Perf Verify
AI-generatedSummary
Verify text for factual accuracy and route results to Verified or Issues Found based on confidence and retraction thresholds.
Inputs
- text (required) — The text to verify for factual accuracy
- mode — Verification depth — faster modes are quicker but less thorough
- options — Collection of additional verification settings such as confidenceThreshold, maxRetractedClaims, etc.
Output shape
a list of verification result objects per input item
Each result includes original_text, status, composite_confidence, total_claims, corrected_claims, uncertain_claims, retracted_claims, mode, latency_ms; optional corrected_text and claims array are included when enabled
Examples
Example 1: Check an article before publishing
mode: standard, confidenceThreshold: 0.9, includeClaims: true, includeCorrectedText: true
Example 2: Validate a batch of FAQ answers quickly
mode: fast, confidenceThreshold: 0.7, maxRetractedClaims: 1, includeClaims: false
Example 3: Deep audit of a press release
mode: thorough, confidenceThreshold: 0.95, maxRetractedClaims: 0, includeClaims: true, includeCorrectedText: true