Public demo: read-only, fake AI provider, resets every hour. Run the scenarios to see new rows appear here.
Console
More
loading

AI spend

Every call through Governor is capped, can be held for approval, and is signed into the audit trail.

Organization

Spend across every key, against the account caps.

Waiting for approval

Your decision is signed into the audit trail together with a hash of exactly what this row showed you. Deny and Defer ask for a reason; Approve smaller runs it with a lower output cap.

RequestKeyModelWorst caseWhat you are approvingAsked

Keys

One key per agent or app. Each has its own daily, monthly and per-request caps.

KeyTodayThis monthPer requestApproval overStatus

New key

Copy this key now. It will not be shown again.

Open stops

Everything switched off right now, who switched it off, why, and when someone said they would look again. A stop never lifts itself at its review time: its owner is alerted once, and that alert is signed into the audit trail.

StoppedByReasonForReview by

Outage drill

Mark a provider or one model down. Calls for it go to each key's fallback model (set under Caps), or are refused with 503. Every step is signed into the audit trail, and Verify chain checks nothing reached a target while it was down.

DownReasonFromUntilReroutedRefusedCost vs asked-for model

Alerts

Caps reached, loops broken, approvals asked for and kill-switch changes.

Audit trail

A hash-chained, signed record of every call and decision. Verify it here, or download a proof for one record.

#TimeEventKeyModelCostDetail

Your .key

Link a .key name and you can open this console with the name instead of the admin token. Approvals and kill-switch changes made that way are signed into the audit trail as name.key.

No name yet? Get one