whether the case requires dialogue with the municipality and the heritage authority, and gives the right terms (listed and protected under plan- og bygningsloven, as opposed to formally protected). A direct link to kulturminnesok.no is included.

Limits

239 registrations within 300 m, but the search is incomplete in a dense city (398 in the grid cell, 300 read): narrow it down with niva or vernetype. The protection date “1900-01-01” is a placeholder, not a real date. The register does not give outlines, SEFRAK buildings or cultural environments. The descriptions are the register’s own text and may mention architects and clients: consider whether to reproduce them.

Evidence from the run

2026-10-23: 239 registrations within 300 m (incomplete, 398 in the grid cell); 255818-0 “Nedre Bakklandet”: apartment block, 1908, brick, “Vernet etter PBL”, protection act empty.

UC-045Snow load last winter at Geilo: peak value and melt dateNVEconsulting engineer (buildings), Consulting engineering firm doing condition assessments and damage cases · Property and construction
LookupVerified 23 Oct 20261 tools

Sources NVE

Situation

After a snowy winter the engineer wants to assess whether a holiday building or a hall at Geilo was exposed to high snow load, and when the snow disappeared, for a damage case or a condition assessment. Modelled snow load for one place and one winter is hard to get without a model of your own.

The question

“What was the snow load (kg/m²) at Geilo in winter 2025/26, when was the peak, and when was the snow gone?”

How the AI does it

  1. NVE

    nve_get_grid_timeseries(lat 60.534, lon 8.206, tema “snowload”, fra 2025-11-01, til 2026-04-30) returns daily values in kg/m² and a summary (max, mean).

  2. Read the date of the maximum and the last day with a value above 0.

Result

Maximum snow load 108.9 kg/m² on 9 March 2026 (the grid cell is at 859 m a.s.l.), an average of 40.8 over the period, and zero again from 24 April.

Why it matters

Quick, verifiable documentation for damage cases and condition reports, and a basis for assessing the need for snow clearing.

Limits

Modelled load for a 1 km grid cell, not characteristic design snow load under NS-EN 1991-1-3, and not local snowdrifts, drifting or roof shape. Use the Eurocode and the national annex for calculation.

Evidence from the run

2026-10-23: snowload for Geilo (859 m a.s.l.), 181 days with no missing values: max 108.9 kg/m² (2026-03-09), mean 40.83, zero from 2026-04-24.

UC-046Site facts for the sales prospectus: radon, flooding, quick clay and cadastreKartverketNVDB (Norwegian Public Roads Administration)Geonorge+2estate agent and surveyor, Estate agency · Property and construction
WorkflowVerified 23 Oct 20265 tools

Sources Kartverket, NVDB (Norwegian Public Roads Administration), Geonorge, Norwegian Environment Agency, Riksantikvaren

Situation

The estate agent wants the local conditions included in the sales prospectus and to avoid random omissions. The information is held by NGU, the Norwegian Water Resources and Energy Directorate (NVE), the Norwegian Environment Agency and the Norwegian Mapping Authority (Kartverket).

The question

“Make a site facts appendix for Nedre Bakklandet in Trondheim: cadastral number, elevation, radon, flood and quick clay mapping, contaminated ground nearby and whether the building is protected. Include sources and any restrictions on use.”

How the AI does it

  1. KartverketNVDB

    describe_location returns the address, coordinates, 3.5 m a.s.l. and the nearest road.

  2. Kartverket

    kartverket_eiendom_naer_punkt(radius 15) returns cadastral number 409/89 (parcel).

  3. Geonorge

    geonorge_query_dataset: Radon aktsomhet (dc0605f3-2301-4abe-a91f-6da42464c281) returns “lav aktsomhet” and “meget lav sannsynlighet (< 15 %)” for an indoor measurement above 100 Bq/m³; Kvikkleire (a29b905c-6aaa-4283-ae2c-d167624c08a8) and Flomsoner (e95008fc-0945-4d66-8bc9-e50ab3f50401) return the mapping areas.

  4. Norwegian Environment AgencyRiksantikvaren

    miljodir_check_location(radius 150) lists contaminated ground, and riksantikvaren_search_heritage_sites shows that the building is registered as “Vernet etter PBL”.

Result

A finished appendix with sources and licence, including the dataset’s own restriction on use: the radon map cannot be used to predict radon in individual buildings, and a measurement is the only reliable answer.

Why it matters

A consistent and verifiable appendix where the right caveat comes with the data.

Limits

No ownership data, only the cadastral number. Mapped zones are not a condition report or a quick clay investigation, and no hit in a flood zone does not mean there is no flood risk. Radon must be measured. The list of contaminated ground is truncated (at least 8).

Evidence from the run

2026-10-23: cadastre 409/89; radon “lav aktsomhet” (indoor measurement: very low probability); quick clay mapping area “Trondheim 1621-4”; flood “Flom Nidelva”; at least 8 contaminated ground sites within 150 m; the building “Vernet etter PBL”.

UC-047Who comes to Tromsø? Foreign guest nights by countryOwn pageStatistics NorwayNorges Bankmarketing manager at a destination company, Tourism organisation in Northern Norway · Transport, travel and tourism
WorkflowVerified 23 Oct 20263 tools
Open the scenario →

Sources Statistics Norway, Norges Bank

Situation

The marketing manager is planning media buying and marketing for next season together with the agency. She wants to see which countries the guest nights come from and whether a stronger krone explains the changes, and the figures are in a Statistics Norway (SSB) table with over 40 countries.

The question

“How did guest nights in Tromsø change in August 2026 compared with August 2025, by country? Which markets are growing and which are falling, and is it true that the krone exchange rate against the euro explains it?”

How the AI does it

  1. Statistics Norway

    ssb_search_tables “overnatting kommune” returns table 12894 (guest nights by municipality, type of accommodation and the guests’ country of residence, monthly 2024M01–2026M08).

  2. Statistics Norway

    ssb_get_data 12894 with Region 5501 (Tromsø), InnKvartering1 *, Landkoder2 * and Tid 2025M08 and 2026M08. The variables are called InnKvartering1 and Landkoder2; the first attempt with the wrong names returned an error message that showed the right names.

  3. Norges Bank

    norgesbank_get_exchange_rates with valuta=EUR, frekvens=maaned and fra=2025-10-01 returns monthly averages for the krone exchange rate.

  4. The AI calculates the change per country and compares euro countries with the USA and the UK.

Result

Hotels in Tromsø in August: foreign guest nights 36,414 → 39,256 (up 7.8%). The USA up 18% (4,974 → 5,859), the UK 33%, Italy 21% and Spain 54%, while Germany is down 16% (5,383 → 4,521) and the Netherlands 36%. The krone became stronger against the euro (monthly average 11.84 in December 2025 against 10.80 in September 2026), but the growth from Italy and Spain shows that the exchange rate alone does not explain the change.

Why it matters

The marketing manager and the agency get a data basis for geotargeting and budget allocation per market, and can test a hypothesis against figures instead of assuming.

Limits

Only guest nights at commercial accommodation establishments (not guests, revenue or day visitors), and the series only starts in 2024. Small countries have small numbers and noise, and co-variation with the exchange rate is not causation. The exchange rate is a monthly average; October has not been published yet.

Evidence from the run

Run 2026-10-23: Statistics Norway 12894 Tromsø August, hotels: abroad 36,414 → 39,256, USA 4,974 → 5,859, Germany 5,383 → 4,521. Norges Bank EUR monthly average 11.8448 (2025-12) → 10.8003 (2026-09).

UC-048Day plan for guests in Tromsø: flights, buses and polar nightOwn pageAvinorMET NorwayKartverket+1managing director of a tourism company, Activity company in Tromsø · Transport, travel and tourism
WorkflowVerified 23 Oct 20263 tools
Open the scenario →

Sources Avinor, MET Norway, Kartverket, Entur

Situation

The company picks up guests at the airport and sells northern lights and winter experiences. Arrivals, buses, weather and light conditions are fetched today from four different websites.

The question

“How many flights land in Tromsø in the next three hours and are any delayed, what is the weather like, when does the nearest bus to the city centre leave, and when is the sun up on 21 December?”

How the AI does it

  1. AvinorMET Norway

    get_airport_overview with flyplass TOS, retning ankomst and timer 3: arrivals with status, weather and weather warnings in one call.

  2. KartverketEntur

    find_departures_near with the airport’s coordinates (69.6833 N 18.9189 E), transportmiddel buss and radius 800 m: the nearest stop and the next departures with real-time data.

  3. MET Norway

    met_get_sun_and_moon for Tromsø 2026-12-21: polar night, highest position of the sun and the moon’s path.

  4. Ask the AI to make a day sheet for the guides.

Result

One sheet with arrivals and delays, weather, the next bus and the light conditions for the day.

Why it matters

Guides and customer service do not have to search in several apps, and real-time data on departures and delays makes it easier to look after guests.

Limits

Avinor’s times are estimates and can be up to 3 minutes old, and the source “Flydata fra Avinor” must be shown as a link. The stop nearest the airport’s reference point is not necessarily the airport bus stop. No northern lights forecast, and the weather forecast is a point forecast.

Evidence from the run

Run 2026-10-23 at about 10:45: 14 arrivals at Tromsø (ENTC) in the next 3 hours, 1 delayed and 0 cancelled; weather 8.1–9.5 °C, wind max 5.9 m/s, no weather warnings. The stop Skognesvegen (495 m): lines 420, 42 and 40 with 2–3 minutes’ delay. 21 December: polar night, the sun at its highest −3.1° at 11:42, the moon 44° at 21:23.

UC-049Tenders in the Nordics: open bus tenders in Sweden and Norway in one conversationTEDDoffintender manager at a bus company, Norwegian bus company considering Sweden · Transport, travel and tourism
WorkflowVerified 23 Oct 20263 tools

Sources TED, Doffin

Situation

The company wants to know which bus service contracts are open in the neighbouring countries now, and how large they are. TED has its own filters and languages, and Norwegian tenders are in Doffin, the Norwegian procurement database, so you have to look in two places.

The question

“Show open tenders for bus services in Sweden right now: buyer, value and deadline. Look at the largest one in detail, and check whether there are similar open tenders in Norway.”

How the AI does it

  1. TED

    ted_search_notices with land SWE, cpv 60112000 (public road transport services) and aapne true: 4 open notices.

  2. TED

    ted_get_notice 585078-2026 returns the details of the largest, Region Sörmland (SEK 2.2 billion).

  3. Doffin

    doffin_search_notices with cpv 60112000 and status aktiv checks Norway.

  4. Repeat with land DNK and FIN. The AI makes a Nordic tender map and recommends which ones fit.

Result

Four open Swedish tenders, among them Region Sörmland “Busstrafik Södra Sörmland 2029” (SEK 2.2 billion, deadline 7 Dec, start of operations December 2029), Östgötatrafiken (SEK 110 million, deadline 15 Dec) and Länstrafiken i Västerbotten (SEK 30 million, deadline 30 Oct). In Norway there are no active ones on the same CPV code today.

Why it matters

The company gets a Nordic picture of tenders in one conversation and can decide early whether to spend time on a Swedish bid or wait for Norwegian ones.

Limits

land is the buyer’s country, with one country per search. The texts are in Swedish or English (sprak chooses the language), and the TED notice does not give award criteria for this one. The tender documents are held by the buyer. Norwegian buyers may use CPV codes other than 60112000, so 0 hits in Doffin is not the same as no tenders.

Evidence from the run

Run 2026-10-23: ted_search_notices land SWE, cpv 60112000, aapne true returned 4 in total, among them 585078-2026 (2,200,000,000 SEK, deadline 2026-12-07). ted_get_notice 585078-2026 returned buyer type regional authority, NUTS SE122 and an empty list of criteria. doffin_search_notices cpv 60112000, status aktiv returned 0 in total.

UC-050Toll stations, tunnels and bridges in Bergen for heavy vehicle routesKartverketNorwegian Environment AgencyEntur+2transport planner, Distribution company with heavy vehicles in the Bergen region · Transport, travel and tourism
Combined toolVerified 23 Oct 20261 tools

Sources Kartverket, Norwegian Environment Agency, Entur, NVDB (Norwegian Public Roads Administration), Trafikkdata

Situation

The planner wants to calculate tolls and see route obstacles for heavy vehicles in Bergen. Toll stations, tunnels and bridges are in the National Road Database (NVDB) of the Norwegian Public Roads Administration, but otherwise have to be fetched per road or object type.

The question

“Which toll stations with rates for small and large vehicles, which tunnels (length, year opened, subsea) and which large bridges are there in Bergen municipality?”

How the AI does it

  1. KartverketNorwegian Environment AgencyEnturNVDBTrafikkdata

    find_in_area(kommune “4601”, kategorier [“veg”], maxPerKategori 6) reads NVDB for the municipality’s box (29.8 × 40.0 km).

  2. Increase maxPerKategori (up to 100) for the full list: the field avkortet shows that the lists have been truncated.

Result

Bomringen Bergen (Ferde AS): 6 stations shown, rate 38 (small vehicle) and 88 (large vehicle) in the NVDB data. Tunnels such as Knappetunnelen (6,500 m, subsea, opened 2010) and Råtunnelen (2,200 m, 2022), and bridges such as Sotrabrua (suspension bridge, 1,236 m, 1971), Askøybrua (1,057 m, 1992) and Nordhordlandsbrua (floating bridge, 1,615 m, 1994).

Why it matters

A basis for a route and cost model for heavy vehicles. The credit line (“Inneholder data under norsk lisens for offentlige data (NLOD) tilgjengeliggjort av Statens vegvesen”) is included.

Limits

Only 6 per category were shown when the call was run (avkortet). The box is a rectangle and includes neighbouring municipalities (a bridge in Samnanger was included). The rates are NVDB properties, not discounts, hourly rules or ferries. No real-time closures, height or weight restrictions or driving time (route finding is not connected).

Evidence from the run

2026-10-23: box 29.8 × 40.0 km; toll stations “Ferde AS - Bomringen Bergen” 38/88; Knappetunnelen 6,500 m (2010, subsea); Sotrabrua 1,236 m (1971).

UC-051Svalbard guide: place names and polar night in LongyearbyenNorwegian Polar InstituteMET Norwayguide and tour leader, Tourism operator on Svalbard · Transport, travel and tourism
WorkflowVerified 23 Oct 20262 tools

Sources Norwegian Polar Institute, MET Norway

Situation

The guide needs correct place names with explanations and wants to know when the sun disappears. According to the integration’s README, the place name register of the Norwegian Mapping Authority (Kartverket) has only the settlements on Svalbard, while the Norwegian Polar Institute has its own register with tens of thousands of names.

The question

“Which official place names are there within 3 km of Longyearbyen centre and what do they mean, and is it polar night on 21 December?”

How the AI does it

  1. Norwegian Polar Institute

    npolar_search_placenames(lat 78.2232, lon 15.6267, radiusKm 3, maxTreff 6) returns official names nearest first, with type and definition.

