Check shelf and display photos against the planogram
Store staff send a photo of each display, and a decision model checks it against the planogram for that bay, flagging gaps, wrong products, or missing price tags for the store lead to fix.
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
Brands and small retail chains agree a layout for each shelf or display, and keeping it right across locations depends on someone checking. Here, staff photograph a bay at opening, and Jev answers a fixed set of questions about it with the planogram described in text alongside the image: is the bay fully stocked, are any unexpected products present, is the price tag visible on each shelf, and how closely does it match overall on a four-level scale. A Box answer marks the largest gap. Bays where every answer clears a 0.85 confidence are logged as checked. Bays with a problem or a low-confidence answer go to the store lead, who decides what to restock or move and whether to raise it with the supplier.
Example workflow
- Staff take a photo of each assigned bay in a shared Google Form or a Slack channel each morning.
- Make attaches the text planogram for that bay from a Google Sheet and sends both to Jev with the image preview header.
- Jev answers Noul questions for fully stocked, unexpected products present, and price tags visible, a Score question rating match from poor to exact on four levels, and a Box question marking the largest gap.
- Code logs every answer and confidence in a sheet per store and per bay.
- Bays with a problem or a low-confidence answer are posted to the store lead in Slack with the photo and the box drawn on it.
- A weekly summary of repeat gaps by bay is shared with whoever places stock orders.
Tools
- Jev (TypeSafe AI)
- Google Forms
- Google Sheets
- Make
- Slack
- Looker Studio
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.
- Glare, angles, and partial photos can hide products — guardrail: a Noul question asks whether the whole bay is visible, and unclear photos are sent back for a retake.
- A text planogram may not capture every facing — guardrail: start with a few high-value bays and add detail only where the checks prove useful.
- Staff may feel watched rather than helped — guardrail: share the purpose openly and use the results for restocking, not for individual performance reviews.
Related recipe
Weekly KPI Digest from Google Sheets with Source-Cited Numbers: An agent reads your defined metrics from Google Sheets, compares them to last week and targets, and drafts a plain-English digest where every number cites its cell. The owner approves before it goes to Slack or email.
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:
- 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)