Sort customer vehicle damage photos before booking an estimate
An auto body shop asks customers to upload damage photos when they request an estimate, and a decision model sorts each by damage area and size so the estimator books the right slot.
Uses image input, which TypeSafe AI labels a preview feature for Jev. Test it on your own sample images before relying on it.
An example pattern written by Agents Lookup. We have not yet found a public build of this exact job.
The use case
Body shops get estimate requests with a few phone photos, and booking the wrong slot length wastes a bay. Jev looks at the photos and makes two decisions: which area of the vehicle is damaged, and how large the job looks on a fixed scale from minor to major. It also asks whether the photos are clear enough to judge at all. Requests where every answer clears a 0.8 confidence get a suggested slot length. Unclear photos trigger a request for better ones, and low-confidence answers go to the estimator. Jev never gives a price or decides on insurance matters. The estimator reviews the suggestion, books the slot, and writes the estimate after seeing the vehicle.
Example workflow
- Customers submit an estimate request form with two to six photos of the damage.
- Make sends the photos and the customer's description to Jev with the image preview header.
- Jev answers a Choice question, which area is damaged, with the allowed answers front, rear, driver side, passenger side, roof or glass, and multiple areas, a Score question rating size as minor, moderate, or major, and a Noul question, the photos are clear enough to judge.
- Code maps the area and size to a suggested slot length and writes it to the booking sheet with the confidence scores.
- If the photos are unclear, Claude drafts a polite request for better photos, which staff review and send.
- The estimator confirms or changes the slot and books it with the customer.
Tools
- Jev (TypeSafe AI)
- Jotform
- Make
- Google Sheets
- Claude
- Gmail
- Slack
Risks and guardrails
- Image input is a preview feature in Jev 1.14 and TypeSafe says it is not production-ready yet — guardrail: test it on a few hundred of your own photos against a person's labels before relying on it, and keep a person checking every answer until the results hold up.
- Photos hide damage behind panels — guardrail: suggestions are for booking only, and the estimate is always written after an inspection.
- Customers may treat the suggestion as a quote — guardrail: no size or slot answer is shown to the customer, only the booking time.
- Low-light photos skew the size rating — guardrail: unclear photos trigger a request for new ones rather than a guess.
Related recipe
Support Ticket and Review Triage with Policy-Cited Draft Replies: An agent tags every helpdesk ticket and public review by intent, urgency, and sentiment, routes it, and drafts a reply that cites your help center. People approve refunds and all public replies.
Want it set up for you? Request a quote.
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- Jotform MCP server (official page): www.jotform.com/developers/mcp (checked 2026-11-09)
- Make MCP server (official page): developers.make.com/mcp-server (checked 2026-11-09)
- Claude (official page): claude.ai (checked 2026-11-09)
- Gmail (official page): developers.google.com/workspace/gmail (checked 2026-11-19)
- Slack MCP server (official page): docs.slack.dev/ai/slack-mcp-server (checked 2026-11-09)