  2. MET Norway

    met_get_sun_and_moon(lat, lon, dato “2026-12-21”) shows the state (“moerketid” or “midnattssol”), the height of the sun and the moon.

  3. For the origin of a name there is npolar_get_placename (not run: the text can mention people).

Result

29 names within 3 km (Longyearbyen 154 m, Sysselgarden, Skjeringa, Svalbard kirke, Lia and more), each with a definition and a link to the Norwegian Polar Institute’s page. On 21 December it is polar night: the sun is 11.7° below the horizon at 11:55, while the moon is 35.8° high at 21:37.

Why it matters

Correct, approved names with credit (CC BY 4.0) in guide scripts and signs, and planning around light and darkness without looking things up in several sources.

Limits

The definitions come in English by default (Nynorsk is available). No avalanche, ice or polar bear warnings and no weather forecast for Svalbard are fetched here. The origin text for a name can mention people with years.

Evidence from the run

2026-10-23: 29 names within 3 km (Longyearbyen 154 m); met_get_sun_and_moon for 2026-12-21: state “moerketid”, height of the sun −11.66° at 11:55, moon 35.81° at 21:37.

UC-052Oslo–Bergen train on Monday and deviations before departureEnturtravel coordinator in a company, Mid-sized company with a lot of business travel · Transport, travel and tourism
WorkflowVerified 23 Oct 20262 tools

Sources Entur

Situation

The coordinator is booking a train trip to Bergen on Monday morning and wants to see departure, arrival and whether there are deviations. The ticket apps do not always show deviations clearly.

The question

“How do I get by train from Oslo S to Bergen stasjon on Monday 26 October from 07, and are there any deviations at Vy and Ruter right now?”

How the AI does it

  1. Entur

    entur_plan_trip with fraSted Oslo S, tilSted Bergen stasjon, transportmiddel tog and tid 2026-10-26T07:00: departure, arrival, duration and number of changes.

  2. Entur

    entur_get_deviations with kodeverk VYG (Vy) and RUT (Ruter): deviations in force, with lines, stops and time periods.

  3. Ask the AI to summarise and suggest an alternative if a deviation affects the line (F4).

Result

Departure 08:25, arrival 15:00 (395 minutes, direct train F4) and a list of deviations: none at Vy, 42 active at Ruter.

Why it matters

Travel advice and deviations in one answer, without opening several apps.

Limits

Future departures only have the timetable; real-time data only comes on the day. No ticket prices or purchase. Without transportmiddel, flights can come first. Ruter’s deviations do not apply to the Bergen trip, so they should be filtered by line or stop.

Evidence from the run

Run 2026-10-23: Oslo S–Bergen stasjon 2026-10-26: departure 08:25, arrival 15:00, 395 minutes, 0 changes, line F4. Vy (VYG) had 0 deviations; Ruter (RUT) had 42 deviations in force and 11 upcoming, for example line 64, which does not stop at three stops near Stovner because of roadworks.

UC-053The municipality’s climate indicators: new car sales and electricity use per residentStatistics NorwayElhubclimate adviser in a municipality, Municipality that reports on its climate plan · Nature, environment and climate
WorkflowVerified 23 Oct 20263 tools

Sources Statistics Norway, Elhub

Situation

The climate adviser needs fixed indicators for the climate plan: the electric car share of new car sales and electricity use per resident. The figures are held by Statistics Norway (SSB, the vehicle register) and Elhub and have to be combined with population figures.

The question

“What share of new passenger cars in Steinkjer, Trondheim and Oslo were electric in 2025, and how much electricity do households in Steinkjer use per resident at the moment?”

How the AI does it

  1. Statistics Norway

    ssb_search_tables “kjøretøy kommune” returns table 12906 (vehicles registered for the first time by municipality and fuel type, 2019–2025).

  2. Statistics Norway

    ssb_get_data 12906 with Region 5006, 5001 and 0301, Drivstofftype * and Tid 2025. The share is calculated from the row “Personbiler” per fuel type.

  3. ElhubStatistics Norway

    get_municipality_electricity_use for 5006 fetches Elhub’s consumption (households, business, industry) and Statistics Norway’s population figures in one call.

Result

Electric passenger cars as a share of new registrations in 2025: Steinkjer 97.2% (616 of 634), Trondheim 96.9% (6,397 of 6,603) and Oslo 96.3% (22,625 of 23,493). In Steinkjer, 2.97 GWh was used over two days (20–21 October 2026): industry 42%, households 37% and business 21%. Households used 46.2 kWh per resident over the two days (24,007 residents), that is about 23 kWh per day.

Why it matters

Climate reporting gets indicators that can be updated with one question, and the municipality sees that new car sales are now electric almost everywhere, so attention can move to the car fleet, charging and electricity use.

Limits

First-time registrations are new car sales, not the whole car fleet. Statistics Norway has no greenhouse gas emissions per municipality (only national tables), so municipal emissions have to be fetched from the Norwegian Environment Agency (not tested here). The Elhub figures are a short window from two to three days back, not annual consumption.

Evidence from the run

Run 2026-10-23: Statistics Norway 12906 electric passenger cars of total 2025: Steinkjer 616/634, Trondheim 6,397/6,603, Oslo 22,625/23,493; get_municipality_electricity_use(5006): 2,966,839 kWh over 48 hours, households 46.2 kWh per resident.

UC-054Red-listed species and protected areas around a planning area on JærenArtsdatabankenNorwegian Environment Agencybiodiversity adviser, Consultancy that prepares the knowledge base for an impact assessment · Nature, environment and climate
WorkflowVerified 23 Oct 20262 tools

Sources Artsdatabanken, Norwegian Environment Agency

Situation

The consultant has to describe the biodiversity at a planning area by Orrevatnet in Klepp. Records of threatened species are in Artskart, while protected areas, habitat types and Ramsar sites are in several maps at the Norwegian Environment Agency.

The question

“We are assessing a planning area at Orre in Klepp. Which threatened species (CR, EN, VU) have been recorded since 2023 within a box of a few kilometres, and which protected areas, habitat types and Ramsar sites are within two kilometres?”

How the AI does it

  1. Artsdatabanken

    artsdata_search_observations with a box (58.70–58.74 N, 5.50–5.58 E), status CR, EN and VU and fraDato 2023-01-01 returns the total number of records and the first page (treffPerSide 5).

  2. Norwegian Environment Agency

    miljodir_check_location(lat 58.72, lon 5.54, radius 2000) returns 38 hits across eleven sections.

  3. A box over 25 km is rejected with the error code INVALID_QUERY: use a smaller box or a municipality number.

Result

A knowledge base with the number of records of threatened species and which ones (willow tit, hen harrier, goshawk and rook on the first page, reported via Norsk Ornitologisk Forening/Birdlife), the protected areas Orrevatnet Nature Reserve (1996) and Jærstrendene (2003), the Ramsar site “Jæren våtmarkssystem”, habitat types with value A (Orrevatnet, Horpestadvatnet, Vassmyra) and water bodies in poor condition.

Why it matters

The consultant gets a structured starting point in minutes and sees where fieldwork and additional mapping are needed.

Limits

Artskart has individual records, not population sizes, and does not sort by date. Observer and free text are not included, and the position accuracy is up to 300 m. The answer is the first page of 17,756 records, so it has to be paged through or narrowed down. Absence of records is not absence of the species, and sensitive species are kept out of the Norwegian Environment Agency’s layers. Must be supplemented with a field survey.

Evidence from the run

2026-10-23: 17,756 CR/EN/VU records since 2023-01-01 in a 4.4 × 4.6 km box (first page: willow tit VU, hen harrier EN, goshawk VU, rook VU, all in Klepp); miljodir_check_location returned 38 hits, among them Orrevatnet Nature Reserve (protected 1996-12-20) and Ramsar “Jæren våtmarkssystem”; a box of 33 km returned INVALID_QUERY.

UC-055Alien species with very high risk in Bergen: a priority listArtsdatabankennature management officer, Municipality (nature and environmental protection) · Nature, environment and climate
LookupVerified 23 Oct 20261 tools

Sources Artsdatabanken

Situation

The municipality has to set priorities for controlling harmful alien species and needs records from the last season with species, place and accuracy, without searching through the whole of Artskart.

The question

“Which alien species with a very high risk assessment (SE) have been recorded in Bergen since 1 June 2025, how many records are there, and which sources do they come from?”

How the AI does it

  1. Artsdatabanken

    artsdata_search_observations(kommunenummer “4601”, status [“SE”], fraDato 2025-06-01, treffPerSide 5) returns the total number and the first page.

  2. Page through with `side` for more, or narrow down with taxonId (from artsdata_search_taxa) to one species.

Result

2,113 records. First page: broom, garden lupin, hybrid yew, sycamore and goat’s beard, all reported by Norsk botanisk forening in June 2025 with position accuracy of 10–150 m.

Why it matters

A quick ranking of where control has the greatest effect, and the records have coordinates and accuracy.

Limits

The order is Artskart’s own, not by date, and there are at most 100 records per page. The category is the species’ status on the list today. Finder and free text are not shown. Only registered records: absence is not proof of absence. Garden plants may be included.

Evidence from the run

2026-10-23: 2,113 SE observations in Bergen since 2025-06-01; first page: broom (Cytisus scoparius), garden lupin (Lupinus polyphyllus), sycamore (Acer pseudoplatanus), all from Norsk botanisk forening.

UC-056Reuse of shelf infrastructure: shut-down fields, facilities and pipelinesNorwegian Offshore Directorateadviser on CCS and transition, Consultancy for the energy transition · Nature, environment and climate
WorkflowVerified 23 Oct 20263 tools

Sources Norwegian Offshore Directorate

Situation

The adviser is considering whether shut-down shelf infrastructure can be reused for CO2 storage or shore power. She first needs an overview of what has been shut down, removed or abandoned.

The question

“How many fields, facilities and pipelines on the Norwegian shelf are shut down or being decommissioned, and which fields are examples?”

How the AI does it

  1. Norwegian Offshore Directorate

    sodir_search_fields with grupperPaa status: fields split into producing, shut down and approved for production.

  2. Norwegian Offshore Directorate

    sodir_search_fields with status Shut down and grense 5: examples of shut-down fields.

  3. Norwegian Offshore Directorate

    sodir_search_facilities with grupperPaa fase: facilities by phase.

  4. Norwegian Offshore Directorate

    sodir_search_pipelines with grupperPaa fase: pipelines by phase.

  5. Ask the AI to make a table and suggest filters for the next step (field, operator, medium).

Result

A table with the number of fields, facilities and pipelines per phase, and examples of shut-down fields.

Why it matters

A factual starting point for further mapping of reuse, without opening one fact page at a time.

Limits

The MCP has no layers for CO2 storage, geology or seismic data, and no ownership shares, technical condition or depths for pipelines either. Phase in the register is not an assessment of suitability. 79 facilities have no phase in the register.

Evidence from the run

Run 2026-10-23: 142 fields (102 in production, 29 shut down, 11 approved for production; examples Atla, Brynhild, Cod, Edda, Flyndre). 1,229 facilities (891 in operation, 94 removed, 41 shut down, 22 decommissioned, 20 laid up). 83 pipelines (70 in operation, 8 decommissioned, 3 abandoned in place, 2 under installation).

UC-057Low-income children: own municipality versus the country in the public health overviewFHImunicipal public health coordinator, Municipality that prepares an overview of health status and the factors that affect it · Health, care and public health
WorkflowVerified 23 Oct 20264 tools

Sources FHI

Situation

The coordinator has to prepare the municipality’s public health overview and find which groups need action. The statistics banks of the Norwegian Institute of Public Health (FHI) have over a hundred tables with technical names and municipal figures at several levels, so it takes time to find the right table and code.

The question

“What share of children (0–17 years) in Steinkjer and Bodø live in households with persistent low income, compared with the country? Find the right table at FHI.”

How the AI does it

  1. FHI

    fhi_list_sources shows the source “nokkel” (Folkehelsestatistikk, Helsedirektoratet), which has municipal figures.

  2. FHI

    fhi_search_tables kilde=nokkel (119 tables) shows that the titles are technical, for example table 403 “Vedvarende_lavinntekt_kommunegrense”.

  3. FHI

    fhi_get_table_info 403 shows the dimensions GEO (country → county → municipality → city district), AAR (three-year windows), ALDER, INNVKAT, MAAL (EU50 and EU60) and MEASURE_TYPE (count, share, ratio).

  4. FHI

    fhi_get_data 403 with GEO 0, 5006 and 1804, AAR siste(1), ALDER 0_17, INNVKAT 0, MAAL ant_eu60 and MEASURE_TYPE RATE.

Result

In 2022–2024, 10.1% of children aged 0–17 in the country lived in households with persistent low income (EU60, municipal median), compared with 7.4% in Steinkjer and 7.5% in Bodø. The same table can give figures by immigrant background and for city districts.

Why it matters

The public health overview gets a figure with source and definition in a few minutes, and the coordinator can go on to subgroups and city districts without opening FHI’s statistics bank.

Limits

The measure variable is called “EU60 (kommunal median)”, so the low-income threshold appears to be set from the municipality’s own median income; read FHI’s metadata (fhi_get_table_info with inkluderMetadata) before the figures are used to compare municipalities. FHI hides small numbers (value null with a flag). The latest period is the three-year window 2022–2024.

Evidence from the run

Run 2026-10-23: nokkel/403 (published 2026-10-06) returned 2022–2024, 0–17 years: Whole country 10.08%, Steinkjer 7.42%, Bodø 7.54%.

UC-058Weekly infection surveillance: norovirus by countyFHIinfection control adviser in a health trust, Health trust responsible for infection control and emergency preparedness · Health, care and public health
MonitoringVerified 23 Oct 20262 tools

Sources FHI

Situation

Outbreaks of vomiting and diarrhoea in nursing homes and kindergartens require rapid alerts. The adviser checks the statistics bank of the Norwegian Institute of Public Health (FHI) by hand every week to see whether norovirus is increasing in the county.

The question

“How many positive norovirus samples have there been per county in the last four weeks, and is there an increase in my county? Tell me if there are more positives than the week before.”

How the AI does it

  1. Table 769 “Oversikt gastroenterittvirus” in the source msis has weekly samples by agent, age group and county. A search in msis for “kikhoste” returned no hits (the titles use other words), so the tables have to be listed or searched with other words.

  2. FHI

    fhi_get_data msis/769 with Agens Norovirus, xPrøveåruke siste(4) and MEASURE_TYPE Antall_positive.

  3. Run weekly (the table was last published on 21 October 2026) and let the AI compare the last week with the previous one and alert on an increase.

Result

Week 39 of 2026: 35 positive norovirus samples in the whole country, 5 in Oslo, 5 in Akershus and 4 each in Vestfold, Rogaland and Vestland. The response holds 512 cells for the last four weeks (age groups and geography).

