REVTECH AI AGENT GOVERNANCE CHECKLIST

Content ID: RT-20260827-07
Use: one worksheet per agent workflow.

1. WORKFLOW CONTRACT
[ ] Objective and recurring trigger are explicit.
[ ] Expected output and accountable business owner are named.
[ ] Approved systems, records, fields and definitions are documented.
[ ] Prohibited actions are explicit.
[ ] Quality, completion and business measures have a baseline.

2. RISK TIER
[ ] Impact of a wrong or late output is understood.
[ ] Customer and commercial exposure are recorded.
[ ] Reversibility is known for every proposed action.
[ ] Required human judgment is named.
[ ] Risk tier determines the approval posture.

3. PERMISSIONS
[ ] System access is limited to the named workflow.
[ ] Object, field and record-population boundaries are documented.
[ ] Read, prepare and write permissions are separated.
[ ] Rate or volume limits are set where appropriate.
[ ] Approvers follow existing roles and access rights.

4. REVIEW AND EVIDENCE
[ ] Review items show the proposal, evidence and affected records.
[ ] Approve, edit, reject and escalate paths are tested.
[ ] Review deadlines and escalation owners are named.
[ ] Consequential or hard-to-reverse actions retain a human gate.

5. AUDIT AND RECOVERY
[ ] Inputs, policy, proposal, approver, edits and final action are recorded.
[ ] Pause control and incident owner are documented.
[ ] Affected records and time windows can be isolated.
[ ] Rollback or corrective action is defined for reversible changes.
[ ] Restart requires an approved rule, permission or evaluation change.

6. SCALE DECISION
[ ] Acceptance, edit, rejection and exception rates are reviewed.
[ ] Workflow completion and business outcomes are reviewed together.
[ ] Scope widens only when evidence supports it.
[ ] The next governance review has an owner and date.

RevTech provides fully managed AI agents for GTM. RevTech runs the agents;
customers govern the objectives, permissions, approvals and outcomes.