Governance & Audit Hub
Oversight, lineage, and AIMS roll-up.
Human oversight maturity
Four operating modes along the autonomy ladder. Co-Pilot: human drives, agent assists. HITL: human approves every action. HOTL: human reviews by policy / sampling. HOOTL: agent runs autonomously, humans handle exceptions only.
Total use cases
14
Fully autonomous
4
29% of portfolio
Avg autonomy score
57%
Weekly actions
4,710
Escalations · 30d
19
3 ready to promote
Adoption pipeline
Use cases progress left-to-right as trust, safety and efficiency evidence accumulates. Click a card to review and promote.
Co-Pilot
Baseline · suggestions only
3
HITL
Per-action approval
4
HOTL
Policy-based · interruptible
3
HOOTL
Autonomous · exception only
4
Use case register
All use cases with scorecards and promotion status.
| Use case | Stage | Autonomy | Weekly | Escal. 30d | Safety | Efficiency | Trust | Action |
|---|---|---|---|---|---|---|---|---|
Draft email replies Suggest responses to customer emails | Co-Pilot | 5% | 340 | 0 | — | |||
Code snippet suggestions Inline code completions | Co-Pilot | 10% | 1,200 | 0 | — | |||
Knowledge base search Suggest relevant docs | Co-Pilot | 2% | 520 | 0 | — | |||
Schedule meetings Find slots + send invites | HITL | 33% | 180 | 0 | ||||
Create pull requests Generate PR from diff | HITL | 40% | 95 | 0 | — | |||
Expense report approval Flag policy violations | HITL | 28% | 210 | 0 | — | |||
Jira ticket assignment Suggest assignee | HITL | 45% | 320 | 0 | ||||
Auto-triage support tickets Categorize + priority | HOTL | 80% | 540 | 5 | — | |||
Server auto-scale decisions Add/remove nodes | HOTL | 75% | 42 | 2 | — | |||
Database query optimization Suggest index + rewrite | HOTL | 85% | 28 | 1 | ||||
Auto-remediate common incidents Restart + rollback | HOOTL | 98% | 15 | 1 | — | |||
Customer chatbot resolution Full conversation | HOOTL | 95% | 890 | 8 | — | |||
Password reset requests Verify + reset | HOOTL | 99% | 210 | 2 | — | |||
Cache warmup Preload hot keys | HOOTL | 97% | 120 | 0 | — |
Agent:
RACI
ResponsibleD. Park
AccountableS. Chen
ConsultedModel Risk, Legal, Privacy Office
InformedInternal Audit, Board Risk Committee
Incidents
Latency spike during sanctions lookup
INC-1000 · Dec 27 00:00
medium
Change history
| When | Author | Change |
|---|---|---|
| Dec 30 00:00 | D. Park | Prompt v3.2 — tightened sanctions phrasing |
| Dec 23 00:00 | L. Rossi | Added CaseManager.write:notes capability |
| Dec 11 00:00 | K. Nakamura | Model upgrade gpt-4o |