The TRACE Protocol Five Dimensions of Accountable AI
Every AI response is scored across Transparency, Reasoning, Auditability, Compliance, and Explainability. TRACE transforms opaque model outputs into structured, evidence-backed governance artifacts.
Grounded in 3 source documents with 6 evidence chunks. 0 chunks have high relevance (>70%).
206 knowledge graph entities across 19 types connected to your query. Evidence chain spans 3 independent sources.
Full audit trail logged with SHA-256 hash chain. Each retrieval step recorded with timestamps and provenance.
Data processed in eu-central-1 (GDPR compliant). EU AI Act Art. 50 transparency notice included.
Response structured for readability. Source attribution for every claim.
Based on 6 document chunks and 205 knowledge graph entities across 3 independent sources.
How TRACE Scoring Works
From query to evidence chain in four steps
Submit Query
User sends a natural-language query to any connected LLM through the TraceGov interface.
CERI Injects Governance
Before the LLM responds, CERI embeds relevant regulatory constraints directly into the inference context.
TRACE Scores Generated
The response is scored across all 5 dimensions: Transparency, Reasoning, Auditability, Compliance, and Explainability.
Evidence Chain Recorded
Every score, source, and decision is hash-verified and written to the immutable audit trail with full provenance.
TRACE Capabilities
Everything you need to score, explain, and audit AI at scale
Structural Scoring
Each TRACE dimension receives an independent score derived from structural analysis of the AI response, not surface-level heuristics.
5 Confidence Tiers
GREEN (verified), BLUE (high confidence), ORANGE (moderate), RED (low confidence), MAROON (unverifiable) -- instant visual signal.
6 Explainability Metrics
Source attribution, reasoning chain depth, uncertainty quantification, bias detection, factual grounding, and regulatory alignment.
GDPR Art. 22 Auto-Trigger
Automatically detects when an AI response constitutes automated decision-making and triggers the mandatory human review workflow.
Real-Time Scoring
TRACE scores are computed in real time as the LLM streams its response. No batch processing, no delays.
API Access
Full REST API access to TRACE scores, dimensions, and evidence chains for integration into your existing compliance workflows.
136days until enforcement
Score Your First AI Response
See TRACE in action on your own queries. Free plan includes full scoring.