The Veracity engine evaluates how faithfully AI models reproduce cited grounded search sources by scoring survived, lost, and distorted claims to compute an overall fidelity score.

flowchart LR
    A[Grounded Response] --> B[Resolve URLs]
    B --> C[Fetch Content]
    C --> D[Assess Claims]
    D --> E[Survived Claims]
    D --> F[Lost Claims]
    D --> G[Distorted Claims]
    E --> H[Compute Fidelity Score]
    F --> H
    G --> H