Claim form to structured data

The brief

Claims are rejected or delayed because a form field or document is missing. Build a tool that reads a scanned claim form and bills into validated JSON, and lists anything missing before the customer files the claim. Good looks like: a schema, accuracy per field on labelled forms, and a correct missing-documents list.

What you will build

1. Define the claim schema and required documents
2. Extract fields with a vision-capable LLM and structured outputs
3. Validate and list what is missing
4. Measure accuracy on 30 labelled forms
5. Build a simple upload-and-check page

Free, open data to build it with

• FUNSD: form understanding in noisy scanned documents ↗ FUNSD
• CORD: receipt understanding dataset ↗ Clova AI (GitHub)

Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.