Ansvar's answer

Ansvar is grounded, not generated. Where most compliance AI tools ask a language model to recall what a regulation says, Ansvar retrieves the actual provision text from official sources and returns it with its citation, CELEX identifiers for EU law, national identifiers like BWBR for Dutch statute, licensed clause text for standards. If the source doesn't have it, Ansvar says so rather than inventing something plausible.

The second difference is the delivery model. There is no portal to log into. Ansvar is an MCP gateway: it plugs into the AI assistant your team already uses, so compliance intelligence shows up inside the workflow instead of in yet another dashboard. Coverage spans 90+ jurisdictions and 260+ frameworks, alongside deterministic engines for things like CVSS environmental scoring and structured workflows for NIS2 gap analysis, threat modelling, and DPIA/FRIA.

Everything runs on EU-sovereign infrastructure, and the COMPANY tier writes every query to an eIDAS-anchored Merkle-chain audit ledger, so you can prove to an auditor what was asked, what was answered, and that the record hasn't been altered.

Ansvar's answer

Ansvar is built on the Model Context Protocol (MCP), which is how the gateway exposes retrieval, workflow, and scoring capabilities to AI assistants such as Claude. The backend is a federated set of MCP servers, one per jurisdiction or domain, sitting behind a single gateway that handles authentication, tenancy, rate limiting, and audit logging.

Underneath: structured legal corpora ingested from official sources (EUR-Lex, national legislative registers, licensed standards bodies), deterministic rule engines for risk scoring rather than model inference, and a Merkle-chain audit ledger anchored via eIDAS-qualified timestamping. Infrastructure is hosted entirely in the EU on Hetzner, with Kubernetes for orchestration. Bring-your-own-key model inference means customer data never trains anyone's model.

Ansvar's answer

Three reasons.

Verifiability. Every answer carries a source link back to the official text. GRC platforms give you workflow but no legal substance; general-purpose AI gives you fluent answers you can't cite in an audit. Ansvar gives you the provision and the pointer to it.

Licensed content, legitimately. Standards text isn't freely available. Ansvar holds a licensing agreement with SIS (Swedish Standards Institute) to serve ISO/IEC 27001:2023 and 27002:2022 clause text through the gateway — so you get the actual control wording rather than a paraphrase of uncertain provenance.

Sovereignty and auditability by default. EU-hosted infrastructure, bring-your-own-key at every tier, and a tamper-evident audit trail. For regulated European organisations, that removes the procurement objections that stall most AI tooling before it starts.

Ansvar's answer

Security, compliance, and legal teams inside EU-regulated organisations, the people who have to demonstrate conformity, not just claim it. Concretely: CISOs and security architects working through NIS2, CRA, and DORA obligations; compliance and privacy officers running DPIAs and gap analyses; and in-house counsel who need to know what a provision actually says in a given member state.

Sector-wise, the strongest fit so far has been medtech, energy, financial services, industrial software, and public sector — organisations operating across multiple jurisdictions where the same obligation is transposed differently in each one, and where getting it wrong is expensive.

Secondarily, consultancies and advisory firms who bill for regulatory analysis and need it to be defensible.

Ansvar's answer

Ansvar came out of frustration on the practitioner side. I've spent my career in cybersecurity architecture, automotive (ISO/SAE 21434, UNECE R155/R156), financial services, and Dutch government, and the same pattern kept repeating: enormous effort spent locating what a regulation actually requires, and then re-locating it every time someone asked again.

When LLMs arrived, the obvious move was to ask them. They were confidently wrong often enough to be dangerous. A hallucinated article number in a compliance assessment isn't a minor error; it's a finding.

"Ansvar" is Swedish for responsibility or accountability, which is the point. The system is built so an AI can only tell you what a real source says, and always shows you the source. Ansvar Systems AB was founded in Sweden in October 2025 to build it, and it runs on EU infrastructure because European regulated organisations shouldn't have to send their compliance questions across an ocean to get answers about European law.