Evidence-Grounded AI Workflow
From live recall evidence to explainable safety triage.
This demonstration retrieves current U.S. Consumer Product Safety Commission recall evidence, evaluates transparent prioritization factors, preserves an auditable tool trace, and limits public narration to checksum-protected recordings that passed human perceptual review.
01
Live CPSC Evidence
02
Explainable Triage
03
Auditable Decision Support
Agent workflow running
Retrieving recall evidence and evaluating transparent safety factors.
The demonstration could not complete.
An unexpected error occurred.
Executive Briefing
Highest-priority recall
Source Evidence
Recall evidence
Hazard
Incidents and injuries
Recommended action
Explainability
Triage score factors
—
Governed Narration
Approved voice segments
Approved references only
Audio is available only where the exact spoken text matches a checksum-protected recording that passed human perceptual review. Other evidence remains text-only.
0
Approved audio segments
0
Text-only segments
Disabled
Dynamic public TTS
View text-only evidence segments 16 withheld
These segments remain visible for evidence completeness but are not publicly narrated because they have not yet passed perceptual validation.
Evidence Governance
Runtime controls
Agent Observability