Deterministic classification
Every system is scored against structured EU AI Act Article 6 and Annex III criteria — the same inputs always produce the same tier, with the matched criteria recorded.
AI risk management software for B2B SaaS
Most risk-management tools stop at a one-time score. Govarna classifies each system deterministically against the EU AI Act, keeps every reassessment instead of overwriting the last one, and maps live controls to the tier it lands in.
Best fit: teams that need a defensible, auditable AI risk register — not a spreadsheet of guesses or a black-box score.
Not a numeric risk score, model monitoring, or a GRC platform replacement.
A register, not a guess
Every classification is reproducible, sourced, and kept. Your team confirms or overrides — Govarna structures the decision and its history.
Every system is scored against structured EU AI Act Article 6 and Annex III criteria — the same inputs always produce the same tier, with the matched criteria recorded.
Reassess a system when its use case or data changes. Prior classifications stay in the record instead of being replaced, so you can show what changed and when.
High-risk systems automatically pull in the control set that applies to that tier. Minimal-risk systems don’t inherit obligations they don’t have.
Attach the document, screenshot, or export that supports a control — linked back to the specific system and risk tier it evidences.
Why deterministic, not a survey
A risk-scoring survey can drift between reviewers, or between the same reviewer on different days. A rules engine applied to structured facts about the system can't — which is what makes the result defensible when a buyer, auditor, or regulator asks how you got there.
The free deployer assessment takes about three minutes and tells you whether the EU AI Act likely applies to you before you set up a workspace.
Article 50 transparency duties have applied since 2 August 2026. High-risk timelines are scheduled later under the May 2026 political agreement, pending legal adoption.
Questions before you start
No. Govarna classifies each AI system against the EU AI Act’s actual Article 6 and Annex III criteria — a deterministic rules engine, not a weighted survey. The output is a risk tier (minimal, limited, high, or prohibited) with the specific criteria that produced it, not an opaque score.
Controls also map to the NIST AI RMF. Govarna does not currently map to ISO/IEC 42001 or other frameworks — we’d rather say that plainly than list it as included.
You reassess it. The prior classification stays in the record rather than being overwritten, so the history shows exactly what changed, when, and why the tier moved.
Not directly today. Govarna is a standalone workspace for the AI-specific parts of risk management — inventory, classification, controls, and evidence. Exports are PDF and JSON for handoff to whatever your team already uses.
Govarna applies deterministic rules and shows its reasoning. Your team confirms or overrides the result and records why — the tool structures the decision, it doesn’t make it unaccountably.
The Starter plan is $499 per month and includes a 14-day free trial with no credit card required. Higher tiers add unlimited systems and multi-framework control mapping.