Resources

FAQ

Quick answers on connecting agents, verifying proof and protecting data.

• How do I connect an agent? — Two paths: non-technical (pick from a list of platforms, code handoff sent to your technical team by email) or technical (paste a single code block directly, API key included).
• How is a proof verified? — Each recorded trace gets an HMAC integrity code (shared-secret, verifiable by GoVidence) and, for newer traces, an Ed25519 signature anyone can check with the published public key. Verification shows the signed content was not changed afterward; it does not prove the action itself happened or was complete. Demo and older traces may not carry the Ed25519 signature.
• How does a requirement turn 'green' (Met) in a compliance report? — A requirement automatically turns green once GoVidence can prove it with real signed data — for example, the number of months of retained logs, a notification recorded before an automated action, or a logged human review. Nothing is marked as met by default: it's always based on real evidence in your traces. Some requirements, however, can never be verified by GoVidence, no matter how much data you send — for example an external bias audit or an internal policy document. These are clearly shown as 'Client responsibility — not verifiable from logs', so you always know what GoVidence covers and what remains your responsibility.
• What happens to PII (personal data)? — Client-side redaction is your responsibility. GoVidence also applies pattern-based redaction to input and output on the server as a safety net; it is not an exhaustive check.
• Where is data hosted? — GoVidence data is hosted on Lovable Cloud.

Glossary

Key AI compliance terms, explained simply.

• HMAC — A shared-secret integrity code showing data hasn't been altered since it was recorded; it is not a publicly verifiable signature.
• Ed25519 — An asymmetric signature verifiable by anyone with the matching public key, covering only the fields included in the signed payload.
• Watcher — GoVidence's signing component, separate from the agent, that receives and signs the traces your systems submit.
• Audit trail — The complete, tamper-evident history of actions, usable as evidence.
• Risk score — A severity indicator assigned to an action based on its context and the applied policy.
• Swarm behavior — Unusual coordinated activity across multiple agents, detected automatically.

About

GoVidence turns AI agent actions into externally signed, understandable evidence to support your audits — with personal data redacted on your side and no external AI involved in interpreting it.