TriageMind
Back to home
Evaluation

Performance dashboard

Measured proof from the eval harness over 27 labeled cases — plus live analytics from your current session. (generated 2026-06-09)

Classification accuracy
70%
Escalation accuracy
70%
Guardrail violations
0
target: 0
Avg confidence
78%

Route distribution

✓ Auto-resolve18
📷 Request evidence3
↗ Escalate6

Severity distribution

LOW21
MED1
HIGH5

This session

No triages yet. Run one in the live demo and it shows up here.

Per-case results

CaseSeverityRouteConfidenceResult
t1LOW📷 Request evidence90%
t2HIGH↗ Escalate35%
t3LOW✓ Auto-resolve90%
t4LOW✓ Auto-resolve90%
t5LOW✓ Auto-resolve90%
t6LOW✓ Auto-resolve90%
t7LOW📷 Request evidence90%
t8MED↗ Escalate70%
t9LOW✓ Auto-resolve90%
t10LOW✓ Auto-resolve90%
t11LOW✓ Auto-resolve90%
t12LOW✓ Auto-resolve90%
t13HIGH↗ Escalate35%
t14LOW✓ Auto-resolve90%
t15LOW✓ Auto-resolve90%
t16LOW✓ Auto-resolve90%
t17HIGH↗ Escalate45%
t18LOW✓ Auto-resolve90%
t19LOW✓ Auto-resolve90%
t20LOW✓ Auto-resolve90%
t21LOW✓ Auto-resolve90%
t22LOW✓ Auto-resolve90%
t23LOW📷 Request evidence90%
t24LOW✓ Auto-resolve90%
t25LOW✓ Auto-resolve90%
t26HIGH↗ Escalate20%
t27HIGH↗ Escalate20%

Numbers reflect the current stub baseline. Run npm run eval to regenerate after the Foundry agent is wired in.