As of 13 August 2026, AI can only partly trace a product batch through your supply chain.

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, power-user skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open your stock, purchasing, goods-in, production, delivery and sales systems and export the records that contain the target batch or lot identifier.
   2. Gather the matching supplier documents, goods-in notes, production records, stock movements, delivery notes and customer shipment records, keeping each source reference and date.
   3. Remove unrelated personal information, then paste the records into the prompt with the target batch or lot ID, the direction of trace and the date range filled in.
   4. Ask the model to produce the chronological trace and to keep confirmed, inferred and missing links in separate sections.
   5. Compare every row in the drafted trace with its named source record, checking identifiers, quantities, dates, locations and any split or merge against the original systems.
   6. Send each flagged gap or conflict to the relevant supplier, warehouse, production or quality colleague and add their confirmed evidence to the records.
   7. Run the prompt again with the new evidence, then have a quality or operations colleague approve the final trace before using it for stock decisions, release, withdrawal or recall.

   Prompt

   Trace the product batch through the supply-chain records below. Treat the records as the only source of truth and invent nothing.
   
   Batch or lot to trace: [BATCH OR LOT ID]
   Scope of trace: [SUPPLIER TO CUSTOMER, CUSTOMER TO SUPPLIER, OR BOTH]
   Date range: [DATE RANGE]
   
   Records:
   [PASTE PURCHASE ORDERS, SUPPLIER LOT RECORDS, GOODS-IN RECORDS, PRODUCTION RECORDS, STOCK MOVEMENTS, DELIVERY NOTES, SALES ORDERS, CUSTOMER SHIPMENTS AND ANY RELEVANT QUALITY RECORDS HERE]
   
   Produce:
   1. A chronological trace table with one row for each confirmed movement or transformation.
   2. For every row, include the source record name or reference, date, item, batch or lot identifier, quantity and location or organisation where available.
   3. Separate confirmed links, inferred links and missing links. Do not present an inference as fact.
   4. Flag duplicate identifiers, quantity mismatches, date conflicts, unexplained splits or merges, and records that cannot be joined.
   5. State whether the batch can be traced forward, backward or both, and identify the exact point where traceability stops.
   6. List the specific records or people needed to resolve each gap.
   7. End with a short exception summary for a quality or operations colleague.
   
   Do not estimate quantities, dates or destinations. Quote the relevant source reference for every conclusion. If the records do not support an answer, write "not established from the supplied records".

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