Why it matters

Early warning without manual work, and a weekly report with source and dated week for the infection control doctor and the management of nursing homes and kindergartens.

Limits

Covers laboratory-confirmed positive samples (not all cases) at county level, with a lag in reporting. No municipal figures. Small numbers may be hidden (flag), and weeks must be compared with care.

Evidence from the run

Run 2026-10-23: msis/769 (published 2026-10-21) returned for week 2026-39: 35 positive norovirus in the whole country, Oslo 5, Akershus 5, Vestfold 4; 512 cells for the last four weeks.

UC-059Air quality, traffic and contaminated ground at a kindergarten siteMET NorwayKartverketNVDB (Norwegian Public Roads Administration)+1municipal property adviser, Municipality considering premises for a new kindergarten · Health, care and public health
WorkflowVerified 23 Oct 20263 tools

Sources MET Norway, Kartverket, NVDB (Norwegian Public Roads Administration), Norwegian Environment Agency

Situation

Before the municipality rents or buys premises for a kindergarten, it wants to check air quality, traffic load and whether the outdoor area is near registered contaminated ground. The data is held by the Norwegian Meteorological Institute (MET Norway), the Norwegian Public Roads Administration and the Norwegian Environment Agency.

The question

“What is the air quality hour by hour today at Nedre Bakklandet in Trondheim, how much traffic is there in the street, and is there any registered contaminated ground or industry nearby?”

How the AI does it

  1. MET Norway

    met_get_air_quality(lat, lon, timer 12) returns an hourly forecast for the basic statistical unit (Rosenborg 3): air quality index, NO2, PM10, PM2.5 and O3.

  2. KartverketNVDB

    describe_location returns AADT 1,900 (2021), speed limit 30 km/h and 3% heavy vehicles.

  3. Norwegian Environment Agency

    miljodir_check_location(radius 150) shows sites with contaminated ground (degree of impact and status) and any industrial and treatment plants.

Result

Air quality “lite” (low, index 1.45–1.65) in the morning, but a peak of “moderat” (2.27) at 17 with NO2 60 µg/m³, PM10 76 and PM2.5 15. Low traffic (AADT 1,900), at least 8 contaminated ground sites within 150 m (one with “ikke akseptabel forurensning” and unresolved status) and no industrial or waste water plants.

Why it matters

A factual basis for assessing the outdoor area and airing (times to avoid), and pointers to fact sheets that must be followed up before a lease.

Limits

The MET Norway figures are a model forecast for the basic statistical unit, not measurements at the kindergarten, and here only for 12 hours. Noise, radon and sun conditions are not included. A registration of contaminated ground does not mean that the condition is harmful, so fact sheets and any measures must be read. AADT is from 2021, and the list of contaminated ground is truncated.

Evidence from the run

2026-10-23: basic statistical unit Rosenborg 3: index 1.45–1.65 “lite” and 2.27 “moderat” at 17 (NO2 60.1; PM10 76.2; PM2.5 15.3 µg/m³); AADT 1,900; at least 8 contaminated ground sites within 150 m (one “ikke akseptabel”, unresolved).

UC-060Benchmark research: UiB versus UiO, NTNU and UiT, with channel levelCristinHK-dir (DBH)R&D manager / research administration, University (central R&D department or faculty) · Education, research and culture
WorkflowVerified 23 Oct 20264 tools

Sources Cristin, HK-dir (DBH)

Situation

The management wants a quick look at how its own institution is doing compared with the other broad-based universities. The figures are in two different systems (DBH and Cristin) and in the register of publication channels, and it takes hours to put them together by hand.

The question

“How have the University of Bergen’s publication points developed from 2021 to 2024 compared with UiO, NTNU and UiT? How many scientific articles did UiB register in Cristin in 2024, and what level does Tidsskrift for samfunnsforskning have?”

How the AI does it

  1. Cristin

    cristin_search_institutions (navn “Universitetet i Bergen”) returns Cristin id 184.

  2. HK-dir (DBH)

    hkdir_search_tables (sok “publiseringspoeng”) finds DBH table 373 (“Vitenskapelig publisering, ikke skjermet”) and 374 (the same with channel name and quality level).

  3. HK-dir (DBH)

    hkdir_get_data on table 373 with utvalg Årstall between 2021 and 2024 and Institusjonskode 1110, 1120, 1130 and 1150, grupperPa Institusjonskode and Årstall.

  4. Cristin

    cristin_search_results with institusjon 184, kategori ARTICLE and publisertFraAar = publisertTilAar = 2024 returns totaltAntallTreff.

  5. HK-dir (DBH)

    hkdir_get_data on the register of publication channels (table 851, journals) with Original tittel like %samfunnsforskning% returns the level per year and Open Access status.

IDs used: DBH table 373 and 851, the institution codes 1110 (UiO), 1120 (UiB), 1130 (UiT), 1150 (NTNU), Cristin institution 184.

Result

A comparison table with publication points and percentage change per institution, the number of Cristin articles for 2024 and the channel level of a chosen journal, with source references.

Why it matters

R&D management gets a basis for decisions in minutes and sees that the decline is shared across the sector, not specific to its own institution.

Limits

DBH publication points and Cristin counts measure different things and cannot be compared directly. “Antall publikasjoner” in table 373 does not count each publication once (UiO 2024: 14,163 against 4,921 author shares), so use publication points or author shares. The figures are not adjusted for size. DBH requires a source reference and that AI-generated presentation is labelled.

Evidence from the run

Run 2026-10-23: table 373 returns UiB 7,526 publication points in 2021 and 6,382 in 2024 (−15%), and in 2024 UiO 12,819, NTNU 12,187 and UiT 3,947. cristin_search_results (institution 184, ARTICLE, 2024) returned 1,364 hits. Table 851 returned Tidsskrift for samfunnsforskning with level 1 (2025 and 2026) and Open Access DOAJ.

UC-061The Spanish flu in Norwegian newspapers 1918–1920: full-text searchNational Library of Norwayhistorian / local historian, University, local history society or museum · Education, research and culture
WorkflowVerified 23 Oct 20261 tools

Sources National Library of Norway

Situation

The historian wants to see how much the local newspapers wrote about the Spanish flu and which issues everyone can read. Searching the online service of the National Library of Norway and counting hits per year by hand is cumbersome.

The question

“How many newspaper issues in the National Library of Norway mention “spanskesyken” in 1918, and how many in 1919–1920? Give me links to three open examples from different parts of the country.”

How the AI does it

  1. National Library of Norway

    nb_search_items with sok “spanskesyken”, sokI “fulltekst”, mediatype “aviser”, aar 1918 and aarTil 1920 returns totaltAntallTreff.

  2. Repeat with aar 1918 alone (treffPerSide 1 to save space) and subtract to get 1919–1920.

  3. Each record has tilgang.fulltekstApen and a link; pick the ones that can be read without a Norwegian IP address.

  4. The AI puts together a small time series and a list of links. At most 5,100 records per search can be read, so narrow it down by year and media type.

IDs used: no fixed IDs; the search uses mediatype “aviser” and years.

Result

The number of newspaper issues per period and a list of open links to specific newspaper issues from 1918 and 1919.

Why it matters

Mapping the sources takes minutes, and the selection for an article, exhibition or research note can be chosen from open sources without logging in.

Limits

Full-text search returns newspaper issues where the OCR text contains the word, not the number of mentions, and the hit does not show why it matched; OCR errors cause both missed and false hits. The MCP gives only metadata and links (never the newspaper text), and according to the README extracts are refused for newspapers. Newspapers contain named private individuals, so use them for analysis of topics and spread, not to map individuals.

Evidence from the run

Run 2026-10-23: nb_search_items (fulltekst, aviser, 1918–1920) returned 5,310 hits, 1918 alone 2,110 (so 3,200 in 1919–1920). The first three hits (Østfinmarkens Folkeblad and Skotfos Avis 6 November 1918, Tvedestrand og Omegns Avis 14 February 1919) have tilgangFra EVERYWHERE, lisens ccbyncnd and a link to nb.no.

UC-062Literature search: Norwegian articles on AI with open archive and journal levelCristinHK-dir (DBH)master’s student, University or university college · Education, research and culture
WorkflowVerified 23 Oct 20262 tools

Sources Cristin, HK-dir (DBH)

Situation

The student is writing a master’s thesis on artificial intelligence in higher education and needs Norwegian peer-reviewed articles from recent years that can be read without a paywall. The student also wants to know whether the journals count as academic channels.

The question

“Find Norwegian academic articles since 2024 with “kunstig intelligens” in the title, with a link to an open archive where there is one, and check whether the journals are level 1 or 2.”

How the AI does it

  1. Cristin

    cristin_search_results with tittel “kunstig intelligens”, kategori ARTICLE and publisertFraAar 2024 returns a list of hits (30 hits) with DOI, journal, NVI level and link to an open archive.

  2. Repeat in English (“artificial intelligence”) and narrow it down by institution or year.

  3. HK-dir (DBH)

    hkdir_get_data on the channel register (table 851 for journals, 850 for publishers) returns the level per year and Open Access status for a journal.

  4. The AI makes an annotated bibliography with DOI, level and open link.

IDs used: category code ARTICLE, channel register table 851.

Result

An annotated bibliography with DOI, journal, channel level and link to an open archive.

Why it matters

The student saves hours on searching and quality checks and finds sources that can be read without a paywall.

Limits

Cristin covers only results registered by Norwegian institutions and gives no full text, only links. The title search only matches the words in the title, and the results cannot be sorted by relevance (only newest, oldest or category). Channel level applies to Norwegian publication channels. Author names must not be used to search for people.

Evidence from the run

Run 2026-10-23: cristin_search_results (tittel “kunstig intelligens”, ARTICLE, from 2024) returned 30 hits; the first two are articles in Magma forskning og viten (2026-10-14, NVI level 1, with a link to an open archive), among them “Kunstig intelligens i regnskap og revisjon: Hvilken kompetanse blir viktig for nyutdannede?”. Table 851 returned Tidsskrift for samfunnsforskning with level 1 and Open Access DOAJ.

UC-063Freely available editions: check the licence and find pages in open booksNational Library of Norwaylibrarian, Public or school library · Education, research and culture
WorkflowVerified 23 Oct 20264 tools

Sources National Library of Norway

Situation

The library is making a digital exhibition on polar history and wants to link to books that everyone can read, wherever they live. Otherwise each edition has to be checked on nb.no, one by one, to see whether it is open or can only be read in Norway.

The question

“Find editions of “Fram over Polhavet” that are freely available to everyone, show the licence and subject terms, and find the pages where the word “isen” appears in the 1897 edition.”

How the AI does it

  1. National Library of Norway

    nb_search_items with sok “Fram over Polhavet”, mediatype “bøker”, kunFriVerk true and kunDigitalt true returns 9 hits, among them the 1897 edition and two volumes from 1942.

  2. National Library of Norway

    nb_get_item returns subject terms (polarekspedisjoner, arktis), place, place of publication, extent and tilgang.fulltekstApen.

  3. National Library of Norway

    nb_get_item_fragments (sok “isen”) returns the page numbers where the word appears; the tool only works for open books and journals.

  4. National Library of Norway

    nb_get_related_items checks whether the work is linked to other records (empty for this work).

IDs used: Sesam ID 2abe617e50dd1c905885396f563ea1a5 (URN:NBN:no-nb_digibok_2009060910001).

Result

The exhibition list with links, licence and subject terms, and page numbers for selected words.

Why it matters

The library can build reading material for the whole country without opening each record by hand, and can reuse subject terms and licence information in its own cataloguing.

Limits

The extracts are only the word and a few characters with the page number, not running text; the text itself is read on nb.no. nb_get_related_items is empty for multi-volume works like this one. The catalogue gives both antallSider (681, scanned pages) and extent (553 pp., printed), which must not be confused. Most of the collection is protected by copyright and can only be read in Norway.

Evidence from the run

Run 2026-10-23: nb_search_items returned 9 hits; the first hit (URN:NBN:no-nb_digibok_2009060910001, published 1897 by Aschehoug) has friVerk true, lisens publicdomain and fulltekstApen true. nb_get_item returned subject terms and extent 553 pp., nb_get_item_fragments returned page numbers 26 and 33 for “isen”, and nb_get_related_items returned an empty list.

UC-064Weekly tender alert: new communications and advertising tendersOwn pageDoffinbid manager at a communications agency, Agency that supplies the public sector · Selling to the public sector and business development
MonitoringVerified 23 Oct 20262 tools
Open the scenario →

Sources Doffin

Situation

Notices for advertising, PR and communications services are spread across several CPV codes, and the deadlines are short. The agency finds the tenders too late, or spends time reading notices without knowing whether they fit.

The question

“Show all open tenders for advertising, PR and communications services in Doffin right now. Include the buyer, deadline, estimated value and how the bids will be evaluated, and tell me which ones are worth going for.”

How the AI does it

  1. Doffin

    Every Monday: doffin_search_notices with cpv 79340000 (advertising and marketing), type konkurranse, status aktiv and fra = the previous Monday.

  2. Repeat for cpv 79416000 (PR) and 79341000, since buyers code differently. The Norwegian Industrial Property Office (Patentstyret), for example, uses 79416000.

  3. Doffin

    doffin_get_notice on each new notice for award criteria, duration, qualification deadline and link to the tender documents.

  4. The AI ranks by value, deadline and fit and writes a short weekly report.

Result

A weekly list. Today there are five active hits under 79340000: Vygruppen AS, advertising agency services (NOK 80 million, deadline 9 Nov), Ruter AS (NOK 42 million, 3 Nov), Harstad kommune and others (NOK 4.5 million, 16 Nov), Kartverket (NOK 2 million, 9 Nov) and a dynamic purchasing system. In addition there is Patentstyret’s framework agreement under 79416000 (deadline 10 Nov).

Why it matters

The agency sees deadlines and values in the same week the notice comes out, and does not have to read every notice to see whether it fits. That leaves more time for the tenders it will actually put effort into.

Limits

The weighting of the criteria is not always in the notice: Patentstyret gives criteria without weights, and Vygruppen gives none. The tender documents are held by the buyer (Mercell, Ivalua) and are not read here. Patentstyret has two almost identical notices that have to be merged, and status aktiv also includes schemes that run until 2030.

Evidence from the run

Run 2026-10-23: doffin_search_notices cpv 79340000, type konkurranse, status aktiv returned totalt 5. doffin_get_notice 2026-115718 (Patentstyret) returned four criteria without weights, 48 months and deadline 2026-11-10; 2026-115844 (Vygruppen) returned 80,000,000 NOK and an empty list of criteria.

UC-065Which municipalities should we visit first? Growth in the elderly as a sales signalStatistics Norwaysales manager for welfare technology (personal alarms), Supplier to the municipal health and care sector · Selling to the public sector and business development
WorkflowVerified 23 Oct 20261 tools

Sources Statistics Norway

Situation

The sales team has to set priorities for which municipalities get a visit before the budget round. They want to start where the elderly population is growing fastest and use of services is high, but the figures are in several tables.

