Verity Lex: Building an AI Readiness Assessment Tool for Government Courts

Short summary

A government worker built Verity Lex during OpenAI Build Week to assess whether courts are ready to adopt AI, using GPT-5.6 for document retrieval and extraction while keeping scoring in a deterministic TypeScript rule engine. The key design principle: the AI handles perception (finding and reading documents) but never judgment (assigning compliance scores), ensuring reproducibility that government buyers can audit. The tool currently covers Santa Barbara's superior court and returns an AI-readiness score anyone can recompute from cited evidence.

• •Verity Lex separates AI perception (document retrieval/extraction) from deterministic judgment (rule-based scoring) for government compliance
• •The tool reads public court records against published legal standards and returns a reproducible readiness score
• •Design principle: the model has no path to the score by construction, ensuring auditability for public-sector procurement

Generated with AI, which can make mistakes.