Skip to content
TruthVouch Docs

Core Concepts

TruthVouch provides three integrated layers of AI governance to help organizations navigate, govern, and trust their AI systems.

Three Layers of AI Governance

1. Request/Response Scanning

The Governance Pipeline inspects every AI interaction before and after it happens:

  • Real-time threat detection — Identify PII, injection attacks, toxic content
  • Policy enforcement — Block requests that violate organizational rules
  • Compliance checking — Ensure GDPR, HIPAA, SOC2 compliance

2. Behavior Monitoring

The Sentinel Agent monitors how your team uses AI tools on their devices:

  • Desktop monitoring — Track which AI tools are used (ChatGPT, Claude, Gemini, Copilot)
  • Local enforcement — Policies enforced offline without cloud delays
  • Encrypted reporting — Usage data securely reported to your dashboard

3. Truth Verification

Truth Nuggets let you certify factual content with cryptographic proof:

  • Content certification — Mark verified facts with tamper-proof certificates
  • Hallucination detection — Compare AI outputs against your verified knowledge base
  • Trust scoring — Rate AI responses for factual accuracy

Key Concepts Explained

Hallucination Detection

Detect when AI models generate false or unsupported claims by comparing their responses against your verified facts.

Learn more: Hallucination Detection

Natural Language Inference (NLI)

Determine if one statement logically follows from another — the core technology powering hallucination detection.

Learn more: NLI Scoring

Neural Cache

Intelligent multi-layer caching speeds up fact verification from seconds to milliseconds by recognizing repeated and similar queries.

Learn more: Neural Cache

Embeddings

Convert text into semantic vectors that capture meaning, enabling fast semantic matching even when questions are phrased differently.

Learn more: Embeddings

Governance Pipeline

A 15-stage scanning pipeline that processes every LLM request and response, enforcing policies and detecting threats in milliseconds.

Learn more: Governance Pipeline

Trust Score

A 0-100 score representing your organization’s AI trustworthiness based on hallucination detection, correction rates, and compliance.

Learn more: Trust Score

GEO Scoring

Evaluates how well your content is positioned and discoverable across AI engines like ChatGPT and Claude.

Learn more: GEO Scoring

Audit Trail

Complete hash-chained record of every decision and action, with cryptographic tamper-detection for compliance audits.

Learn more: Audit Trail

Content Certification

Cryptographically sign your verified content so anyone can verify it hasn’t been modified.

Learn more: Content Certification

Multi-Provider Verification

Simultaneously verify claims across 9+ AI engines (ChatGPT, Claude, Gemini, Perplexity, and more) to catch hallucinations no single model would miss.

Learn more: Multi-Provider Verification

Next Steps

Ready to explore? Start with:

Or jump to the Platform Guide to start configuring TruthVouch for your organization.