Prior-Authorization Checklist Agent LIVE — runs in your browser

The full agent, no installation: policy retrieval and the completeness check run as JavaScript in this page, and the two reasoning steps are real LLM calls made directly from your browser. Yellow trace lines are live model calls — the pause is the model reading the policy.

Preset scenarios
1 · Payer
2 · Procedure requested
3 · Documents in the submission
Execution trace
Waiting for a run...

Ask about this project — architecture & design Q&A

A chatbot grounded in this project's fact sheet. Ask how the agent works, where the policy documents live, what's deterministic vs. LLM, how it was tested, or why we made a design choice.