LexiRule provides a complete end-to-end pipeline — from raw document to live eligibility form — with AI assistance and human oversight at every step.
Upload any PDF, DOCX, or paste a URL. Gemini Vision "reads" the document page by page, preserving tables, charts, and complex layouts as accurate Markdown — far beyond what traditional text extractors can handle.
Our extraction agent (powered by OpenAI) reads the Markdown and identifies three types of rules: binary qualifications, weighted scoring criteria, and conditional date-based logic — all output as a structured JSON RuleBase.
A second Gemini Vision agent takes the drafted rules and the original PDF and visually verifies each rule — finding missed sections and providing bounding-box evidence for every match. No hallucinations go undetected.
Admins review and edit extracted rules through a clean, hierarchical interface — no JSON required. Each rule can be edited, reordered, approved, or deleted with full audit logging. View the original document right alongside the rules.
Once rules are approved, LexiRule generates a minimal, adaptive questionnaire. Only relevant questions are shown. The entire scoring and evaluation is done with pure logic — no AI, no network call, no guessing. Results are explainable to every applicant.