Home · Business · Operations & Logistics · Quality control

PARTLY

As of 13 August 2026, AI can only partly inspect packaging for 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 costsNo comparable packaging inspection alternative price is provided in the supplied tool data.

If this goes wrong: damaged stock is accepted, reaches a customer or weakens your position in a supplier dispute.

What to actually do

  1. 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.

  2. Use a tool built for this

    Second choice

  3. Do it yourself

    The distant third

    A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

    How to actually do it

    1. Open your packaging inspection specification or supplier acceptance criteria and copy the rules for dents, tears, punctures, crushing, stains, seals and labels.
    2. Photograph each package against a plain background in even light, including the front, back, both sides, top, bottom, every seal, every corner and any suspected damage.
    3. Name the image files with a package or consignment identifier, then place all photographs for the same package together.
    4. Paste the prompt into a chatbot, replace the bracketed criteria with your current inspection rules and attach the photographs.
    5. Compare every reported finding with the relevant photograph, removing any claim that is not visibly supported and marking unclear images for another photograph.
    6. Physically inspect the listed hidden or unclear areas, including the contents, underside, internal cushioning and seal condition, and record the final accept, reject or quarantine decision in your normal quality record.
    7. Have a colleague compare the photographs, AI notes and physical findings before releasing stock or sending a supplier claim.

    Prompt

    Inspect the attached packaging photographs for visible damage. Use only what can be seen and do not infer hidden damage. For each package, report: 1. package identifier if visible, 2. damage type such as dent, tear, puncture, crushing, stain, water damage or broken seal, 3. exact location, 4. apparent severity as minor, significant or severe, 5. confidence as high, medium or low, and 6. whether the image is insufficient to judge. Quote the visual evidence for each finding. Do not declare the packaging safe, unsafe, fit for sale or unfit for sale. Do not make a final accept, reject or quarantine decision. End with a list of surfaces or details that need a physical inspection. Our inspection criteria are: [PASTE YOUR PACKAGING DAMAGE CRITERIA].

    Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

What it gets wrong

  • AI cannot see the underside, inside or back of packaging that was not photographed.
  • AI cannot feel softness, dampness, loosened contents or structural weakness through an image.
  • AI cannot reliably apply an unstated tolerance, such as whether a dent is acceptable for your product and customer.
  • AI cannot take responsibility for releasing, rejecting or quarantining stock.
  • AI cannot create strong supplier evidence when the photographs do not show scale, timing, chain of custody or the condition on receipt.

What caps this at PARTLY: physical presence, 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.

AxisScore (0–2)
Output2
Inputs1
Verification1
Liability1
Effort delta1
Total6 / 10

The methodology and its thresholds are published in full.

FAQ

Can ChatGPT inspect damaged packaging from photos?
Partly. It can describe visible dents, tears, punctures, crushing and broken seals from clear photographs, but it cannot inspect hidden surfaces or physically test the packaging. A person must confirm the finding and make the stock decision.
Can AI tell if a package is safe to sell?
No, not reliably from photographs alone. AI can flag visible damage against your criteria, but your quality process must decide whether the packaging and contents are fit for sale.
What photos do I need for an AI packaging inspection?
Take clear, well-lit photographs of every side, the top, bottom, corners, seals and any suspected damage, with the package identifier visible where possible. Include a scale or measurement reference when the size of a dent, tear or puncture matters.
Can AI replace a warehouse quality check?
No. It can screen photographs and produce inspection notes, which may reduce repetitive recording work. It cannot handle packages, inspect hidden damage or carry responsibility for releasing stock.

Nearby answers

All Quality control rulings

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.