Scan a product photo lives at the top of the Add inventory item page (`/app/inventory/items/new`). It takes one photo of a product or its label — a bottle, box, tool, or anything else you'd stock — and returns a structured guess at the item's name, brand, SKU, category, unit of measure, size, and a short description. Nothing is written to your inventory from this step: the fields it fills in land in the same Add inventory item form you'd otherwise type into by hand, and the item is only created when you review those fields yourself and submit the form through the normal Create item button.
Scanning a product
Drop or snap a single photo into the "Scan a product photo" panel at the top of the form — the tool works from one image per run, so pick the clearest shot of the label or packaging you have. Click "Extract details" (it stays disabled until a photo is attached, and shows a brief loading state while the read is in progress). When it succeeds, the fields further down the form — name, brand, SKU, category, unit, size, and description — get pre-filled from what the AI found, alongside a confidence percentage and, when there's something worth flagging, a short note.
Reviewing what came back
The extractor only fills in what it can actually read off the label. For most fields — name, brand, SKU, size, description — anything it can't make out comes back blank rather than guessed. Category and unit of measure are the two exceptions: since the form needs some value in those fields, an unreadable label falls back to a generic bucket (Other category, Each unit) rather than staying blank, so those two are worth a second look after every scan. It also deliberately never touches cost, price, or on-hand stock count, since a photo of a label can't tell you what you paid for something or how many you have on the shelf. Everything it does fill in sits in the same editable inputs you'd use if you'd typed it yourself, so it's worth a quick check before you save — especially the category field, since the AI is picking its single best match from a fixed list of shop-supply categories (ceramic coating, detail chemical, tint tool, PPF supply, and so on) and won't always land exactly where you'd have filed it.
Things to know
- Plan requirement: unlike some of SalesThumb's other vision-AI tools, this one isn't locked to a specific plan tier — it's available regardless of plan. The only thing that can block it is the shop's own 30-day AI spend allowance (and any AI Booster credits on file to cover it).
- Stub mode: if the workspace doesn't have AI configured, if the shop's spend allowance is used up with no Booster credits to cover it, or — on the rare occasion the AI's reply doesn't come back in a form the app can use — the button still runs but doesn't get a real read of your photo. The app is built to catch this on your behalf: a non-live result never gets auto-filled into the form; instead you'll see a short note explaining why nothing was applied, so you can fill the item in by hand.
- Category and unit always get a value. Since the form needs one, an unreadable label defaults to a generic bucket (Other category, Each unit) rather than coming back blank — worth a second look after every scan.
- One photo per run. If you want to try a different angle or a clearer shot of the label, swap the photo and run it again.
- It never guesses money or stock. Cost, price, minimums, reorder points, and opening stock all stay exactly as you left them — those come from you, not the photo.
- It doesn't save anything by itself. The extractor only proposes values for the form fields above it; creating the actual inventory record still goes through the same Create item action as always, so you can edit, clear, or override anything before it's saved.
Frequently asked questions
Q: Does scanning a photo save the inventory item automatically?
A: No. It only pre-fills the Add inventory item form — you still review the fields and click Create item yourself to actually save anything.
Q: How many photos can I scan at once?
A: One per run. Drop or snap a single photo of the product or its label, then click Extract details.
Q: Will it fill in cost, price, or how many I have on hand?
A: No. It only reads what's printed on the label — cost, price, and stock counts are never guessed or filled in; you enter those yourself.
Q: What does the confidence percentage mean?
A: It's the AI's own estimate of how sure it is about what it read off the label. Treat a low number as a cue to double-check the fields before you save.
Q: What plan do I need for this?
A: None specifically. Inventory photo extract isn't locked to a plan tier the way some of SalesThumb's other AI tools are — it's only gated by the shop's own AI usage allowance.
Q: What happens if AI isn't live for my shop when I click Extract details?
A: The button still runs, but since there's no real read of your photo, the app blocks the result from auto-filling the form and shows a short note instead explaining why — for example, that AI isn't configured on this deployment, or that your shop's AI spend allowance for the period is used up. You can then fill the item in by hand.