Receive FNOL data (claim number, VIN, contact info) and trigger a photo-capture request to the claimant via SMS or email link pointing to a mobile photo-upload flow.
Once photos are uploaded, POST the image set along with VIN and loss description to the estimatics provider's API (e.g., Mitchell, CCC Intelligent Solutions, or Audatex — consult the provider's partner portal for their specific endpoint and auth method) to initiate an AI estimate job.
Poll the estimate job status endpoint until the status reaches COMPLETE, then retrieve the structured estimate response containing labor hours, part costs, and total repair amount.
Compare the returned estimate against your carrier's total-loss threshold (typically replacement cost value minus salvage value relative to ACV; consult your carrier guidelines for the exact threshold) to route to repair vs. total-loss workflow.
Write the estimate data back to the claim record in your claims management system and attach the PDF estimate as a claim document.
If the estimate requires a reinspection or supplement, invoke the reinspect endpoint (provider-specific) with the original estimate ID and updated damage notes.
Known gotchas
Photo quality gates — blurry, insufficient, or missing angle photos cause the AI engine to return a low-confidence estimate or reject the job entirely; implement front-end guidance (overlay prompts, required angle checklist) before submission.
Estimatics providers update labor rate databases and part prices on a schedule; estimates created days apart for similar losses may differ materially — document the estimate date in the claim record.
Supplement cycle creep is common: the initial photo estimate often understates damage revealed during teardown; design the workflow to accept supplement submissions without overwriting the original estimate record.
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp