Can Pangram tell you whether a product photo is usable?
A supplier sends a polished product photo, and you hesitate before adding it to your shop. Pangram’s new Image 1.0 can flag signs of AI and help you decide what to ask the supplier. Use it as an extra review step when the image’s origin is unclear. You still need to check that the photo shows the product accurately and that you have permission to use it.
A new clue for the person reviewing the photo
TLDR AI and The Rundown AI highlighted Pangram Image 1.0 in their November 5 editions. Pangram’s November 4 launch adds a heatmap that highlights areas it identifies as AI-generated, alongside image and video detection through its API. The company also announced support in its Chrome extension.
If Pangram flags an area of the photo, ask the supplier what was changed there. Pangram’s published accuracy figures come from its own evaluations; they are not a guarantee about the next image in your inbox. We have not tested the detector.
Discovery: TLDR AI, November 5Discovery: The Rundown AI, November 5Pangram’s Image 1.0 launch and evaluation
Illustrative example: one lamp, three different questions
Imagine a fictional homewares shop receiving a picture of a brass lamp on a sideboard. This is an illustrative example, not a client story or a detector test.
Suppose the supplier says it generated the room around an unchanged photograph of the lamp. A detector flag on the background would give the owner something specific to discuss. The owner still needs to compare the lamp’s colour, proportions and included accessories with the actual item.
Or the picture could be a genuine photograph of a similar lamp from another collection. A correct classification as a real photograph would still leave you with a picture of the wrong product.
The picture might show the right lamp, while the supplier cannot explain who provided the photograph or whether the shop may use it. The detector cannot supply that missing permission. Ask for an approved image before adding it to the listing.
Keep a short photo review record
Pangram says its detector analyses image content rather than relying on watermarks or metadata. It also displays Content Credentials and camera metadata when available. Those are different kinds of information; none replaces the supplier’s explanation of which product is shown.
We recommend keeping a short review record for each image you are unsure about. Ask a person to review flagged images before rejecting a product listing. Check the product and permission before publishing, even when no AI is detected.
- Record the supplier, product reference and original file so the image can be traced back.
- Note any declared edits and the detector’s result, including the date and any uncertainty.
- Record who checked the visible product details and where permission to use the image was confirmed.
How Pangram analyses images and displays metadataIf your problem is matching photos to product references
Check the current limits before paying
On November 6, Pangram lists its Individual plan at 20 US dollars per month. The plan includes 100 image scans and 10 US dollars of API credit each month. Its Image 1.0 API rate is 25 US cents per image.
Free access is advertised, but the product page gives inconsistent daily limits. Check the limit shown in your session before starting. Do not plan daily work around an assumed number of free scans.
There is also a resolution mismatch. The launch says the minimum has dropped to 256 × 256 pixels, with reduced confidence below 512. The live upload interface still asks for at least 512 × 512 pixels. For a first trial, use a file that meets the displayed requirement; do not enlarge a tiny thumbnail simply to make it pass.
Pangram’s current plans and API pricesCurrent upload requirementsAnnounced resolution change
When the extra check earns its place
Try a small manual review if unfamiliar images regularly arrive and your team struggles to decide what to ask the supplier. Start with public material or images created expressly for testing. Compare the questions and review effort with your usual process before considering an integration.
Wait if you already have a clear source, accurate product photography and permission records. If the real problem is an unidentified product or missing permission, resolve that directly. Add the detector only if it helps your team make those decisions.
Sources
Simplify your photo approval process.
Describe how photos reach your shop and where approval gets stuck. A public image or a fictional example is enough for forvertz to discuss a practical next step.
Discuss your approval processUpdated