Check job-site photos before a milestone invoice goes out
Before a contractor sends a milestone invoice, a decision model checks the site photos against the milestone description and flags any milestone the photos do not clearly show as finished for the project lead.
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
Small contractors bill by milestone, such as framing complete or drywall hung, and invoices sometimes go out before the work is visibly done, which leads to disputes. Jev looks at the latest site photos alongside the milestone text and answers whether the photos show the milestone as complete, how far along the work appears on a four-level scale, and whether the photos are clear and current enough to judge. Milestones that clear a 0.85 confidence are marked ready to invoice. Everything else goes to the project lead with the photos. Jev does not send invoices or decide what the client owes. The project lead confirms progress, sometimes with a site visit, and approves every invoice.
Example workflow
- Crew leads upload end-of-day site photos to a shared folder per job in Google Drive or CompanyCam.
- When a milestone is marked done in the job sheet, Make collects the latest photos and the milestone description.
- Jev answers a Noul question, the photos show this milestone as complete, a Score question rating progress as not started, in progress, nearly done, or complete, and a Noul question, the photos are clear enough to judge.
- Milestones that clear the threshold are marked ready to invoice in the job sheet.
- Unclear or incomplete milestones are posted to the project lead in Slack with the photos attached.
- The project lead approves each milestone, and the invoice is sent from the accounting system by a person.
Tools
- Jev (TypeSafe AI)
- Google Drive
- CompanyCam
- Make
- Google Sheets
- Slack
- QuickBooks Online
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 can show one room while the milestone covers the whole job — guardrail: ask crews to photograph each area listed in the milestone, and flag milestones with too few photos.
- Older photos may be resent — guardrail: code checks upload dates and ignores photos taken before the milestone started.
- A wrong pass could lead to an early invoice — guardrail: no invoice is sent from Jev's answer; the project lead approves each one.
Related recipe
B2B Invoice AR Reminder Agent with Approval Gate: An agent reads open invoices from QuickBooks or Xero, skips blocked accounts, and drafts tiered payment reminders in Gmail that a person approves one by one before anything is sent.
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:
- Google Drive (official page): developers.google.com/workspace/drive (checked 2026-11-19)
- Make MCP server (official page): developers.make.com/mcp-server (checked 2026-11-09)
- Slack MCP server (official page): docs.slack.dev/ai/slack-mcp-server (checked 2026-11-09)
- QuickBooks Online MCP server (official page): github.com/intuit/quickbooks-online-mcp-server (checked 2026-11-09)