ClaimAssist
AI Agent / Insurtech
An AI claims-processing agent for insurers — reads claim documents and photos, extracts structured data, scores severity and routes each claim to the right handler with a confidence gauge.
The problem
- 01
Turning messy claim submissions — PDFs, photos and free-text — into structured data a handler can act on without manual keying.
- 02
Scoring claim severity and fraud risk consistently across policy types so urgent claims aren't buried in the queue.
- 03
Routing each claim to the right handler by specialism and workload, with a confidence gauge the team can trust.
What was built
AI Agents
A claims agent ingests documents and photos, extracts structured fields, and proposes a severity score and route — handlers review and approve.
Backend Functions
Server-side document parsing and extraction logic keeps API keys off the client and returns one consolidated claim payload.
Entities
Claim, policy, document and handler records modelled as related JSON schemas driving the queue and routing rules.
Integrations
InvokeLLM to summarise free-text submissions; SendEmail to notify handlers and claimants at each stage.
The outcome
Shipped in 12 days. Handlers now open a claim with structured data, a severity score and a suggested route already attached — average first-touch time dropped from hours to minutes.
The
terminal.
Send a brief — a sentence or a spec doc. You'll get a fixed price and a build plan back within 24 hours.