CoverageIQ

Inspiration

We were inspired by a frustrating real-world problem in healthcare: answering a simple question like “Does this insurer cover this drug for this condition?” is surprisingly slow and manual. Medical benefit drug policies are buried in PDFs, written differently by every payer, updated on different schedules, and difficult to compare.

What it does

CoverageIQ turns complex insurance drug policy PDFs into structured, searchable data. It helps users quickly find coverage details such as whether a drug is covered, what prior authorization or step therapy rules apply, which indications are approved, and when a policy changed. Instead of manually reading payer documents one by one, users can search, compare insurers, and track policy updates in one place.

How we built it

We built CoverageIQ as a system that converts unstructured insurance drug policy PDFs into structured, queryable data. The platform is designed to: • ingest policy documents from payers • extract important fields such as drug name, payer, coverage status, prior authorization, step therapy, site-of-care rules, and effective dates • store policy versions over time • support semantic search using embeddings • compare policies across payers • surface change logs between document versions

On the frontend and backend side, we focused on building a usable workflow where someone can move from raw documents to actionable policy intelligence much faster than manual review.

Challenges we ran into

One of the biggest challenges was the lack of standardization. Every insurer structures policy documents differently, uses different terminology, and updates documents in inconsistent ways. That makes extraction difficult because the problem is not just reading text — it is identifying meaning reliably across many formats.

Another challenge was handling version tracking. A small wording change in a policy can have a big business impact, so we needed a way to detect and present meaningful differences instead of just storing files.

Accomplishments that we're proud of

We are proud that CoverageIQ tackles a painful, high-friction workflow with a practical solution. Instead of asking users to dig through PDFs manually, the project moves toward a system where policy knowledge becomes structured, searchable, and easier to act on.

What we learned

We learned that document intelligence is not only an AI problem — it is also a data modeling, product design, and usability problem. Extracting text is one step, but making that information trustworthy and useful requires good schema design, versioning, comparison logic, and search.

We also learned how important domain context is in healthcare workflows. Users do not just want “document search”; they want fast answers to operational questions, clear comparisons, and confidence that they are looking at the right version of a policy.

What's next for CoverageIQ

The next step for CoverageIQ is turning the prototype into a production-ready platform for policy intelligence. We want to expand beyond basic tracking into a system that can automatically ingest new payer policies, detect changes between versions, and alert users when coverage rules materially shift.