Start free. Widen what one question can reach.
Free searches one jurisdiction or framework at a time, serves public CVE/KEV/EPSS facts, and runs one described-system assessment a month. Solo searches every jurisdiction in one question and includes two runs. Premium adds case law, CAPEC/CWE/D3FEND enrichment, and the full interview-grounded workflow catalog, including run-scoped effective-risk analysis in applicable vulnerability and ICS workflows. Team adds document-grounded runs, standalone effective-risk tools, persistent scoring contexts, review decisions, and OpenVEX export. Workflow findings cite served evidence or remain unresolved.
What do you need? Cited answers across every jurisdiction — Solo. Evidence a lawyer would demand — Premium. Assessments on your own documents — Team. A signed, provable audit trail — Company. An expert to deliver it — Services.
Prices excl. VAT
Evaluate Ansvar, one jurisdiction or framework at a time
100 search calls / day · fair use
- Sign in with Google, Microsoft, or email — no card, no VAT entry
- Single jurisdiction or framework per question
- Full citation contract
- Public CVE details, KEV status, EPSS scores, exploits and product search
- 1 workflow run / month — threat model, NIS2/DORA/CRA/AI Act gap analysis or DPIA, watermarked report
- Every jurisdiction in one question
- Legal evidence layer (case law, preparatory works, agency guidance)
- Audit ledger
No credit card required
One seat — a person or an agent — with the whole fleet in one question
750 search calls / day · fair use
- Every jurisdiction and framework in one question
- 49 licensing-audited countries · 262 security frameworks
- 750 calls a day — headroom for real research
- Shows when case law exists for what you searched — reading it is Premium
- 2 workflow runs / month — the same seven types as Free, watermarked report
- Legal evidence layer (case law, preparatory works, agency guidance)
Cancel anytime
Premium is one seat. When compliance is a function — more deliverables a month, grounded in your own documents, seats for the team and its agents — these two step up.
A compliance function — workflows on your own documents, SSO
From 1 seat — €490/month
50,000 search calls / day · org-pooled · fair use
- Workflows on your own documents — DPIA, gap analysis, tender review — 20 runs / seat / month
- Standalone effective-risk tools, persistent asset contexts, review decisions and OpenVEX export
- Document upload — paragraph-level, tamper-evident citations
- Shared workspaces, SSO, quota pooled across the org
- Service credentials — standing agents on their own seats
- Audit ledger
€490/mo · 1 seat × €490/mo
Cancel anytime
Regulated orgs that need the signed audit ledger
500,000+ search calls / day · org-pooled · more by contract · fair use
- Per-tenant cryptographic audit ledger — signed receipts
- Dedicated Hetzner cluster by agreement, fully shielded from other tenants
- Custom retention windows, SLAs, and premium support
We reply in writing
One question to the gateway is one call against your quota — no matter how many of the 224 servers it fans out to. Quotas are sized for agents, not typing speed: a seat can belong to a person or to an agent, and a research agent may spend hundreds of calls on one deliverable.
A workflow run is one assessment taken from start to finish: the workflow asks its questions in your own client and produces an exportable document with factual findings cited or marked unresolved. The output can be a threat model, DPIA, gap analysis, or tender review. On Free and Solo the included runs cover seven types: STRIDE threat model, gap analysis (generic or NIS2, DORA, CRA, EU AI Act), and DPIA on a system you describe, and the report comes back as a watermarked render carrying a self-asserted banner, or as structured JSON. The full interview-grounded catalog and case law inside the run start at Premium; unwatermarked branded exports and document-grounded runs are Team and up.
Need the deliverable, not the tool? Fixed-scope engagements, priced per engagement: see consulting services
Every tier, side by side.