The question

“Rank Steinkjer, Trondheim and Bodø by how much the number of people aged 80+ rises from 2026 to 2040, and show what share of the over-80s use home-based services in Steinkjer. Which municipality should we call first?”

How the AI does it

  1. Statistics Norway

    ssb_get_data 14746 (regional projections) with Region 5006, 5001 and 1804, Alder agg_Funksjonell5, ContentsCode Personer and Tid 2026 and 2040.

  2. Statistics Norway

    ssb_get_data 12209 (KOSTRA health and care) with the region variable KOKkommuneregion0000 for use of services.

  3. The AI ranks by percentage and absolute growth and suggests a priority list.

Result

Number of people aged 80+ from 2026 to 2040: Trondheim 9,025 → 15,471 (up 71%, 6,446 more), Bodø 2,649 → 4,337 (up 64%) and Steinkjer 1,666 → 2,362 (up 42%). Trondheim grows most in both absolute and relative terms; in Steinkjer 32.6% of the over-80s use home-based services.

Why it matters

The sales team gets a data-driven ranking in minutes, and the same setup can be run for more municipalities with the list of municipality codes (max 1,000 rows per call).

Limits

Only three municipalities have been tested; a ranking of all municipalities has to be done in several calls. The number of older people is not the same as a need to buy: budget, decision-making authority and existing agreements are outside the data. The projection is the main alternative from Statistics Norway (SSB).

Evidence from the run

Run 2026-10-23: Statistics Norway 14746, 80+ from 2026 to 2040: Trondheim 9,025 → 15,471, Bodø 2,649 → 4,337, Steinkjer 1,666 → 2,362; KOSTRA 12209 Steinkjer 32.6% of the over-80s with home-based services.

UC-066Competitor analysis: which public contracts do the agencies win?EnhetsregisteretDoffinTEDtender manager at a communications agency, Agency that sells to the public sector · Selling to the public sector and business development
Combined toolVerified 23 Oct 20263 tools

Sources Enhetsregisteret, Doffin, TED

Situation

The agency wants to compare its own public awards with a competitor’s, to see buyers, sizes and who it comes up against. Doffin, the Norwegian procurement database, cannot search by winner directly, so the work is cumbersome and is rarely done.

The question

“Compare which public framework agreements TRY AS and Dentsu Norge AS have won in Doffin. Who are the buyers, how large are the agreements, and how did the buyer weight the last one TRY won?”

How the AI does it

  1. EnhetsregisteretDoffin

    get_company_procurements with organisation number 980129942 (TRY AS) and 927445875 (Dentsu Norge AS). The tool looks up the entity in Enhetsregisteret, the Central Coordinating Register for Legal Entities, and finds the awards by exact registered name.

  2. Doffin

    doffin_get_notice on the latest award (2026-116097, NTNU): number of bids, weighting and duration.

  3. TED

    ted_search_notices with land NOR, cpv 79340000 and type resultat shows the awards above the EU threshold (369 hits), with the winner names in the list itself.

  4. The AI makes a table: number of awards, buyers, ceiling and which criteria decide.

Result

TRY AS has 12 awards in Doffin as supplier, among them NTNU (NOK 12 million, 2026), Norad (NOK 8.1 million, 2026) and Vygruppen (media agency services, 2025). Dentsu Norge AS has 2, but large ones: Entur (NOK 90 million) and Innovasjon Norge (NOK 300 million). The NTNU agreement was weighted 20% price and 80% quality, with 12 bidders.

Why it matters

The agency sees the pattern: many medium-sized agreements versus a few large ones, which buyers choose quality over price, and who it comes up against. That gives a better basis for which tenders to go for and how the bid should be built.

Limits

The supplier side matches on exact registered name: a winner whose name is written differently in the notice is not found, and sole proprietorships are never searched. The ceiling is not revenue, and agreements that have not been advertised are not included.

Evidence from the run

Run 2026-10-23: get_company_procurements 980129942 returned somLeverandor.totalt 12 (NTNU 2026-10-16, Norad 2026-02-09, Vygruppen 2025-11-14); 927445875 returned totalt 2 (Entur NOK 90 million, Innovasjon Norge NOK 300 million). doffin_get_notice 2026-116097 returned antallTilbud 12 and Pris 20 / Kvalitet 80. ted_search_notices land NOR, cpv 79340000, resultat returned totalt 369 with winner TRY AS on 715551-2026.

UC-067Supplier to the hospitals: open procurements at SykehusinnkjøpDoffinsales manager at a supplier of medical equipment, Supplier to the health trusts · Selling to the public sector and business development
MonitoringVerified 23 Oct 20262 tools

Sources Doffin

Situation

The hospitals buy through Sykehusinnkjøp HF, and the notices are many and detailed. The salesperson needs a weekly overview of what is open, who it is being bought for and which deadlines apply.

The question

“Which procurements does Sykehusinnkjøp HF have open now? List the three newest with the deadline, which health trust it is being bought for and the estimated value.”

How the AI does it

  1. Doffin

    doffin_search_notices with kjoper Sykehusinnkjøp HF and status aktiv: 63 open tenders.

  2. The response says which buyer name was used and suggests other names (such as the Midt-Norge division), which have to be searched separately.

  3. Doffin

    doffin_get_notice on the ones that fit, for lots, criteria and link to the tender documents.

  4. Run weekly with a fra date, and let the AI send new hits to the sales team.

Result

63 open tenders. The newest are urological products for Helse Nord (21 lots, NOK 12 million, deadline 26 Nov), replacement of the RO plant for dialysis at Helse Bergen (deadline 23 Nov) and replacement of lifts at Martina Hansens Hospital (deadline 23 Nov).

Why it matters

The salesperson picks up deadlines without following several portals, and quickly sees which lots fit the products.

Limits

Buyers with several names in Doffin, the Norwegian procurement database, are separate buyers: Sykehusinnkjøp’s divisions and the air ambulance must be searched separately. The tender documents (Ivalua) are not read, and many lack an estimated value.

Evidence from the run

Run 2026-10-23: doffin_search_notices kjoper Sykehusinnkjøp HF, status aktiv returned totalt 63, kjoper.brukt SYKEHUSINNKJØP HF and two other names suggested, and 2026-116477 (urological products, 21 lots, 12,000,000 NOK, deadline 2026-11-26).

UC-068Cod landings in Vardø: value, volume and price per kilo 2024–2025Directorate of Fisheriesbusiness development officer, Coastal municipality in Finnmark · Selling to the public sector and business development
WorkflowVerified 23 Oct 20263 tools

Sources Directorate of Fisheries

Situation

The municipality wants to document the importance of fisheries in an application for business development funding. The landing figures are in a table with 38 million cells that is hard to query without knowing the codes.

The question

“How much cod was landed in Vardø in 2024 and 2025, and what was the value?”

How the AI does it

  1. Directorate of Fisheries

    fiskeridir_stat_search_tables: find F06009b, catch by Norwegian vessels by species and gear per landing municipality (2018–2025).

  2. Directorate of Fisheries

    fiskeridir_stat_get_table_info with table F06009b and variable Art with sok torsk: the code 0201001 Torsk, and the variables Fangstår, Landingskommune, Redskap and Statistisk variabel.

  3. Directorate of Fisheries

    fiskeridir_stat_get_data with the selection Fangstår 2024 and 2025, Landingskommune 5634* and Art 0201001 Torsk: round weight (tonnes) and catch value (NOK 1,000).

  4. Ask the AI to calculate the price per kilo and change from 2024 to 2025.

Result

2025: 2,894 tonnes worth NOK 133.5 million. 2024: 3,278 tonnes worth NOK 118.1 million. The price per kilo rose from NOK 36.0 to NOK 46.1.

Why it matters

The municipality gets figures with a source for the application and can repeat the query for other species and municipalities.

Limits

Landing municipality is where the catch was landed, not the vessel’s home municipality. The label 5634 VARDØ (2024-) suggests that earlier years are under other municipality codes. Catch value is first-hand value, not export value, and the price per kilo is calculated here and not given by the source.

Evidence from the run

Run 2026-10-23: cod landed in Vardø in 2024 was 3,277.71 tonnes and 118,053.61 (NOK 1,000), in 2025 it was 2,893.85 tonnes and 133,516.21 (NOK 1,000). The first attempt returned Cloudflare 502, a new attempt worked.

UC-069EU regulation to Norwegian law: the Machinery Regulation and arbeidsmiljølovenOwn pageStortingetEUR-LexLovdatalawyer in an industry association for machinery manufacturers, Industry association in the technology industry · Law and compliance
WorkflowVerified 23 Oct 20265 tools
Open the scenario →

Sources Stortinget, EUR-Lex, Lovdata

Situation

Members ask when the EU’s new Machinery Regulation applies in Norway and which Norwegian rules are changed. The answer is in four sources (the government’s list of cases, EUR-Lex, Lovdata and Stortinget, the Norwegian parliament), and the lawyer has to search each one separately.

The question

“The government is going to put forward an amendment to arbeidsmiljøloven on the EU Machinery Regulation. When will the case come to Stortinget, when does the regulation apply in the EU, and which provisions in current Norwegian rules deal with machinery?”

How the AI does it

  1. Stortinget

    stortinget_hent_ventede_saker: find case 110403 “Endringer i arbeidsmiljøloven - sikre tilstrekkelig hjemmel for gjennomføring av EUs maskinforordning i norsk rett” from Arbeids- og inkluderingsdepartementet, expected October 2026.

  2. EUR-Lex

    eurlex_get_act with celex 32023R1230 (the Machinery Regulation): iKraftFra, eeaRelevant and endretAv show what applies when, and that the regulation has been amended by 32026R1744.

  3. Lovdata

    lovdata_get_law with id LOV-2005-06-17-62 (arbeidsmiljøloven) and sok “maskin” returns the sections that mention machinery (§ 4-4, § 5-5, § 6-2, § 7-2 and one more).

  4. Lovdata

    lovdata_get_provision with id LOV-2005-06-17-62 and paragraf 5-5 returns the text on the duties of manufacturers, importers and suppliers of machinery, with the authority to issue regulations in the seventh paragraph.

  5. Lovdata

    lovdata_search_laws with sok “maskiner” and type forskrift (page 2) finds Maskinforskriften, FOR-2009-05-20-544.

IDs used: case 110403, CELEX 32023R1230 and 32026R1744, LOV-2005-06-17-62, FOR-2009-05-20-544.

Result

A short memo with the timing of the Norwegian amendment, the EU dates of entry into force, which sections of arbeidsmiljøloven and which regulation are affected, all with links.

Why it matters

The lawyer does not have to put four systems together by hand and can send a first draft of a letter to members the same day. The risk of using the wrong date or the wrong section goes down.

Limits

The Lovdata text is an extract from the open data packages and can be older than the current wording, so check lovdataUrl before you quote. The tools do not give preparatory works, case law or local regulations. EUR-Lex shows that the regulation is EEA-relevant, but not whether and when it has been incorporated into the EEA Agreement. The list of announced cases gives only title, ministry and month, not the content of the proposal.

Evidence from the run

Run 2026-10-23: announced case 110403 (October 2026). EU 2023/1230 has iKraftFra 2023-07-19, 2024-01-20, 2024-07-20, 2026-10-20 and 2027-01-20, is EEA-relevant and amended by 2026/1744 and 2024/2748. Lovdata returned arbeidsmiljøloven § 5-5 (eight paragraphs) and Maskinforskriften (last amended 2026-01-01), extract fetched 2026-10-23.

UC-070Annual accounts: Norges Bank’s exchange rate at 31 DecemberNorges Bankaccountant/auditor, Accounting firm with export clients · Law and compliance
LookupVerified 23 Oct 20261 tools

Sources Norges Bank

Situation

Monetary items in foreign currency have to be translated in the annual accounts, and the rate has to come from an official source that the auditor can verify. 31 December and 1 January can be days without a rate.

The question

“What was Norges Bank’s USD/NOK rate on 31 December 2025, and which date applies if the day falls on a public holiday?”

How the AI does it

  1. Norges Bank

    norgesbank_get_exchange_rates with valuta=USD, fra=2025-12-29 and til=2026-01-05, frekvens dag.

  2. Check that the date is in the response; public holidays and weekends have no rate, so the previous working day is used.

  3. Without arguments the tool returns the latest rate for about 40 currencies, also as an overview of which currencies are available.

Result

USD/NOK on 31 December 2025: 10.0791 (29 Dec: 10.0518; 30 Dec: 10.0587; 2 January 2026: 10.0661; 5 January: 10.099). 1 January 2026 has no rate (public holiday).

Why it matters

The auditor gets a verifiable source and date, and the same setup works for all currencies and balance sheet dates, also monthly and annual averages (frekvens maaned and aar).

Limits

Gives the rate, not the accounting rules: which rate to use (balance sheet date, transaction date or average) follows from regnskapsloven and good accounting practice. Norges Bank’s terms require a source reference and that the content is not changed. Currencies that have ceased to exist (Bulgarian lev after 2 January 2026) have their last rate on 31 December 2025.

Evidence from the run

Run 2026-10-23: USD/NOK 2025-12-31 = 10.0791 and 2026-01-02 = 10.0661; no rate for 1 January 2026.

UC-071Customer due diligence at onboarding and over time: company, roles and statusEnhetsregisteretAML officer at an accounting firm, Accounting and advisory firm · Law and compliance
WorkflowVerified 23 Oct 20262 tools

Sources Enhetsregisteret

Situation

Hvitvaskingsloven requires customer due diligence at onboarding and ongoing monitoring. The firm checks status, roles and address by hand in several windows, and rarely follows up the portfolio after onboarding.

The question

“Do customer due diligence on ADMENTO AS (989176013): is the company active, who holds roles such as managing director, board and auditor, and where is it located? Then check my whole client portfolio for bankruptcy and liquidation.”

How the AI does it

  1. Enhetsregisteret

    brreg_get_company_profile with organisation number 989176013 and medPosisjon true returns the entity, roles and geocoded address.

  2. Enhetsregisteret

    brreg_search_enheter with organisation numbers as a comma-separated list (all the clients) returns the status per entity: konkurs, underAvvikling and underTvangsavviklingEllerTvangsopplosning.

  3. Enhetsregisteret

    Run step 2 weekly. brreg_search_enheter with konkurs true (and kommunenummer) shows the bankruptcy date for new cases.

  4. The AI makes a control form with findings and flags changes since last time.

Result

The company is active (no bankruptcy or liquidation), is located at Heimdal in Trondheim, has 313 employees and is part of a group. Roles: managing director, chair of the board, two board members and BDO AS as auditor. The portfolio check covered three companies in one call.

Why it matters

The due diligence is documented quickly, and the portfolio monitoring can run automatically. The firm finds out about a client’s bankruptcy or liquidation in the same week it happens.

Limits

No shareholders or beneficial owners, no PEP or sanctions lists and no date of birth: only what is registered about roles. Role holders who are individuals must be handled under the data protection rules. The status is the register’s, not a risk assessment.

