How it works

Proofroom in four simple steps

You run AI agents. Proofroom is a flight recorder for them: hash-chained evidence, Action Receipts, and a live page you can share. No safety certification, no magic green ticks, just an honest audit trail.

Your agent's material actions become linked Action Receipts on a hash chain

The big picture

Your agent does work in the real world. You send short evidence events to Proofroom. Material actions become Action Receipts on an append-only chain. Anyone you share the link with can see what was declared, what was recorded, and what is not evidenced.

1
Sign in and create your organisation
- Create a free account with your email.
- Give your organisation a name. This is the workspace where your agents and evidence live.
Sign in →
2
Register your agent and use case
- An Agent Passport says who the agent is (name, vendor stack).
- A Use Case Passport says what it is allowed to do for one specific job: scope, allowed actions, prohibited actions, and who oversees it.
- Proofroom covers use cases, not vague “AI in general”.
Read setup guide →
3
Send evidence as your agent works
- Create an API key in Settings (shown once).
- As your agent does meaningful work, POST short evidence events to Proofroom: what happened, who did it, when.
- Events link together in a tamper-evident hash chain. Important actions automatically get an Action Receipt (PRF-XXXXX).
- Send proof, not payloads: sensitive content is stripped at ingestion.
MCP & API docs →
4
Share your live proof room
- Every use case gets a live proof room: status, coverage score, chain integrity, and Action Receipts.
- Share the link with customers, auditors, or procurement. Generate an Evidence Pack when they need a snapshot.
- If evidence stops flowing, status shows Degraded. The live room is always more current than a PDF.
See a live example →