
explorer
How does organisation-level governance work?
Fleet-wide attribution, policy gates and revocation with proof
Your agents did 4,183 things today. The organisational question is whether you can prove every one, and CueCrux's Governance tier exists to make the answer yes across an entire fleet, not just a single machine.
The first pillar is attribution. Every cost and every action resolves through the same chain: fleet, then agent, then verb, then receipt. Nothing in that chain is estimated. Where metered credits are spent on heavy lifts such as dense rerank, extraction or attestations, each credit spend mints a signed receipt of its own, so the bill and the audit trail are the same artefact. Finance and security stop arguing from dashboards and start arguing from receipts.
The second pillar is policy gates. Plans and work items carry risk classes, and high-risk actions route through human gates that cannot auto-approve past a timeout. If no human approves, the action does not happen. This is the property auditors actually probe: a gate that quietly waves work through when nobody is watching is not a control.
The third pillar is revocation that propagates. When you pull an agent's access, the revocation is a signed pull, each agent acknowledges it individually, and when zero holders remain you get an attestation saying so. Offboarding an agent or rotating away a compromised credential ends with proof of removal, not an assumption.
Governance is priced at £49 per user per month, on top of a local daemon that is free forever and a Pro tier at £19. The hosted app and live fleet feed land with the launch gate; the daemon-side machinery described here is what they surface.