Evidence from the run

Run 2026-10-23: brreg_get_company_profile 989176013 returned 5 roles (DAGL, chair of the board, two board members, auditor BDO AS 993606650) and position 63.3613, 10.3706. brreg_search_enheter with three organisation numbers returned totalElements 3, all with konkurs false. konkurs true for AS in Trondheim returned 114 hits with a bankruptcy date (for example 2025-11-06).

UC-072What has changed in the AI Act this year? Compliance check against EUR-LexEUR-Lexcompliance officer, Norwegian software company with AI products · Law and compliance
WorkflowVerified 23 Oct 20262 tools

Sources EUR-Lex

Situation

The company has built its compliance plan on the AI Act (EU) 2024/1689, but has heard that it has been simplified. The adviser needs to know what the amendment is called, when it applies and whether it is marked as EEA-relevant.

The question

“Has the AI Act been amended since 2024, what is the amending act called, when did it enter into force, and is it marked as relevant for the EEA? Which other regulations does it amend?”

How the AI does it

  1. EUR-Lex

    eurlex_search_acts with sok “artificial intelligence” (optionally iKraft true and eeaRelevant true) finds CELEX 32026R1744.

  2. EUR-Lex

    eurlex_get_act with celex 32026R1744 returns the title, iKraftFra, eeaRelevant and which acts it amends (32024R1689, 32023R1230 and 32018R1139).

  3. EUR-Lex

    eurlex_get_act with celex 32023R1230 (the Machinery Regulation) shows that it has been amended by 32026R1744.

Result

A short memo with the amending act, entry into force, EEA marking and the acts affected, with a link to the legal text.

Why it matters

The compliance plan is not built on outdated wording, and the company gets a date to work towards.

Limits

EEA marking means that the EU wrote the act with the EEA in mind, not that it has been incorporated into the EEA Agreement or applies in Norway. The tool gives metadata and a link, not the legal text or Norwegian deadlines for adaptation.

Evidence from the run

Run 2026-10-23: Regulation (EU) 2026/1744 of 2026-07-08 (Digital Omnibus on AI): iKraftFra 2026-07-27, eeaRelevant true, amends 2024/1689, 2023/1230 and 2018/1139.

UC-073Licence and credit line check before the client deliveryproject manager or lawyer, Agency delivering a data-driven solution for a client · Law and compliance
LookupVerified 23 Oct 20261 tools

Situation

Before launch the team needs to know which sources can be reused, what the credit line must say and which sources lack a clarified licence.

The question

“We use figures from Statistics Norway, Elhub, Statnett and the Norwegian Directorate of Elections in a client launch. What must we credit, and which sources have an unclear licence?”

How the AI does it

  1. Call get_data_licences without an argument, or with kilde for each source

  2. Read licence, credit, terms page and certainty (bekreftet, indirekte, fra_repoet, ikke_funnet)

  3. Let the AI write the credit text and a risk list of the sources that need to be clarified

Result

A table per source with licence, credit text, terms page and how certainly the licence has been established.

Why it matters

Avoids the client publishing data without the right credit, or building on a source without a clarified licence.

Limits

An overview of what has been read on the sources’ pages, not legal advice: check the terms before you promise anything to a client. The licence applies to the data, not to the personal data in it.

Evidence from the run

Run 2026-10-23: 39 entries. Statistics Norway and Elhub have CC BY 4.0 with the credit “Kilde: …”. Statnett and the Norwegian Directorate of Elections have no licence found, and Statnett’s API can be shut down without notice. In total 19 confirmed, 11 indirect, 8 with no licence found and 1 from the repo.

UC-074Situation overview in one minute when the warnings go outOwn pageKartverketNorwegian Coastal AdministrationMET Norway+1crisis communications adviser on call, Consultancy with an emergency preparedness agreement for a client with operations along the coast · Agencies, marketing and communications
Combined toolVerified 23 Oct 20261 tools
Open the scenario →

Sources Kartverket, Norwegian Coastal Administration, MET Norway, NVE

Situation

When a storm or flood is forecast, the adviser needs to know what warnings have actually been issued where the client operates, before the first statement is written. The warnings are held by the Norwegian Meteorological Institute (MET Norway), the Norwegian Water Resources and Energy Directorate (NVE) and the Norwegian Coastal Administration.

The question

“Give me a combined situation overview for Bergen now: weather warnings from MET Norway, avalanche, flood and landslide warnings from NVE, and navigation warnings at sea nearby.”

How the AI does it

  1. KartverketNorwegian Coastal AdministrationMET NorwayNVE

    Call get_warnings_for_place with the place (and sjoRadiusKm if the client has ships or a quay)

  2. Read hoyesteNivaa and the sections vaer, snoskred, flom, jordskred and sjo

  3. Let the AI write a holding statement and a list of what needs to be followed up

Result

One combined answer with the highest warning level and each source separately, with sources and licence, as a basis for the first statement.

Why it matters

One call instead of checking yr.no, varsom.no and the Norwegian Coastal Administration in turn when time is short.

Limits

Not a substitute for the official warnings. A section that is null means that the service is down or does not cover the place, not that no warning has been issued. NVE warns per municipality and region, and the Norwegian Coastal Administration’s times lack a time zone.

Evidence from the run

Run 2026-10-23 for Bergen: highest level 1 (green), no weather warnings from MET Norway, green level for flood and landslide in the municipality, the avalanche region Hordalandskysten “not assessed”, no navigation warnings within 10 km.

UC-075Meeting preparation: municipal profile before a meeting with the mayor of BodøKartverketStatistics NorwayNorwegian Directorate of Electionspublic affairs adviser at a communications agency, Consultancy with a client that wants to launch a project in Nordland · Agencies, marketing and communications
Combined toolVerified 23 Oct 20261 tools

Sources Kartverket, Statistics Norway, Norwegian Directorate of Elections

Situation

The agency is preparing the client’s first meeting with the mayor and the chief municipal executive in Bodø. Population, neighbouring municipalities and election results are held by the Norwegian Mapping Authority (Kartverket), Statistics Norway (SSB) and valgresultat.no, and the municipal profile is usually made by hand.

The question

“Make a one-page municipal profile of Bodø for a meeting with the mayor: population trend over the last five years, neighbouring municipalities, and who won the 2025 Storting election and the 2023 municipal council election in the municipality.”

How the AI does it

  1. KartverketStatistics NorwayNorwegian Directorate of Elections

    describe_municipality with kommunenummer 1804 fetches name, county and neighbouring municipalities (Kartverket), population on 1 January for the last five years (Statistics Norway table 07459) and the latest Storting and municipal council elections (Norwegian Directorate of Elections) in one call.

  2. The AI puts the figures into a profile with headings for population, neighbouring municipalities and the political picture.

  3. The field “sources” in the response is used as the source line in the profile.

Result

Bodø (Nordland) has 10 neighbouring municipalities and 53,638 inhabitants on 1 January 2026, down from 53,725 the year before (the peak year 2025). Storting election 2025: Ap 30.8%, FrP 23.7% (up 11.2 percentage points), Høyre 16.0%, turnout 79.1%. Municipal council election 2023: Høyre 34.3% and 14 of 39 seats, Ap 16.9%.

Why it matters

The profile is ready in seconds and can be made the same way for all the country’s municipalities, so the agency can prepare a whole round of municipalities without spending a day per municipality.

Limits

Only party shares and turnout at municipal level, not the people behind them or local issues (journal entries and political decisions require other tools). The seats field is 0 for the Storting election at municipal level and must be ignored. The municipal council election is from 2023, and the next one is in 2027.

Evidence from the run

Run 2026-10-23: describe_municipality(1804) returned 10 neighbouring municipalities, 53,638 inhabitants (2026), Ap 30.8% and FrP 23.7% in the 2025 Storting election and Høyre 34.3% (14 seats) in the 2023 municipal council election.

UC-076Labour market profile for a recruitment campaign in KristiansandStatistics Norwayemployer branding strategist at an agency, Agency with an industrial client that is recruiting in Sørlandet · Agencies, marketing and communications
WorkflowVerified 23 Oct 20262 tools

Sources Statistics Norway

Situation

The client lacks skilled workers and wants to build an employer campaign. The strategist needs facts about how many people live and work in the region, how many commute in and what it costs to move there.

The question

“How large is the labour force in Kristiansand, how many work there without living there, and how have house prices in the region developed over the last year? Use it to say whether the campaign should be aimed at commuters or at people from outside.”

How the AI does it

  1. Statistics Norway

    ssb_search_tables “befolkning alder kommune” returns table 11618 (population and employed persons, municipality, sex and age, 4th quarter 2009–2025).

  2. Statistics Norway

    ssb_get_data 11618 with Region 4204 (Kristiansand), Alder *, ContentsCode * and Tid set to the latest year returns population by place of residence and employed persons by place of residence and by place of work for 15–74 and 20–66 years.

  3. Statistics Norway

    ssb_get_data 07221 (price index for existing dwellings, 2015=100) with Region * and Tid 2025K3 and 2026K3 returns the region “Agder og Rogaland uten Stavanger”.

  4. The AI calculates net in-commuting and the employment rate and writes a recommendation on the target group.

Result

In 2025 there were 63,645 employed persons with their place of work in Kristiansand, but only 58,973 employed residents (15–74 years): nearly 4,700 more jobs than employed residents, that is net in-commuting. The employment rate for 20–66 years is 73.8% (53,990 of 73,130). House prices in the region Agder and Rogaland except Stavanger rose 6.9% over the last year (147.4 → 157.5).

Why it matters

The campaign can be aimed at commuters or people moving in, and the client gets facts to build the message on instead of assumptions. The same setup works for any municipality.

Limits

Table 11618 only counts 15–74 and 20–66 years and says nothing about occupation, skills or what the candidates want. Education level (09429), industry (07091, 13470) and pay (12852) are in other municipal tables that were not run here. The house price region is wider than the municipality.

Evidence from the run

Run 2026-10-23: Statistics Norway 11618 Kristiansand 2025, 15–74 years: 63,645 employed by place of work against 58,973 by place of residence; 07221 Agder og Rogaland uten Stavanger 147.4 (2025K3) → 157.5 (2026K3).

UC-077Press story on house prices with the right figures and credit lineOwn pageStatistics NorwayPR adviser, Communications agency with an estate agent as client · Agencies, marketing and communications
WorkflowVerified 23 Oct 20262 tools
Open the scenario →

Sources Statistics Norway

Situation

The client wants a regional press release on the housing market. The adviser has to find the latest official figures, calculate the change and give the source and licence correctly, or the story falls at the journalist’s first source check.

The question

“Find the latest price index for existing dwellings from Statistics Norway for Bergen, Oslo and the rest of the country, calculate the change over the last 12 months, and write three press angles with a credit line.”

How the AI does it

  1. Statistics Norway

    ssb_search_tables “prisindeks brukte boliger” returns table 07221 (quarterly, latest 2026K3, updated 12 October 2026) and 07230 (annual).

  2. Statistics Norway

    ssb_get_data 07221 with Region *, Boligtype 00 (all dwelling types) and Tid 2026K3, and one more call for 2025K3.

  3. The AI calculates the change per region and formulates angles. “sources” in the response provides the credit line: Statistisk sentralbyrå, table 07221, licence CC BY 4.0.

Result

Last 12 months (Q3): Bergen up 11.2% (147.9 → 164.4), Møre og Romsdal and Vestland except Bergen up 7.1%, Agder and Rogaland except Stavanger up 6.9%, Northern Norway up 6.6%, the whole country up 3.3%, Oslo including Bærum up 0.1% (170.3 → 170.5). Possible angle: “Bergen climbs 11 per cent while Oslo stands still”, with Oslo still at the highest level (170.5 against 158.3 for the whole country).

Why it matters

The story can be produced in an hour with the right figures, the right period and a credit line that stands up to the journalist’s source check.

Limits

The index is regional (12 regions), not municipality or city district, and shows price trends, not NOK per square metre. Unadjusted figures for Q3 must be compared with unadjusted figures for Q3; seasonally adjusted figures show somewhat different growth. The licence is CC BY 4.0, so the source must be credited.

Evidence from the run

Run 2026-10-23: Statistics Norway 07221 Bergen 147.9 (2025K3) → 164.4 (2026K3) = +11.2%; Oslo med Bærum 170.3 → 170.5; Whole country 153.2 → 158.3.

UC-078Pitch dossier: what agency services has the prospect bought, and when?DoffinRegnskapsregisteretbusiness developer at an advertising and communications agency, Agency in communications, advertising and media · Agencies, marketing and communications
WorkflowVerified 23 Oct 20263 tools

Sources Doffin, Regnskapsregisteret

Situation

The agency is considering pitching to Entur AS and wants to know how large the company is, what agency services it has bought, who won and how long it is bound by the agreement. Today that means searching Doffin, the Norwegian procurement database, and the accounts separately.

The question

“We are considering pitching to Entur. What media and advertising services have they bought in the last year, who won, how many bidders were there and how long does the agreement run? How large is Entur financially?”

How the AI does it

  1. Doffin

    doffin_search_notices with cpv 79340000 (advertising and marketing services) and type resultat: 290 hits today. Entur’s framework agreement for media consultancy and media buying (2026-114473) is among the newest.

  2. Doffin

    doffin_get_notice 2026-114473: winner, number of bids, ceiling, duration and timeline for the notices.

  3. Regnskapsregisteret

    regnskap_get_accounts for Entur AS (917422575) with antall 2: revenue, margin and equity ratio for 2025 and 2024.

  4. The AI calculates the growth and assesses the timing: the agreement is binding for 3+1+1+1 years, so the next procurement comes around 2029 at the earliest.

Result

A short client dossier: Entur AS had revenue of NOK 840 million in 2025 (+12.5% from 2024) with a 3.3% operating margin. The media agency agreement (ceiling NOK 90 million, 3+1+1+1 years) went to Dentsu Norge AS among 7 bidders in September 2026. Media buying for Entur is therefore locked until around 2029, and pitch time should be spent on clients with open tenders.

Why it matters

The agency avoids spending pitch resources on a client that is bound for several years, and can aim at clients that have something open. The dossier is ready in one conversation instead of a search across several sources.

Limits

The notice gives no award criteria (forsyningsforskriften), and the winner’s own value is not given, only the ceiling. The accounts are figures as submitted, without notes. No contact persons, and agreements that have not been advertised are not included.

Evidence from the run

Run 2026-10-23: doffin_get_notice 2026-114473 returned winner Dentsu Norge AS, antallTilbud 7, ceiling 90,000,000 NOK and notice 2026-04-17 to award 2026-09-16. regnskap_get_accounts 917422575 returned operating revenue 840,055,398 NOK in 2025 against 746,714,342 in 2024.

