Scenario Tuning Agent
Tunes monitoring thresholds and scenarios to cut alert noise.
Watches alert outcomes and proposes threshold and scenario changes, each backtested against historical data. Every change ships with an audit trail and a model-governance memo; the model-risk agent re-checks it and the accountable model owner approves it before go-live.
Authority
Recommend for approval
Team role
Provides specialist analysis
Handoffs
Named collaborators
The role
What it owns and where its authority ends
Desk
Transaction Monitoring
Desk workflow
Alert generation, then triage, enrichment, and disposition (close or escalate), then case creation. The triage agent runs the full loop continuously.
Collaboration
Works within a defined desk workflow
Decision boundary
An accountable reviewer commits the decision or action.
Systems and capabilities involved
Alert outcome warehouse
Backtest sandbox
replays scenarios over history
Model-governance doc store
Handoffs
What this role gives and receives
Capabilities offered
The handoffs name the next owner or specialist and the work that moves between them.
Handoff to
External handoff
Risk's model risk management desk for validation sign-off
Context
What the role needs to do the work
- Current work
- Current scenario under review + backtest results.
- Prior interactions
- Not specified for this role.
- Policies and reference
- Scenario catalog, regulatory expectations (BSA), prior tuning decisions.
- Working method
- Not specified for this role.
Illustrative workflow
How the work moves
Starting point
Rapid-movement scenario firing 4x the alerts of peers with 0.3% conversion.
- 01
Backtest a tightened threshold over 24 months of confirmed cases.
- 02
Confirm no true positives would have been missed.
- 03
Draft the model-governance change memo.
Result
Recommends a tuned threshold cutting ~3,000 alerts/month, pending the model-risk agent's re-check and the accountable model owner's approval.
Checks and boundaries
What must be tested or reviewed
- 01Every proposed change requires a passing backtest, a model-risk agent re-check (SR 11-7) and the accountable model owner's approval.
- 02Champion/challenger comparison before any threshold goes live.
Human authority
An accountable reviewer commits the decision or action.
Keep exploring