FBA Reimbursement Reconciliation with Claude or ChatGPT and Agent Central for Amazon Sellers

What they did

Agent Central connects Seller Central to Claude or ChatGPT and returns three FBA record sets: reimbursements (reason, quantity, amount, ASIN/SKU, order ID), inventory ledger movements (event type, reference ID, quantity, fulfillment center, timestamp) and FBA customer returns (reason, disposition). The suggested monthly run asks for reimbursements approved in the last 60 days, then the ledger movements and returns for the same ASINs and dates, and lists rows with no counterpart.

How it went

These are vendor docs with no customer result. The vendor is upfront about limits: the AI 'can line up the records; it can't file the claim.' The ledger read doesn't include Amazon's adjustment reason codes, and no shared key links a ledger movement to the reimbursement that paid it, so matches are 'candidates to confirm, not proof.' A trial only covers the last 30 days, which is shorter than some claim windows.

We only report what the source says. Results are self-reported unless noted.

Worth knowing

U.S. claim windows depend on the event type: 60 days after Amazon reports a unit lost or damaged in a fulfillment center, and 60 to 120 days after the refund for customer returns. Since 2025, losses before an order are reimbursed at manufacturing cost, not selling price, so keep a cost file per ASIN.

Copy-paste prompt

Pull my FBA reimbursements approved in the last 60 days with reason, quantity and amount. Then pull inventory ledger movements and FBA customer returns for the same ASINs and dates.
List every lost/damaged ledger movement and every refunded return that has no matching reimbursement (match on ASIN/FNSKU, quantity and date). For each, show the claim deadline and the days left. Label every match as a candidate to check in Seller Central, not a confirmed claim.

The source didn't publish its exact prompt. We wrote this starter so you can try the same workflow. Replace anything in [brackets].