Home · Business · Operations & Logistics · Inventory & stock

PARTLY

As of 13 August 2026, AI can only partly manage stock returned by customers.

Most people should hand this to a purpose-built tool.

Can you do it?

15 minutesto a draft.

1 hourto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ita colleague

What the alternative costsA stock-control colleague or warehouse team remains the alternative for physically receiving, inspecting and moving returned goods; no price comparison is provided here.

If this goes wrong: you refund the wrong order, put damaged goods back into saleable stock or lose track of an item in the warehouse.

What to actually do

  1. Use a tool built for this

    The route this page recommends

  2. Do it yourself

    Second choice

    A chat interface, chat-fluent skill, and roughly 1 hour until you can act on the result.

    How to actually do it

    1. Open your returns policy, stock-status definitions and current stock-system field list, then paste them into the prompt.
    2. Export the relevant return records and product or order data from your shop, order system or spreadsheet, remove unnecessary customer personal data, and paste one complete record per return.
    3. Ask the chatbot to produce the triage table, missing-information list and physical-inspection checklist using the supplied prompt.
    4. Give each returned item a physical inspection by checking its condition, completeness, packaging, serial number where relevant and any safety concern, then add those findings to the matching return record.
    5. Paste the updated inspection findings back into the chatbot and ask it to revise the proposed disposition and identify policy or stock-data mismatches.
    6. Compare every proposed status, quantity and action with the current stock system and return policy, correcting any unsupported entry before a colleague approves it.
    7. Have the authorised colleague approve refunds, replacements, write-offs and supplier returns, then update the stock system and send only the approved customer communications.

    Prompt

    Act as an operations assistant for a UK business. Help me manage customer returns using only the information I provide. Do not invent order details, product condition, refund decisions, stock quantities, policy rules or dates. Separate confirmed facts from assumptions and exceptions.
    
    My return policy:
    [PASTE POLICY]
    
    My current stock-system fields and allowed stock statuses:
    [PASTE FIELD LIST AND STATUS DEFINITIONS]
    
    Returned-item records:
    [PASTE ONE ROW OR RECORD PER RETURN, INCLUDING ORDER ID, SKU, QUANTITY, CUSTOMER REASON, RECEIPT DATE, AND ANY INSPECTION NOTES]
    
    For each return, produce a table with: order ID, SKU, quantity, evidence available, missing information, proposed disposition, proposed stock status, refund or replacement action to consider, owner, and next action. Use only these dispositions unless the evidence requires an exception: return to saleable stock, quarantine for inspection, repair or refurbish, supplier return, recycle or dispose, or escalate.
    
    Do not approve a refund, classify an item as safe to resell, or change stock records when the evidence is incomplete. Flag any mismatch between the return record, the policy and the stock data. Then provide a short checklist for the colleague who will physically inspect each item, and a separate list of entries ready for manual entry into the stock system. Mark every item requiring human inspection or approval.

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

  3. Hand it to a person

    The distant third

    A person who owns the outcome does this end to end, worth it when the failure is dear.

What it gets wrong

  • AI cannot receive, open, inspect, photograph or physically move the returned goods.
  • AI cannot reliably distinguish a resaleable item from concealed damage, missing parts, contamination or a safety risk without accurate human inspection evidence.
  • AI cannot decide ambiguous cases such as suspected misuse, fraud or an exception to your policy without an accountable colleague.
  • AI cannot guarantee that a proposed stock movement has been written correctly to your live stock system unless a controlled integration performs and records that action.
  • Your business still carries the consequences of an incorrect refund, stock adjustment or resale decision.

What caps this at PARTLY: physical presence, judgement under ambiguity 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

FAQ

Can AI process customer returns?
Partly. AI can extract return details, sort cases, identify missing information and draft proposed stock actions, but a colleague must inspect the goods and approve consequential decisions.
Can AI decide whether a returned item can go back into stock?
It can make a documented recommendation from inspection notes and your policy. It cannot examine the item itself or take responsibility for deciding that damaged, incomplete or unsafe goods are suitable for resale.
Can AI update my inventory after a return?
It can prepare stock adjustments or populate an approved workflow if your systems support that connection. Check every SKU, quantity and status against the physical item and live stock record before committing the change.
Is it safe to use AI for handling customer returns?
It is suitable for triage, record preparation and routine drafting when a colleague checks the evidence. Do not let it make unattended refund, disposal, safety or stock-status decisions.

Nearby answers

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.