UC-079Full moon over Reinebringen: light, weather and correct creditKartverketMET Norwayphotographer and content producer, Agency producing tourism content for a client in Lofoten · Agencies, marketing and communications
WorkflowVerified 23 Oct 20262 tools

Sources Kartverket, MET Norway

Situation

The team is going to photograph the full moon and the blue hour over Reine and needs times and directions for the sun and moon, a cloud forecast and a credit line that can go in the caption. Today the times come from one website and the weather from another, and the licence has to be found separately.

The question

“We want to photograph the full moon over Reinebringen in Moskenes on Monday 26 October. When does the sun set and the moon rise, in which direction, what are the cloud cover and wind like, and which credit line should we use?”

How the AI does it

  1. KartverketMET Norway

    get_place_weather(sted “Reinebringen”, timer 12) looks up the place (mountain in Moskenes, 448 m a.s.l.) in the place name register of the Norwegian Mapping Authority (Kartverket) and returns an hourly forecast and weather warnings. For Monday, timer up to 240 is used.

  2. MET Norway

    met_get_sun_and_moon(lat 67.9328, lon 13.0885, dato 2026-10-26) returns sunrise and sunset, moonrise and moonset, direction and moon phase.

  3. The credit lines are in the responses: “Data fra Meteorologisk institutt (MET Norway), lisens NLOD 2.0 og CC BY 4.0”, and place names from Kartverket (CC BY 4.0, “© Kartverket”).

Result

Photo plan: on 26 October the sun rises at 07:53 and sets at 15:48 (azimuth 237°), the moon rises at 14:20 in the north-east (azimuth 32°) and the moon phase is 177°, that is nearly full. Today’s forecast shows 94–100% cloud cover and gusts of up to 11.7 m/s on the mountain, with no weather warnings.

Why it matters

Two calls give time, direction, weather and credit line and replace three websites, and the credit line is correct from the start.

Limits

Sun and moon are calculated without terrain: the mountains around Reine can hide both at times. The forecast three days ahead is uncertain and has 6-hour resolution after about two days. No avalanche or safety warning for the mountain hike and no nowcast has been run.

Evidence from the run

2026-10-23: Reinebringen is a mountain in Moskenes (1874), 448 m a.s.l., cloud cover 94–100% and gusts 11.7 m/s; 2026-10-26: sunrise 07:53, sunset 15:48, moonrise 14:20 (azimuth 31.9°), moon phase 177.2°.

UC-080Pitch preparation: what does Statkraft Energi AS own and operate?NVEbusiness developer at an agency, Communications agency · Agencies, marketing and communications
WorkflowPartly verified 23 Oct 20261 tools

Sources NVE

Situation

The agency is going to pitch to a power company and needs a factual basis on what the client actually owns before the first meeting. The power plants are in the register of the Norwegian Water Resources and Energy Directorate (NVE).

The question

“Make a fact sheet on the hydropower of Statkraft Energi AS for the pitch. What is Kvilldal, and who owns it?”

How the AI does it

  1. NVE

    nve_get_hydropower_plant with id 222 (Kvilldal): head, discharge capacity, ownership shares and licences.

  2. Ask the AI to write one A4 page with a source line (NVE) and suggested talking points.

Result

A fact sheet on Kvilldal (1,240 MW) with ownership shares and licences.

Why it matters

The pitch is built on verified figures instead of press releases.

Limits

Shows registered ownership and capacity for one power plant, not finances, strategy or group structure. Employees, wind power and grid areas are not included since get_organisation_footprint was removed from the MCP.

Evidence from the run

Run 2026-10-23: Kvilldal (1,240 MW): Statkraft Energi AS 72%, Lyse Kraft DA 18%, Otra Kraft DA 7.3%. The rest of the fact sheet (employees, number of plants, grid areas) came from get_organisation_footprint, which has been removed, and has been taken out.

UC-081Campaign images with the right credit line from the National Library of NorwayNational Library of Norwaycontent producer / art director, Communications agency with a tourism client · Agencies, marketing and communications
WorkflowVerified 23 Oct 20262 tools

Sources National Library of Norway

Situation

The agency is making a campaign for a coastal destination and wants to use historical images of Hurtigruten. Which images are free to use, and what the credit line should say, have to be cleared image by image on nb.no.

The question

“Find historical images of Hurtigruten that the National Library of Norway has marked as free to use (public domain), and make a credit list with photographer or owner, licence and link for each image.”

How the AI does it

  1. National Library of Norway

    nb_search_items with sok “Hurtigruten”, mediatype “bilder” and kunFriVerk true returns the list of hits.

  2. Narrow it down with aar and aarTil and kunDigitalt true for the period the campaign will draw on.

  3. National Library of Norway

    nb_get_item for each chosen image returns skapere, utgiver, tilgang.lisens (publicdomain), friVerk and lenke.

  4. The AI builds a credit table with id, URN, creator or owner, licence and link, and suggests a credit line.

IDs used: mediatype “bilder”; sample records 0c1e101b496ad1b40ac9acc55363afee, e474a48ed3732bae2a02af4f6c597a75 and d1a42d6158cf1982b313ab7804959e6a.

Result

A finished credit table for selected images, with a link to the original at the National Library of Norway.

Why it matters

The agency avoids rights errors and saves checking each image by hand before production.

Limits

The MCP gives metadata and a link, not the image file itself, so the image has to be downloaded from nb.no. The National Library of Norway’s licence marking is its own; a lawyer should clear commercial use, especially for more recent photographers. The first three hits have no year of publication in the catalogue. Recognisable people in the images need a separate assessment.

Evidence from the run

Run 2026-10-23: nb_search_items (Hurtigruten, bilder, kunFriVerk) returned 25 hits. The first three have licence publicdomain, friVerk true, tilgangFra EVERYWHERE and a link to nb.no, two of them from Norsk folkemuseum (NF.WF 02802 and 02803). nb_get_item on a book shows the fields lisens, skapere, utgiver and lenke.

UC-082Steer the campaign by the weather where the shops areKartverketMET Norwaymedia planner / performance adviser, Media agency with a retail chain as a client · Agencies, marketing and communications
WorkflowVerified 23 Oct 20261 tools

Sources Kartverket, MET Norway

Situation

The client wants to switch creative and move budget between cities according to the weather over the next few days, but the forecasts are fetched by hand from Yr, city by city.

The question

“Check the weather for the next 24 hours in Oslo, Bergen and Tromsø: temperature, precipitation, wind and any weather warnings. Suggest which creative (rain, sun, wind) we should give priority to in each city tomorrow.”

How the AI does it

  1. KartverketMET Norway

    Call get_place_weather for each city (coordinates, place or address) with timer=24

  2. Read the summary (min and max temperature, max wind and gusts, total precipitation) and weather warnings

  3. Let the AI make a table of city × creative trigger with a short reason

Result

A table per city with temperature range, precipitation, wind and warnings, and a recommendation on which creative should get the budget.

Why it matters

The campaign matches the weather people actually have, and the basis for the decision takes minutes instead of checking city by city by hand.

Limits

Forecast, not observation. The tool does not connect to the ad platforms: the budget change itself is made there. The Norwegian Meteorological Institute (MET Norway) must be credited (NLOD 2.0 and CC BY 4.0).

Evidence from the run

Run 2026-10-23 at 09–10 for the next 24 hours from 10: Oslo 7.3–11.6 °C and 3.4 mm precipitation, Bergen 5.4–10.4 °C and 1.5 mm, Tromsø 7.1–10.6 °C, 0 mm and gusts up to 12 m/s. No weather warnings in any of the cities.

UC-083Find local news angles on electricity consumption per residentElhubStatistics NorwayPR adviser, PR agency working for an energy company or a heat pump supplier · Agencies, marketing and communications
WorkflowVerified 23 Oct 20261 tools

Sources Elhub, Statistics Norway

Situation

The local press wants local figures and new angles, but the consumption is held by Elhub, the population figures by Statistics Norway (SSB), and the calculation per resident has to be done by hand.

The question

“Compare household electricity consumption per resident in Oslo, Bergen and Vestvågøy for 20 October. What angle can we pitch to the local press, and which caveats need to be included?”

How the AI does it

  1. Call the tool for each municipality (0301, 4601 and 1860) with dato 2026-10-20 and dager=1

  2. Compare husholdningKwh per resident (the consumption group “private” is households)

  3. Let the AI suggest an angle and make a list of caveats, with Elhub and Statistics Norway as the source

Result

kWh per resident per day for each municipality, split into households, business and industry, with sources and licence, ready for a press release or a map.

Why it matters

A local and verifiable news angle in minutes, with the source credits in place.

Limits

Elhub publishes about two days behind, and one day says little about annual consumption. The figure divides by permanent residents, so municipalities with many holiday homes or a lot of industry can be skewed. The reason for the differences (heating, type of housing, climate) is not in the data.

Evidence from the run

Run 2026-10-23 for 20 October: households used 26.4 kWh per resident in Vestvågøy (11,606 residents), 18.5 in Bergen and 15.8 in Oslo. Total per resident 46.5, 35.4 and 33.1 kWh.

UC-084The right name in the right language: Sami place names before printingKartverketcopywriter or language adviser at an agency, Agency with a client that is going to advertise in Finnmark and Troms · Agencies, marketing and communications
LookupVerified 23 Oct 20261 tools

Sources Kartverket

Situation

Ads and signs in Northern Norway must use official names, and for many places both a Norwegian and a Sami name are approved. Finding out what is approved and given priority means looking up the place name register of the Norwegian Mapping Authority (Kartverket), place by place.

The question

“We are going to print ads for Karasjok and Tromsø. What are the approved spellings in Kartverket’s place name register, in which languages, and what are the municipality and county called?”

How the AI does it

  1. Kartverket

    kartverket_sok_stedsnavn with sok “Karasjok” returns the urban settlement with spelling, language, spelling status and the municipality and county names as Kartverket writes them.

  2. Also search for the Sami name (“Kárášjohka”): the same place is listed with a North Sami spelling, and other places with the same name (a river) are included.

  3. Let the AI make a list of names with spelling, language and status per place, and include the credit line from Kartverket: “Alle stedsnavn er hentet fra Sentralt stedsnavnregister (SSR) ©Kartverket”.

Result

A list of names per place with approved spelling, language and status, and the municipality and county names in the same form as Kartverket uses them.

Why it matters

The right name in the right language in ads and signs, with a source, without searching place by place.

Limits

The register gives the approved spelling and status, not advice on which language the campaign should use. The search also hits other places with the same name, so look at type and municipality. Names that are not registered are not here, and Kartverket requires credit for place names.

Evidence from the run

Run 2026-10-23: “Karasjok” returned the urban settlement (place number 868181, the spelling “godkjent og prioritert”, Norwegian) in the municipality “Kárášjohka - Karasjok” in the county “Finnmark - Finnmárku - Finmarkku”. “Kárášjohka” returned two hits: the same urban settlement with a North Sami spelling, and a river. “Tromsø” returned the city in the county “Troms - Romsa - Tromssa”.

UC-085Media training: real questions from question time and how the minister answeredStortingetmedia trainer, Consultancy that gives media training to managers in health trusts and public organisations · Agencies, marketing and communications
WorkflowVerified 23 Oct 20263 tools

Sources Stortinget

Situation

Media training works best when the questions are real. The questions put to a minister in question time are short and thematic, and the answers show how a case can be defended, but finding them for a topic means searching several parts of stortinget.no.

The question

“Write ten realistic journalist questions to a hospital director on bed capacity and emergency preparedness, based on what Stortinget has actually asked the minister of health and care services in question time in 2025–26. Show how the minister answered one of the questions, and which follow-up contributions were made.”

How the AI does it

  1. Stortinget

    stortinget_hent_sporretimesporsmal with sesjonid 2025-2026 and til “helse” lists the questions (109 answered questions to the minister of health and care services), with topic and dates.

  2. Stortinget

    stortinget_hent_enkeltsporsmal returns the full text of the question. The answer to an oral question is not there, but in the transcript.

  3. Stortinget

    stortinget_sok_i_referat with query “sengeplasser” and publikasjonid for the day of question time (refs-202526-05-27) returns the minister’s answer and follow-up contributions, with speaker, party and time.

  4. Let the AI turn the questions into journalist questions and make a line of response and follow-up for the training.

Result

A training set: ten realistic questions sourced from question time, the minister’s line of response on one of them and the follow-up questions that were asked.

Why it matters

The training is built on actual questions and answers instead of guesses, and the set can be made for each topic in a short time.

Limits

The questions are asked by members and not by journalists. The transcript ID has to be found first (the day the question was answered), and the search requires all the words to be in the same contribution. The extracts are short: read the whole contribution before quoting anything. Members and ministers hold public roles.

Evidence from the run

Run 2026-10-23: 109 answered question time questions to the minister of health and care services in 2025–26. The four newest (27 May 2026) are about, among other things, 7,000 beds for the wounded in war, maternity wards in summer and the cancer clinic in Bergen. The transcript refs-202526-05-27 (208 contributions) returned 7 hits on “sengeplasser”, among them the minister’s answer that 7,000 beds are to be made available in a crisis without standing empty at all times.

UC-086Fix the rate on the shared debt? Interest rate, CPI and house prices togetherOwn pageNorges BankStatistics Norwaychair of a housing cooperative board (voluntary role), Housing cooperative in Oslo with a floating rate on its shared debt · Citizens and everyday life
WorkflowVerified 23 Oct 20263 tools
Open the scenario →

Sources Norges Bank, Statistics Norway

Situation

The board is to propose at the general meeting whether part of the shared debt should be fixed. The figures needed (policy rate, yield curve, price growth and the price level of housing) are held by Norges Bank and Statistics Norway (SSB), and no one on the board has time to search.

The question

“What has the policy rate done since 2021, where are the 3-, 5- and 10-year government bond rates now compared with the policy rate, and what have prices of existing dwellings in Oslo and price growth (CPI) done over the last year? Write a one-page decision memo for the housing cooperative.”

How the AI does it

  1. Norges Bank

    norgesbank_get_interest_rates without arguments returns the latest value for all 11 rates (dated 22 October 2026).

  2. Norges Bank

    norgesbank_get_interest_rates with rente=styringsrente, kunEndringer=true and fra=2021-09-01 returns all rate decisions since 0%.

  3. Statistics Norway

    ssb_search_tables “prisindeks brukte boliger” returns table 07221 (quarterly, latest 2026K3). ssb_get_data 07221 with Region *, Boligtype 00 and Tid 2025K3 and 2026K3 returns 12 regions; the regions Oslo including Bærum and Whole country are read out.

  4. Statistics Norway

    ssb_search_tables “konsumprisindeks” returns table 14710 (monthly, 2025=100). ssb_get_data 14710 with Tid 2025M09 and 2026M09.

Result

The policy rate is 4.5% after 18 changes since September 2021 (latest: up from 4.25% on 25 September 2026). The government bond rates are 4.767% (3 years), 4.615% (5 years) and 4.485% (10 years): the curve falls towards the long end. The house price index for Oslo including Bærum is 170.5 (up 0.1% in one year, the whole country up 3.3%), and CPI is up 3.4% from September to September.

