Photos first: using AI vision without misleading anyone
AI can look at a photo and tell you what it probably is, read the text on a label, and spot a scratch you missed. It can also make a tired item look better than it is. Only one of those is a good idea.
Where AI vision earns its keep
- Identification — "what is this?" from a clear photo of the item and any markings
- Reading labels — model numbers, serials, capacities, dates, in seconds
- Checking your photos — ask it to list every flaw it can see. It's a second pair of eyes before the buyer's
- Suggesting the shots you're missing — the back, the ports, the label, the fault
Where to stop
Buyers of used items need to see the actual item they'll receive. Background clean-up and straightening are fine; anything that hides wear, changes a colour or "fixes" a crack is misdescribing, and it comes back as a return.
The simple test: if the buyer held the item next to your photo, would they feel misled? If there's any chance of yes, use the original.
The simple test: if the buyer held the item next to your photo, would they feel misled? If there's any chance of yes, use the original.
A workflow that works
- Shoot in good light: the front, back, label, every flaw close up
- Ask the AI to identify the item and read the label — then check the label yourself
- Ask it to list every visible flaw, and put each one in the description
- Keep the photos honest; let the words do the selling
AI misreads
- Question
- Active discussion
- Answer emerging
- Community solution
- Knowledge article