Skip to content

Why Apier

Sovereign friction is real and Norway should win from it.

Norwegian regulatory infrastructure is excellent for humans, and structurally unfinished for AI agents. Apier is the layer that turns the second statement into the first.

The thesis

Altinn, Brønnøysundregistrene, Maskinporten, and Skatteetaten are public infrastructure that Norwegian companies, accountants, and government services rely on every working day. The portals are documented. The APIs exist. The legal framework is published on Lovdata. What is missing is the layer that translates all of it into a single, deterministic, machine-readable answer that an AI agent can plan against. Agents do not read Norwegian PDF forskrifter. They do not infer from Brønnøysund role codes which person can sign for which entity. They do not keep an Aksjeloven amendment calendar. That work is done once, correctly, by a small number of Norwegian compliance specialists today — and replicated badly, in private, by every team that tries to integrate. Apier exists to do it once and expose it as an API.

What is structurally missing

What is missing is a translation layer between four moving surfaces, not a single endpoint:

  • Signing authority as machine state. Brønnøysund records who may sign for a company as free-text clauses on roles. It does not return the next-step question an agent has: given these roles and clauses, which combination of people can bind the company, and what specifically is missing if none can.
  • MVA thresholds and obligation cadence. The MVA threshold for registration, the bimonthly versus annual termin choice, the auditor trigger at 7 million NOK in turnover, the årsregnskap deadline that shifts on weekends — these are codified in Skatteforvaltningsloven, MVA-loven, Aksjeloven, and a stack of forskrifter. No public API returns the consolidated answer for a given organisation. Every integrator either pays a compliance specialist to read the rules or hard-codes a brittle subset.
  • Brønnøysund roles in Norwegian PDFs. Daglig leder, styremedlem, signaturrett, prokura, innehaver — the role taxonomy that determines who can act for an entity is spread across Brønnøysundregistrene, Aksjeloven, and per-role precedent. The data is published; the operational mapping from role plus action to “yes, this person can sign that filing” is not.
  • Maskinporten as a credential lifecycle. Maskinporten is a clean OAuth2 server. Around it sit virksomhetssertifikat issuance, Samarbeidsportalen onboarding, scope approval cycles, key rotation, and the test versus production environment split. The cost of doing this once per integrating team is the cost of a senior engineer for several weeks per team, every year.

Who Apier is for, and who it is not for

Apier is for AI agents and the developers who run them, not for end-users who need a portal.

  • AI agent platforms. Claude, Cursor, MCP-compatible harnesses, LangChain or AutoGen pipelines, custom orchestrators. The tools the agent calls should be deterministic, dated, and stamped with their source. Apier ships that shape over both REST and the Model Context Protocol.
  • Integrators wiring agents into Norwegian company authority. The team building an internal agent that reads the signing rules for a counterparty, checks an auditor threshold, or reads the filing calendar before acting. Apier removes the integration tax so the integrator can focus on the workflow.
  • Foreign companies operating in Norway. Norwegian company law and its registries are famously specific. Apier returns the same sourced, dated answer regardless of which country the agent is reasoning from, so the team does not need a Norwegian specialist on staff to ship.
  • Not the audience: individual taxpayers, end-users filing for themselves, or teams with one-off lookup needs. The agencies' own portals already serve those well, and Apier is intentionally not in that lane.

FAQ

Why route through Apier instead of calling Altinn, Brønnøysund, or Skatteetaten directly?
The agencies expose raw data. Apier turns that raw data into sourced, dated answers an AI agent reads before it acts: who holds signing authority as the register codes it, when the next deadline lands in Europe/Oslo, which MVA termin a given organisation owes. Direct calls are free and reasonable if your software needs one row from one register. If your agent needs to plan or act across registers, the integration cost flips toward Apier.
Is the data live or cached, and how fresh?
Brønnøysund company state is reconciled on read with a short cache; signaturrett and prokura are read from Brønnøysund's Fullmakttjenesten on every request because they are the register's coded signing rules for the company. Every response carries a _meta block with rulebook_version, data_freshness, and last_verified so an agent can decide whether to replan or proceed. Apier is deterministic: same inputs and rulebook version produce byte-identical outputs.
Where is data hosted? Does anything leave Norway or the EU?
Primary data lives in Supabase EU (Stockholm). Build and edge hosting is on Vercel EU edge with a migration off Vercel planned before real customer data flows. The well-known attestation at /.well-known/data-sovereignty publishes the subprocessor list, regions, and DPA status in a machine-readable shape so procurement teams and agents can verify it without asking us. See also our trust + transparency page.
Who is the typical Apier user, and who is not?
The typical user is an AI agent platform, an integrator wiring an agent harness to Norwegian company authority, or a foreign company that needs to act in Norway without hiring a Norwegian specialist to read the registries. Apier is not for end-users filing their own taxes (that is Skatteetaten's surface), and not for one-off lookups, where calling Brønnøysund directly is cheaper.

Related reading

  • /sandbox: the zero-auth public sandbox surface, with curl-able examples of obligations, deadlines, and authorization results.
  • /docs: the developer docs, including the OpenAPI spec, the MCP server config, and the Compliance Explainer error catalogue.
  • /use-cases/altinn-migration: the Altinn 2 to Altinn 3 migration page, with the access-package mapping table and the 19 June 2026 deadline status (the cutoff has passed).
  • /blog/altinn-3-migration-for-developers: the developer-facing migration guide, with code samples and the structured-error envelope shape.
  • /blog/maskinporten-guide: the Maskinporten primer covering client setup, token lifecycle, and the Altinn 3 system user model.
  • /norwegian-government-apis-for-foreign-companies: the cross-border access sequence for companies without a Norwegian registration.
  • /faq: site-level questions: which government systems Apier reaches, how an agent authenticates, and what the sandbox does and does not prove.