Insurance policies in India are designed to be signed, not read. The average policy is 50+ pages of dense legalese, hiding critical "gotchas" like Room Rent Capping, Co-pay clauses, and ambiguous Waiting Periods. Most people don't know what they bought until their claim gets rejected.

TermScope is a forensic audit engine that "reads" the fine print so you don't have to. You upload your policy PDF, and our AI pipeline extracts, analyzes, and benchmarks the clauses against a strict "Risk Rubric." No sales agents, no bias—just raw code decoding the contract.

Under the Hood (The Stack):
I didn't want this to be just another "PDF wrapper," so I built a Multi-Model Architecture to ensure forensic accuracy:

• Smart Routing: The system intelligently classifies the policy type on the fly to select the correct audit framework.

• Neuro-Symbolic Logic: We don't just ask the AI to "find risks." We use a Hybrid Pipeline where the LLM performs extraction, but a deterministic Rule Engine applies the scoring (Red/Yellow/Green). This eliminates the "hallucinated risk" problem common in pure-LLM wrappers.

• Ground Truth Injection: Critical financial metrics (like CSR) are cross-referenced against real-time IRDAI data, not read from outdated PDFs.

• Current Status: v0.1 (MVP). The core Forensic Engine is live and stable for Health & Term Life. The UI is raw.

Help Needed: I need "Stress Testers." Upload an old policy you have lying around. Does the AI find the risks you know about? Did it hallucinate a clause? Roast my logic in the comments.