AI Governance

Put a person behind every AI action.

Connect your agent through CLAIIM to check who is accountable and whether a requested action is allowed. The controlled gateway applies the decision on its execution path. Every gate decision is recorded.

Runs in your environment. Your policies, your database, your proof.

ALLOW — send reply
DENY — approve refund
ALLOW — file ticket
ALLOW — schedule task
DENY — release code
ALLOW — read document
CLAIIM Gateillustration
support-aiasks to send customer replychecking…
finance-aiasks to approve refundwaiting
hr-aiasks to send offer letterwaiting
Integrates through SDK and HTTP pathsBuilt-in audit logAlways one responsible personPolicy before execution
How it works

Simple by design.

Integrated agents send requests to the gate. Enforcement depends on the execution path and destination credentials. See coverage and current status.

01

AI agents ask for permission

An integrated agent submits the intended action to CLAIIM first. Gate-only integrations rely on the agent respecting the decision; the controlled gateway provides a separate execution path.

02

CLAIIM checks your rules

CLAIIM matches the request against the policies your team set. Requests are allowed, denied, or held for review according to your policies. A decision alone does not prevent alternate execution.

03

A person is always on the record

Every decision is linked to a responsible person and the rule that allowed or denied it. Chron keeps an append-only application audit record. Independent verification is in development.

Chron — built-in audit log

Every decision leaves proof.

Review who was responsible, what rule applied, and where the record lives. The examples below illustrate decisions, not live agent activity.

Who was responsible?

A named person, always

Every decision is linked to a real person in your team. AI agents act under a named anchor. No anonymous actions.

anchor: r.okafor@support role: team-lead org: /support/tier-1
What rule applied?

The exact policy, on record

CLAIIM records which policy version made the call. If rules change later, the record shows what was in effect at the time.

policy: send-reply-v4 matched: customer-facing-ok version: 2026-03-11
What happened?

The request and the decision

The requested action and gate decision are recorded. Reported gateway delivery does not independently verify downstream success.

verdict: ALLOW action: send:customer-reply recorded: 09:12:44Z
Where is the proof stored?

Chron, your built-in log

An append-only record in your own environment, exportable for review. Independent verification of the record is in development.

store: chron://your-org integrity: application append-only export: JSON · HTML proof
Use cases

One team, any AI agent.

These are example policy scopes. Start with one agent and an integration whose execution path you can control. Destination connectors vary by use case.

Support teams

ALLOW draft customer replyDENY issue refund

Let agents handle routine support work while keeping refunds and account changes human-approved.

Finance teams

ALLOW read payment dataDENY approve payment

Give agents visibility into financial data without letting them approve or move money.

Engineering teams

ALLOW deploy to stagingDENY release to production

Let deployment agents work freely in staging while blocking production releases without review.

HR and operations

ALLOW draft offer letterDENY send without review

Agents can prepare documents and correspondence. A person approves before anything goes out.

Get early access

We are onboarding teams now. Join the preview and see CLAIIM in your environment with a scope agreed with our team.