Enziu

Insurance policies span 50–100+ pages of legal language that 8.8 million claimants were burned by in 2024. ENZIU is the Universal Insurance Transparency Engine built to reverse this asymmetry. Upload any insurance PDF — any carrier, any format — and receive a structured ENZIU Index scoring your policy across Clarity, Coverage Alignment, and Claim Efficiency. Unlike existing tools, ENZIU audits the document you actually signed. Summarization is not auditing. A summary tells you what a policy says. An audit tells you what it means for you — scored against objective criteria, with reproducible grades that hold up to scrutiny. ENZIU is a two-agent AI system: - Agent 1 (Llama 4 Scout 17B, 890K context) — extracts facts across 18 categories as a licensed bad-faith attorney: exclusions, waiting periods, appeal rights, jargon density, and 8 general legal risk categories - Agent 2 (Llama 3.3 70B) — scores Clarity (0–30), Coverage (0–40), and Claim Efficiency (0–30); applies two-source red flag detection; outputs the ENZIU Index (0–100) with letter grade A+ to F Same policy always yields the same grade (±1 band). Temperature 0.0. No chunking — full policy in one pass. Privacy-first: Zero server storage. AES-256-GCM client-side encryption. No PII — only a bcrypt-hashed voucher passphrase. Recovery Vault keyed by SHA256(voucher_code). Built on AMD Developer Cloud. Production inference via NScale.