The Sentinel
Real-time observability dashboard for all agent activity, performance, and compliance metrics.
Governance Tasks1 pending
PENDINGCase Analyst87% confidence+18% accuracy boost predicted
Missing Liability Clauses
View prompt diff
CURRENT PROMPT
Extract all parties, timelines, and claims from the uploaded PDF document.
PROPOSED CHANGE
Extract all parties, timelines, and claims from the uploaded PDF document. YOU MUST also explicitly extract and list any INDEMNIFICATION or LIABILITY LIMITATION clauses. If none exist, state 'NO LIABILITY CLAUSES FOUND'.
Global Deflection Rate
78.4%+2.1% (7d)
Tasks resolved without human intervention.
View details →Average Confidence Score
89.3%-0.5% (7d)
Aggregated certainty across all active models.
View details →Active Guardrail Interventions
0Total Logged
Times an agent generated a warning/error.
View details →Cost per Successful Resolution
$0.42-$0.05 (7d)
Token costs vs. human baseline.
View details →Agent Drift Health
LIVE| Agent Identity | Model | Current Accuracy | Performance Trend | Status |
|---|---|---|---|---|
| Doc Reviewer | claude-3-5-sonnet | 96.5% | — | Healthy |
| Case Analyst | gpt-4o | 82.1% | — | Warning |
Runtime Logs
No logs generated yet. Run a workflow to see data.