Home · Business · Operations & Logistics · Quality control
As of 13 August 2026, AI can only partly check your stock for visible damage.
This still needs a person who signs their name to it.
Can you do it?
5 minutesto a draft.
30 minutesto something you’d act on.
Cost, all in£0
Skill neededchat-fluent
Who has to check ita colleague
What the alternative costsThe supplied tool list does not include a price for a human stock inspection.
If this goes wrong: damaged stock is accepted or good stock is rejected, causing returns, waste, customer complaints or a dispute with your supplier.
What to actually do
Hand it to a person
The route this page recommends
A person who owns the outcome does this end to end, worth it when the failure is dear.
Use a tool built for this
Second choiceDo it yourself
The distant thirdA chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.
How to actually do it
- Open your current quality-control procedure and damage criteria, and copy the rules for visible damage into the prompt where indicated.
- Gather clear photographs of each item from all relevant sides, including close-ups of marks, and keep the item code in each filename or alongside each image.
- Upload the photographs to a multimodal chatbot and paste the complete inspection prompt.
- Check that the response has an entry for every item and that each claimed defect is visible in the cited photograph.
- Retake blocked, dark or distant photographs and run those items through the prompt again, keeping any item marked uncertain in the review list.
- Ask a trained colleague to decide every uncertain or consequential case against the quality-control procedure, then record the final accept, repair, discount or reject decision in your stock system.
Prompt
Inspect the attached photographs of stock for visible physical damage only. Work through each item separately and use its item code or filename as the identifier. For each item, report: condition as "no visible damage", "visible damage" or "uncertain"; the exact visible location; a short description of the damage; confidence as high, medium or low; and the photograph or view supporting the finding. Do not infer hidden, internal or functional damage. Do not treat unusual colour, packaging variation or a reflection as damage unless the photograph clearly supports that conclusion. If an angle, image quality or lighting prevents a reliable decision, mark the item "uncertain" and state what additional photograph or human check is needed. Apply these inspection criteria: [paste your written damage criteria here]. Return a table followed by a separate list of items needing a colleague's decision. Do not recommend accepting or rejecting stock unless those rules explicitly say how to decide.
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
What it gets wrong
- Cannot see damage that is hidden, internal, tactile or revealed only when the item operates.
- Cannot reliably separate a serious defect from an acceptable manufacturing variation without your current inspection standard and context.
- Cannot replace a trained inspector's judgement on borderline cases or decide how much damage your business should tolerate.
- Cannot take missing angles, improve a poor physical inspection setup or confirm that the photographs show the correct item.
- Cannot carry responsibility for a wrong accept or reject decision.
What caps this at PARTLY: judgement under ambiguity, verification cost and stakes of error.
How we scored this
Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 1 |
| Verification | 1 |
| Liability | 1 |
| Effort delta | 2 |
| Total | 7 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can ChatGPT check photos of my stock for damage?
- Partly. It can identify and describe clearly visible damage in supplied photographs, but it cannot assess hidden or internal faults and may misread reflections, packaging or poor lighting. A colleague should decide uncertain cases.
- Can AI tell if a product is damaged?
- It can flag visible damage when the photographs are clear and your inspection rules are provided. It cannot reliably decide whether an ambiguous mark is acceptable or detect faults that are not visible.
- Can AI replace a quality inspector?
- No. AI can reduce the time spent sorting photographs and recording obvious findings, but a quality inspector is still needed for unclear views, physical checks and final accept or reject decisions.
- How do I use AI to inspect stock photos?
- Photograph each item from all relevant sides, label the images, provide your written damage criteria and ask the model for an item-by-item table with evidence and an uncertainty flag. Compare every finding with the photograph and send uncertain or consequential cases to a trained colleague.
Nearby answers
- Can AI create a quality control checklist for my business?YES
- Can AI generate quality inspection reports?PARTLY
- Can AI investigate the root cause of product defects?PARTLY
- Can AI trace a product batch through my supply chain?PARTLY
- Can AI audit my supplier remotely?NO
- Can AI check allergen information on my food labels?PARTLY
Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.
The newsletter
AI news, new answers and product picks, straight to your inbox.