Why it matters

The board gets a sourced basis for its decision in minutes, with date and source on every figure, instead of a patchwork of screenshots from several websites.

Limits

Norges Bank does not publish the banks’ lending rates, NIBOR or interest rate swaps, so the choice between fixed and floating rates cannot be settled here; the government bond rates are only a pointer, and forecasts (Pengepolitisk rapport) are not included. The price index applies to the region Oslo including Bærum, not to one housing cooperative.

Evidence from the run

Run 2026-10-23: policy rate 4.5% (2026-10-22) with 18 changes since 2021-09; government bonds 3/5/10 years 4.767/4.615/4.485%; Statistics Norway 07221 Oslo including Bærum 170.3 (2025K3) → 170.5 (2026K3), Whole country 153.2 → 158.3; CPI 14710 100.6 → 104.0.

UC-087Check a home before the bidding round: risk, environment, protection and public transportOwn pageKartverketNVDB (Norwegian Public Roads Administration)NVE+6home buyer, Private individual considering a flat in central Trondheim · Citizens and everyday life
WorkflowVerified 23 Oct 20268 tools
Open the scenario →

Sources Kartverket, NVDB (Norwegian Public Roads Administration), NVE, Norwegian Coastal Administration, MET Norway, Norwegian Environment Agency, Geonorge, Riksantikvaren, Entur

Situation

Before the bidding round the buyer wants to know what lies in and around the address: quick clay, flooding, radon, contaminated ground, protected buildings and public transport. That normally means eight to ten map portals and registers that each have to be read separately.

The question

“I am thinking of bidding on a flat in Nedre Bakklandet in Trondheim. Is there anything I should know about landslide and flood hazard, quick clay, radon, contaminated ground, protection of the building and how I get to and from work by bus?”

How the AI does it

  1. KartverketNVDB

    describe_location(address, kommunenummer 5001) returns coordinates, 3.5 m a.s.l. and the nearest road (30 km/h, AADT 1,900 in 2021).

  2. NVEKartverketNorwegian Coastal AdministrationMET Norway

    get_warnings_for_place and nve_get_hazard_summary return today’s warnings: flood and landslide at green level, the avalanche region “ikke vurdert” (not assessed), no weather warning from the Norwegian Meteorological Institute (MET Norway) and no coastal warnings within 10 km.

  3. Norwegian Environment Agency

    miljodir_check_location(radius 150) shows outdoor recreation areas, the water body Nidelva (chemical status “dårlig”) and sites with contaminated ground.

  4. Geonorge

    geonorge_search_datasets finds Kvikkleire (a29b905c-6aaa-4283-ae2c-d167624c08a8), Flomsoner (e95008fc-0945-4d66-8bc9-e50ab3f50401) and Radon aktsomhet (dc0605f3-2301-4abe-a91f-6da42464c281); geonorge_query_dataset queries each of them at the point.

  5. Riksantikvaren

    riksantikvaren_search_heritage_sites(radiusKm 0.3) returns heritage sites and protection status around the home.

  6. KartverketEntur

    find_departures_near(address, radiusMeter 400) returns stops and real-time departures.

Result

A one-page site profile: the address is in the quick clay mapping area (with hazard zone boundaries) of the Norwegian Water Resources and Energy Directorate (NVE) and in the analysis area “Flom Nidelva”, but outside the mapped inundation areas. The radon awareness level is low, there are at least eight contaminated ground records nearby, the building is registered as an Art Nouveau block of flats from 1908 with protection type “Vernet etter PBL”, and the stop Bakkegata is 98 m away with ten bus routes.

Why it matters

The buyer gets a basis for asking the estate agent and the surveyor the right questions (condition report, geotechnical assessment, insurance) before bidding, instead of learning eight to ten different map services.

Limits

A screening, not a technical or legal assessment. The Geonorge responses give attributes without geometry, so the distance to the zone edge must be checked in NVE’s maps. No flood hits only means that the point is not in a mapped inundation area. The radon map cannot be used for individual buildings (the dataset’s own restriction on use). No owner data, only address and cadastral number. The search of the Directorate for Cultural Heritage (Riksantikvaren) can be incomplete in dense urban areas.

Evidence from the run

Run 2026-10-23 for Nedre Bakklandet: 3.53 m a.s.l., AADT 1,900, highest warning level 1 (green); quick clay 5 hits at the address (2 hazard zone boundaries), flood: analysis area “Flom Nidelva”, radon “lav aktsomhet”; at least 8 contaminated ground sites within 150 m; 239 heritage records within 300 m; Bakkegata 98 m away with real-time departures (1–3 min late).

UC-088Who represents me in Stortinget from Sør-Trøndelag?Norwegian Directorate of ElectionsStortingetengaged voter, Private individual in Trondheim · Citizens and everyday life
Combined toolVerified 23 Oct 20262 tools

Sources Norwegian Directorate of Elections, Stortinget

Situation

A voter wants to know who was elected from the constituency, which party and which committees they sit on, to know who can take up a local issue. The information is on two websites with different names and numbers.

The question

“Who was elected to Stortinget from Sør-Trøndelag in 2025, and which committees do they sit on now? Which of them sit on the energy and environment committee?”

How the AI does it

  1. Norwegian Directorate of ElectionsStortinget

    get_elected_representatives with fylke 16 (the Norwegian Directorate of Elections’ number for Sør-Trøndelag) returns the 10 elected per party, whether they got a levelling seat and the link to the current member with committees.

  2. Stortinget

    stortinget_hent_representanter with fylke and komite (for example ENERGI) shows who from a constituency sits on a particular committee.

  3. You find the constituency number for the 2025 Storting election in underenheter from valg_get_result. Sør-Trøndelag is 16, not 50.

Result

A list of those elected with party, committees and their ID in Stortinget, the Norwegian parliament, which can be used further for votes and questions.

Why it matters

Faster than searching two websites. The voter knows who is dealing with the issue and can write to the right member.

Limits

One of those elected got no hit in Stortinget’s list (leave, government or deputy). The tool does not include candidate lists, age or personal votes. The constituency number is the Norwegian Directorate of Elections’ number, not today’s county number.

Evidence from the run

Run 2026-10-23: constituency 16 Sør-Trøndelag, 10 seats (Ap 4, FrP 2, H 1, SV 1, Sp 1 and Rødt 1 levelling seat), 1 with no hit in Stortinget’s list. The Sp member sits on Energi- og miljøkomiteen. A try with fylke 50 returned NOT_FOUND.

UC-089Avalanches before the ski tour: the warning is off, but Regobs shows today’s avalanchesNVEskier and hiker, Private individual planning an early ski tour in Lyngsalpene · Citizens and everyday life
WorkflowVerified 23 Oct 20262 tools

Sources NVE

Situation

In autumn Varsom has not started issuing avalanche warnings, but there can be snow in the mountains and avalanches can happen. The hiker wants to know what warnings have been issued and what others have actually observed in the last few days.

The question

“I want to go ski touring in Lyngsalpene near Lyngseidet this weekend. What is the avalanche warning, and have avalanches and snow conditions been recorded in the area recently?”

How the AI does it

  1. NVE

    nve_get_avalanche_warning(lat 69.58, lon 20.22, dager 2) returns the warning region (Lyngen) and danger level per day.

  2. NVE

    nve_search_observations(lat, lon, radiusMeter 30000, fare [“snoskred”], dager 365) returns the latest avalanche and snow cover observations with a link to regobs.no.

  3. nve_get_observation(regId) on the most recent ones returns the type of avalanche, size and comment (not run here).

Result

Official warning: Lyngen (region 3010) level 0 “Ikke vurdert” (not assessed) on 23 and 24 October. At the same time two avalanche events were recorded on 22 and 23 October 2026 at about 69.50 N 19.72 E, with 2–3 photos each.

Why it matters

Prevents the false conclusion “no warning means safe”. The hiker gets a fresh, local picture and links to the observations, where the observer is credited.

Limits

The warnings only cover the season and defined regions, and “not assessed” is not “low danger”. Regobs observations are voluntary and can be missing. Weather, wind and avalanche problems are not included in this case. No substitute for varsom.no and your own assessment on site.

Evidence from the run

2026-10-23: nve_get_avalanche_warning Lyngen (3010) level 0 “Ikke vurdert” for 23–24 October; nve_search_observations returned regId 468214 (2026-10-23 at 10:00) and 468210 (2026-10-22 at 16:03), both avalanche events at 69.498 N 19.717 E, and more hits on the next page.

UC-090Speed limits and the sea forecast before a trip in a leisure boatNorwegian Coastal AdministrationMET Norwayleisure boat skipper, Private individual · Citizens and everyday life
WorkflowVerified 23 Oct 20262 tools

Sources Norwegian Coastal Administration, MET Norway

Situation

A skipper is leaving Oslo harbour and wants to know what speed limit applies and what the weather will be. The rules are split between the Norwegian Coastal Administration, the municipality and the port board.

The question

“What speed limit applies to leisure boats at Oslo harbour and in Alver, and what does the sea forecast say for Skagerrak?”

How the AI does it

  1. Norwegian Coastal Administration

    kystverket_check_sea_location with lat 59.9035 and lon 10.7345 (Oslo harbour): speed limit for leisure craft under 24 m, buffer zone, regulations with link and VTS area.

  2. Repeat for another place, for example 60.70321 N 4.90897 E (Alver).

  3. MET Norway

    met_get_text_forecast with type sea_no and omraade Skagerrak: forecast and synoptic situation.

  4. Ask the AI to summarise in three sentences and link to the regulations.

Result

Speed limits with a link to the regulations and a sea forecast in plain text.

Why it matters

The rule comes with its source, instead of rumours from the jetty, and that means less risk of a fine.

Limits

The map layer is not legally binding; the linked regulations apply. The speed limit can apply only within a buffer zone from land (Oslo: 150 and 200, given without a unit). The sea forecast is text per area, not a point forecast. No AIS.

Evidence from the run

Run 2026-10-23: Oslo harbour has 25 knots for leisure craft under 24 m (buffer zone 200, Oslo havnestyre’s regulations) and VTS Horten. Alver (60.703 N 4.909 E) has 5 knots. Skagerrak (inner and outer): “Østlig liten kuling. Moderat sjø. Regn og moderat sikt.” 22–24 October.

UC-091Choosing a school: compare two schools on size, owner and well-beingUdirEnhetsregisteretparent, Household · Citizens and everyday life
Combined toolVerified 23 Oct 20263 tools

Sources Udir, Enhetsregisteret

Situation

Parents are choosing between two schools nearby and have only the schools’ own websites to go on. Open data on size, owner and the Pupil Survey exists, but is in different systems.

The question

“We are considering two schools in Bodø (give the names). How large are they, who owns them, and how many grade 9 pupils say they are bullied compared with the country?”

How the AI does it

  1. Udir

    udir_sok_enheter (register nsr, kommunenummer 1804, or sok on the school name) returns the schools’ organisation numbers.

  2. UdirEnhetsregisteret

    udir_get_school_profile returns pupil numbers, staff, grade levels, ownership structure, KOSTRA group and information from Enhetsregisteret, the Central Coordinating Register for Legal Entities, in one call.

  3. Udir

    udir_stat_get_data (table 150, TidID 202512, TrinnKode 9, Kommunekode 1804 or Organisasjonsnummer for the school) returns the share bullied; the national figure is fetched with EnhetID -12.

  4. The AI summarises and says when the figures rest on small groups.

IDs used: Statistikkbanken table 150 and TidID 202512.

Result

A simple comparison card per school: pupil numbers, staff, grade levels, owner and bullying figures compared with the national average.

Why it matters

Parents get a factual basis in one answer instead of searching several portals.

Limits

Open data has no grades, lower secondary grade points or national tests. The Pupil Survey is self-reported, and figures for grade 9 at one school vary a lot from year to year because the groups are small. Schools with few responses have hidden figures. The figures say nothing about how well the school suits your child.

Evidence from the run

Run 2026-10-23: udir_get_school_profile returned 654 pupils for one of the schools, 132 staff, grades 1–10, owner Bodø kommune and an Enhetsregisteret sub-unit registered 1995-08-10. Table 150 returned 7.7% bullied in grade 9 at one school, 11.5% at the other and 10.2% for the country (2025-26).

UC-092Prototype in a day: “What is happening around this address?”Own pageKartverketNVDB (Norwegian Public Roads Administration)Entur+1developer or creative technologist, Agency pitching a digital service to a municipality or tourism business · Developers, data and AI agents
WorkflowVerified 23 Oct 20263 tools
Open the scenario →

Sources Kartverket, NVDB (Norwegian Public Roads Administration), Entur, MET Norway

Situation

To show a service that combines place, weather, public transport and warnings otherwise takes four integrations, several keys and weeks of work.

The question

“Show what is happening around Karl Johans gate 22 in Oslo: elevation and the nearest road with speed limit, the next departures nearby and the weather for the next few hours.”

How the AI does it

  1. KartverketNVDB

    describe_location returns coordinates, municipality, elevation and the nearest road with speed limit and AADT

  2. KartverketEntur

    find_departures_near returns stops within a radius and the next departures with real time and deviations

  3. KartverketMET Norway

    get_place_weather returns the forecast and weather warnings. Ask the AI to put it together into one card

Result

An address card with terrain, road, public transport in real time and weather, each figure with source and licence.

Why it matters

The client sees a working service in a pitch meeting, not a slide.

Limits

Open data with no guarantee of uptime. Entur and the Norwegian Meteorological Institute (MET Norway) set requirements for how the client identifies itself and gives credit. Distance to the stop is a straight line, not walking distance.

Evidence from the run

Run 2026-10-23: Karl Johans gate 22 is 11.8 m a.s.l., and the nearest road for cars is 39 m away with 30 km/h (AADT missing). Two stops within 230 m (Wessels plass for bus, Øvre Slottsgate for tram and bus) with real-time departures, and one departure has a deviation message (line 13). Oslo senter: 7.3–11.6 °C and 3.4 mm over the next 24 hours.

UC-093Weekly alert: which Statistics Norway tables have been updated since last time?Statistics Norwaydeveloper of an internal news and analysis bot, Analysis department in a media house or agency · Developers, data and AI agents
MonitoringVerified 23 Oct 20261 tools

Sources Statistics Norway

Situation

The analysts want to be told when Statistics Norway (SSB) publishes new figures on prices, labour or population, but there is no subscription per table. Today someone checks the website by hand or finds out too late.

The question

“Which Statistics Norway tables on prices have been updated in the last 7 days? Write a short Slack message with table number, title, latest period and link.”

How the AI does it

  1. Statistics Norway

    Schedule a weekly call to ssb_search_tables with sok “pris” (and later “arbeid” and “befolkning”), oppdatertSisteDager 7 and maxTreff 10.