Three questions, one table: how much you can ask, what a question can reach, and what it produces.
| Free€0 | Solo€29/mo | Premium€249/mo | Team€490/seat/mo | Companycustom | |
|---|---|---|---|---|---|
| Capacity | |||||
| Daily searchesone question = one call, regardless of fan-out — sized for agents, not typing speed; Company's figure is the contractual floor, raised by contract | 100 | 750 | 5,000 per seat | 50,000 org-pooled | 500,000+ org-pooled |
| Parallel requestshow many calls your client or agent can have in flight at once | 3 | 4 | 5 | 16 | 64 |
| Knowledge | |||||
| Scope per questionFree answers one jurisdiction or one framework per question; Solo and up fan out across all 224 servers in one call | one jurisdiction or framework | full fleet | full fleet | full fleet | full fleet |
| Legal evidence layercase law · preparatory works · agency guidance — Solo shows that decisions exist; Premium reads them | — | shows it exists | ✓ | ✓ | ✓ |
| IETF security RFCsTLS, OAuth, and the protocol standards corpus | — | — | ✓ | ✓ | ✓ |
| Security decisions | |||||
| Public vulnerability factsCVE details, CISA KEV status, EPSS scores, known exploits, product search and feed freshness | ✓ | ✓ | ✓ | ✓ | ✓ |
| Threat-pattern enrichmentCAPEC, CWE and D3FEND evidence for security analysis; ATT&CK and ATLAS remain available on every tier | — | — | ✓ | ✓ | ✓ |
| In-workflow effective-risk analysisrun-scoped rescoring inside the vulnerability assessment, deferral dossier and ICS advisory workflows; control simulation in the first two | — | — | ✓ | ✓ | ✓ |
| Standalone asset-context decisionspersistent scoring contexts, standalone effective-risk tools, review decisions and OpenVEX export | — | — | — | ✓ | ✓ |
| Deliverables | |||||
| Structured workflowsone run = one assessment taken start to finish, producing a cited report — Free and Solo pick from seven types (threat model, gap analysis incl. NIS2/DORA/CRA/AI Act, DPIA) on a system you describe, reported as a watermarked render or JSON; Premium runs the full interview-grounded catalog; Team runs them on your uploaded documents | 1 / mo (watermarked) | 2 / mo (watermarked) | 5 / mo | 20 / seat / mo | custom |
| Document uploadparagraph-cited review under your retention policy | — | — | — | ✓ | ✓ |
| SSO & shared workspaces | — | — | — | ✓ | ✓ |
| Service credentialsorg-owned credentials for headless clients — n8n, CI, scheduled agents; each takes a seat of its own with the same tool surface as a signed-in seat on your plan. Create and rotate at Team and up; revoke works on any plan, so a downgrade never strands a live credential | — | — | — | ✓ | ✓ |
| Audit ledgerper-tenant cryptographic receipts, signed | — | — | — | — | ✓ |
Coverage grows continuously — new corpora and jurisdictions ship most weeks. The best way to check what’s live today: connect your AI client and ask the gateway itself (“what can you reach?”). The Free tier is enough for that.
Licensed ISO clauses and controls, delivered to your AI through MCP.
Under a licence agreement with the Swedish Institute for Standards (SIS), Ansvar serves licensed clause and control text through the gateway, cited and attributed to the source standard instead of paraphrased from model memory. The listed standards are live today; any standard SIS publishes can be licensed in on request.
Standards are priced per seat — the same price whether a person or one of your agents holds it. Volume discounts are available; talk to us if you need seats in bulk.
SS-EN ISO/IEC 27001:2023 · SS-EN ISO/IEC 27002:2022 · SS-EN ISO/IEC 27005:2024 · SS-EN ISO/IEC 42001:2026 · SS-ISO/SAE 21434:2021
The four questions procurement asks.
Architecture answers first — billing and contract answers in the FAQ below.
Do I need a new chatbot?
No. Ansvar is one MCP connection for the AI client your team already runs — Claude, Claude Code, Cursor, VS Code Copilot, Open WebUI — or for an agent you run headless. The gateway is a router, not a chatbot.
Where does my data live?
EU-hosted core (Hetzner), with edge and backups disclosed on our subprocessors page. Corpus MCP servers never store client data — your research context stays in your client. Documents you register on Team and Company live in tenant-isolated EU storage under your retention policy. This website runs no tracking cookies and no third-party trackers — one functional cookie remembers your plan when you sign in.
What is BYOK?
Bring your own key. Your client calls its model with your key; there is no LLM inside the gateway. It routes, fans out, and attaches citations — your model traffic never passes through us.
What happens at the quota?
An actionable refusal: your tier, your usage, your reset time, the upgrade path. Never silent degradation, never a thinner answer dressed up as a full one.