Snap a photo to import inventory

Last updated: April 2026

When you have an inventory list on paper, on a screen you can't export from, or scribbled in a notebook, photo capture skips the manual data entry. Snap a photo, AI reads the rows, and you land on the same Google Sheets-style review grid as a CSV upload — fix any cells the AI got wrong, then import.

Two ways to capture

  • Use camera — Add Source → Snap a photo → Use camera. On a phone this fires the native camera; on a laptop you can pick image files from your computer.
  • Show QR for phone — generate a one-time QR code on your laptop, scan it with your phone's camera, and the phone-side capture page lets you snap multiple photos. Photos are pushed back to your laptop session automatically.

Tips for good extractions

  • Lay the sheet flat, light it evenly, and hold the camera roughly square to the page.
  • Get the whole table in one frame. If the table is too long, take multiple photos — each photo becomes its own tab in the review grid.
  • Block letters and clear spacing extract better than cursive.
  • For computer screens — set screen brightness high, dim the room, and avoid glare.

Multiple photos in one import

Stack up to 12 photos in one session. Each photo lands as a separate tab at the bottom-left of the review grid (just like Google Sheets tabs). Rename, fix, or skip rows tab-by-tab, then click Import once everything looks good.

When extraction is wrong

AI extraction isn't perfect, especially on handwriting or cluttered photos. The review grid lets you fix anything inline — required fields with red borders block import until filled. Every cell you correct teaches our system how to read the next photo from your shop better, so accuracy compounds over time.

Privacy

Photos are stored encrypted in your private Techfleet bucket. They're used only to extract the inventory rows and to improve future extractions for your shop. Photos are never shared with other merchants. Auto-purge after 30 days.

NOTE

Photo capture uses Anthropic Claude vision under the hood — about 5–15 seconds per photo. Cost averages a few cents per import; we track and alert if usage spikes.

Was this article helpful?