  2. Each hit has id, tittel, sistePeriode, oppdatert and lenke; the AI filters out what is not relevant.

  3. Optional next step (not run in this test): for relevant tables the bot follows up with ssb_get_table_info and ssb_get_data and puts the figures in the message with a credit line.

Result

The call found 5 tables updated in the last 7 days, among them 09654 “Priser på drivstoff (kr per liter)” (up to and including 2026M09, updated 21 October 2026) and 08264 “Konjunkturbarometeret” (2026K3, updated 22 October 2026).

Why it matters

The analysts are alerted the same day as Statistics Norway publishes, without keeping watch manually, and the bot can fetch the new figures and write a first draft of a memo or news story.

Limits

The answer shows which tables have been updated, not what has changed. The search matches words in the title, so relevant tables with other words can fall outside. “Oppdatert” is Statistics Norway’s timestamp, not the time of a press release.

Evidence from the run

Run 2026-10-23: ssb_search_tables(sok “pris”, oppdatertSisteDager 7) returned 5 hits: 09654 (updated 2026-10-21), 07405, 09242, 08264 and 12786 (updated 2026-10-22).

UC-094Policy radar: an AI agent that sends weekly alerts on the client’s topicsStortingetEUR-Lexdata and AI engineer, Agency or company building internal AI assistants · Developers, data and AI agents
MonitoringPartly verified 23 Oct 20263 tools

Sources Stortinget, EUR-Lex

Situation

The team wants to give the public affairs advisers an agent that every Monday delivers new cases, hearings and EU acts on the client’s topics. Without a single connection the agent has to call several systems with different formats.

The question

“Make a weekly report on the topic of batteries: new cases in Stortinget, questions and hearings with deadlines, EU acts with EEA marking and announced propositions. Mark what is new since last week.”

How the AI does it

  1. Stortinget

    stortinget_hent_ventede_saker (the whole list) is filtered by topic in the agent.

  2. EUR-Lex

    eurlex_search_acts with omfang “forarbeider” and aar 2026 shows the Commission’s proposals (CELEX 52026PC…), for example EU Kids Act (52026PC0681, 2026-09-17).

  3. Stortinget

    stortinget_hent_horinger returns hearing IDs and deadlines that the agent can put in the calendar.

  4. The agent stores last week’s IDs and reports only new ones. All responses have the field “sources” with source and licence, which is included in the report.

Result

A weekly topic report with links and licences, made without manual work.

Why it matters

The public affairs adviser does not have to open several websites every Monday, and the report carries source information so it can be used directly with the client.

Limits

None of the sources has a change log, so the agent has to keep track of “new since last time” itself. For Commission proposals eeaRelevant is often false or null even when the case affects the EEA.

Evidence from the run

Run 2026-10-23: eurlex_search_acts with omfang forarbeider, sok “digital” and aar 2026 returned, among others, the proposal EU Kids Act (52026PC0681, 2026-09-17). The weekly diff itself has not been built.

UC-095Coordinates and cadastral numbers for a GIS extract, without owner dataKartverketNVDB (Norwegian Public Roads Administration)GIS developer, Software house building a property portal · Developers, data and AI agents
WorkflowVerified 23 Oct 20264 tools

Sources Kartverket, NVDB (Norwegian Public Roads Administration)

Situation

The developer has WGS84 coordinates from an address list and needs UTM 33, which parcel the point is on and a check via cadastral number. The developer also wants to know what owner information the open services of the Norwegian Mapping Authority (Kartverket) actually give.

The question

“Give me the UTM 33 coordinates (EPSG:25833) for a WGS84 point in Trondheim, which property (cadastral number) the point is on, and check the cadastral number by looking it up again. What owner data do I get?”

How the AI does it

  1. Kartverket

    kartverket_transformer_koordinat with fra 4258 and til 25833 returns the coordinate in UTM 33.

  2. Kartverket

    kartverket_eiendom_naer_punkt (radius 15, maxTreff 3) returns the nearest parcels with cadastral number, type and distance.

  3. Kartverket

    kartverket_geokode_matrikkel with kommunenummer, gardsnummer and bruksnummer returns the parcel’s position and confirms the lookup.

Result

Coordinate in EPSG:25833, cadastral number (type teig, festenummer and seksjonsnummer 0) and checked position.

Why it matters

A chain (address, coordinate, parcel, cadastral number) that can be run in bulk by an AI agent or a script, without personal data.

Limits

The response has only cadastral number, type, position and distance: no owners, addresses, areas or buildings. Owner data is restricted and is not in the open services, and must not be extracted for private homes either. The point is the parcel’s representative point, not the boundary. Only EPSG codes that Kartverket supports work.

Evidence from the run

Run 2026-10-23 for one point in central Trondheim: the transformation 4258 to 25833 returned UTM coordinates, eiendom_naer_punkt returned three parcels within 3 m (nearest 0 m), geokode_matrikkel confirmed the position, and no owner fields came back in any of the responses.

UC-096Find the right dataset across Statistics Norway, Eurostat and GeonorgeStatistics NorwayEurostatGeonorgedata analyst or developer, Consultancy or agency · Developers, data and AI agents
WorkflowVerified 23 Oct 20263 tools

Sources Statistics Norway, Eurostat, Geonorge

Situation

You know the data exists, but not where: Statistics Norway (SSB), Eurostat and Geonorge each have their own catalogue and their own search.

The question

“Find datasets on house prices in Norway and the EU, and on flood zones that can be looked up at an address. Give me table ID, update date and link.”

How the AI does it

  1. Search each catalogue with simple words (Statistics Norway and Geonorge match whole words)

  2. Compare the hits on latest period and update date

  3. Go further with *_get_table_info / eurostat_get_dataset_info, or query the dataset at a point with geonorge_query_dataset

Result

List of hits with ID, title, latest period, updated date and link to the catalogue.

Why it matters

One conversation instead of three web searches, and the right table with date and link.

Limits

The search matches whole words and often has to be tried with several forms. The hits are catalogue entries, not an assessment of which table fits best.

Evidence from the run

Run 2026-10-23: Eurostat returned 22 hits on “house price” (among them prc_hpi_q, latest period 2026-Q2). Geonorge returned NVE’s “Flomsoner” with WFS and WMS on “flomsone”. Statistics Norway returned only the macro table 12880 on “boligpris”, so the word form matters.

UC-097Test your own AI assistant against ground truth from public registersStatistics NorwayEnhetsregisteretAI developer or quality manager, Company building a customer assistant · Developers, data and AI agents
WorkflowVerified 23 Oct 20262 tools

Sources Statistics Norway, Enhetsregisteret

Situation

An assistant that answers questions on companies, population or politics must be tested against something that is correct. Ground truth from public registers is verifiable and comes with a source.

The question

“Make 20 test questions with ground truth from Statistics Norway and Enhetsregisteret, for example the population of a municipality and the number of employees in a company, and give the source and retrieval date for each.”

How the AI does it

  1. Statistics NorwayEnhetsregisteret

    Fetch the ground truth with ssb_get_population (table 07459) and brreg_get_enhet for a selection of municipalities and companies

  2. Let the AI phrase the questions, and store the answer, source (`sources`) and retrieval date as ground truth

  3. Run the questions against the assistant being tested and count the discrepancies

Result

An evaluation set with questions, ground truth, source and date, which can be run again when the assistant changes.

Why it matters

Measures whether the assistant gives correct figures, with ground truth that others can check.

Limits

Register data changes: the ground truth must be dated and fetched again. Suits facts with one correct answer, not questions of judgement.

Evidence from the run

Run 2026-10-23: Oslo had 724,290 residents in 2025 and 728,714 in 2026 (Statistics Norway 07459). Equinor ASA (923609016) has 21,231 employees, industry code 06.100 and business address in Stavanger (Enhetsregisteret).

UC-098Today’s situation picture for municipal ROS: warnings, landslides and zone mapsKartverketNorwegian Coastal AdministrationMET Norway+2emergency preparedness coordinator, Mid-sized municipality (Trondheim used as an example) · Emergency preparedness and security
MonitoringVerified 23 Oct 20264 tools

Sources Kartverket, Norwegian Coastal Administration, MET Norway, NVE, Geonorge

Situation

Every morning the emergency duty officer wants to see warnings from the Norwegian Meteorological Institute (MET Norway) and the Norwegian Water Resources and Energy Directorate (NVE), which landslides and floods have actually happened recently, and which mapped hazard zones are in the city centre. The information is in Varsom, Regobs and NVE Atlas.

The question

“Give me a situation picture for central Trondheim today: current hazard warnings (weather, flood, landslide, avalanche), recorded landslides and floods in the last 90 days within 15 km, and whether the centre is in a mapped quick clay area.”

How the AI does it

  1. KartverketNorwegian Coastal AdministrationMET NorwayNVE

    get_warnings_for_place(address or coordinate, dager 2) returns the highest warning level and the parts for weather, avalanche, flood, landslide and coastal warnings.

  2. NVE

    nve_search_observations(lat, lon, radiusMeter 15000, fare [“jordskred”, “flom”], dager 90) lists the latest observations in Regobs, newest first.

  3. NVE

    nve_get_observation(regId) returns the text of a single observation: type, volume and what was affected.

  4. Geonorge

    geonorge_query_dataset on Kvikkleire (a29b905c-6aaa-4283-ae2c-d167624c08a8) shows the mapping area and hazard zones.

  5. Run daily, and events and zone hits are entered in the risk and vulnerability analysis (ROS).

Result

Daily situation picture: highest level 1 (green) for 23–24 October and no weather warning from MET Norway. Three landslides have been recorded in the last 90 days (1, 12 and 17 August 2026), and the centre is in NVE’s quick clay mapping area “Trondheim 1621-4”.

Why it matters

Gathers the event log and the day’s warning in one lookup and gives reliable documentation for the ROS analysis and the crisis management team.

Limits

Regobs contains observations from the public and agencies, not a complete list of events. The warnings are no substitute for Varsom and MET Norway, and “today” is counted as the UTC date. Observer names and photos are left out, but the link to regobs.no is included. Quick clay and flood zones give attributes without geometry.

Evidence from the run

2026-10-23: highest level 1; Regobs returned 3 landslides (regId 467539, 467621, 467612) in the last 90 days within 15 km; regId 467612 is a rockfall under about 100 m³ on a county road, under 1 m³ on the road and blocking only the ditch; quick clay mapping area “Trondheim 1621-4”.

UC-099Emergency ports, coastal warnings and NAVAREA before an incident in FinnmarkNorwegian Coastal Administrationemergency preparedness manager in an inter-municipal committee against acute pollution (IUA), Inter-municipal emergency preparedness partnership · Emergency preparedness and security
WorkflowVerified 23 Oct 20263 tools

Sources Norwegian Coastal Administration

Situation

In a shipping accident the officer on duty quickly needs an overview of pre-evaluated emergency ports and which navigational warnings apply in the area. The sources are on different pages at the Norwegian Coastal Administration.

The question

“Which emergency ports has the Norwegian Coastal Administration evaluated in Hammerfest, and which coastal and NAVAREA warnings are in force now?”

How the AI does it

  1. Norwegian Coastal Administration

    kystverket_find_emergency_ports with sok Hammerfest: name, number, category (1–3 by vessel size), IUA region and link to fact sheet (PDF).

  2. Norwegian Coastal Administration

    kystverket_get_coastal_warnings without arguments: warnings in force (lights extinguished, work at sea, danger area for Andøya Space).

  3. Norwegian Coastal Administration

    kystverket_get_navarea_warnings: NAVAREA XIX warnings (rigs with a 500 m safety zone, rocket firing in the Barents Sea, port restrictions).

  4. Ask the AI to make a short situation overview with links to the fact sheets.

Result

One page with emergency ports (9 hits in Hammerfest, the first three in category 2), current warnings with position and time period, and links to the Norwegian Coastal Administration’s fact sheets where the map is.

Why it matters

The officer on duty does not have to search PDFs and warning pages in a time-critical situation, and nothing has to be guessed.

Limits

The emergency ports have no position in the data (the Norwegian Coastal Administration’s map layer gives wrong coordinates) and cannot be searched by distance. The list does not say that a port is safe or free. The warnings are not official (NAVTEX and SafetyNET apply), the times have no time zone, and the texts can contain phone numbers of operators.

Evidence from the run

Run 2026-10-23: 9 emergency ports for Hammerfest (the first three: Akkarfjordbotn no. 50, Fagervika 52 and Fjordbotn 53, all category 2, IUA Vest-Finnmark), 8 coastal warnings (last updated 07:45) and 4 NAVAREA XIX warnings, among them 128/26 on rocket firing in the Barents Sea 25–28 October.

UC-100Daily power preparedness: situation, reservoirs and grid frequencyStatnettNVEElhubemergency preparedness coordinator, Municipality or critical organisation · Emergency preparedness and security
MonitoringVerified 23 Oct 20262 tools

Sources Statnett, NVE, Elhub

Situation

Every morning the emergency preparedness coordinator has to see whether the power situation is normal and whether there are early warning signs. The information is held by Statnett, the Norwegian Water Resources and Energy Directorate (NVE) and Elhub.

The question

“What is the power situation today per price area, how are the reservoirs doing, and is the grid frequency normal?”

How the AI does it

  1. StatnettNVEElhub

    get_power_situation without arguments (all five price areas): Statnett’s level, reservoir filling compared with normal, Elhub’s consumption and production for the latest published day and Statnett’s overview of Norway in MW.

  2. Statnett

    statnett_get_grid_frequency with sekunder 30: frequency per second and the number of measurements outside 49.90–50.10 Hz.

  3. Run it on a schedule every morning and ask the AI to alert when the level is 2 (Pressed) or higher, or the frequency is outside the normal range.

Result

A short morning report: level per price area, reservoir deviation and frequency, with a flag for deviations.

Why it matters

Early warning and one place to look, without collecting from three sources by hand.

Limits

Statnett’s assessment is a judgement at day level, not a forecast, and Elhub is about two days behind. No electricity prices, no grid status (faults and outages) and no warnings beyond Statnett’s level. The frequency is a snapshot of the most recent seconds.

Evidence from the run

Run 2026-10-23: Statnett “Normal” (level 1) in NO1–NO5; Norway produced 18,787 MW and used 16,390 MW at 09:45 (net exports 2,396 MW); NO2 reservoir 52.9% (−30.4 percentage points from the median); average frequency 50.0036 Hz over 31 seconds, 0 measurements outside the normal range.

About the data

How the catalogue was made

Run against real data

Every use case was run through TRY Innsyn against the open sources on 23 October 2026. The “Evidence” field shows what was actually found. The figures are a snapshot and will change.

“Verified” means the figures in the answer were found as described. “Partly verified” means something is missing, and “Evidence” and “Limits” say what.

The use cases were run in Norwegian. The questions on this site are translations.

The limits are always included

“Limits” says what the sources do not cover, so that a use case does not promise more than the data gives. It matters as much as the result.

The examples use real organisations as input. That says nothing about who is a customer or a prospect. No use case maps private individuals.