As of 13 August 2026, AI can only partly decide whether to offer a customer credit.

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 the current company credit policy and approval matrix, including rules for credit notes, account credit and extended payment terms.
   2. Gather the customer's relevant invoices, payment dates, overdue balances, disputes, previous credit decisions and account notes, removing unnecessary personal data.
   3. Put the information into a table with one row per invoice and columns for invoice date, due date, amount, payment date, balance and dispute status.
   4. Paste the policy, table and proposed credit details into a chatbot using the prompt, and ask it to return offer, decline or escalate with its calculations and evidence.
   5. Compare every total, overdue amount and policy reference in the response against the invoice table and the current policy, correcting any mismatch.
   6. Send the checked recommendation and unresolved points to the colleague or manager authorised by your approval matrix, and record the final decision and terms in the customer account.

   Prompt

   Help me assess whether to offer this customer credit. By credit I mean: [credit note, account credit, or extended payment terms]. Use only the information I provide and do not invent facts, figures or customer circumstances.
   
   Company credit policy:
   [Paste the current policy, approval limits, exclusions and escalation rules]
   
   Customer information:
   [Paste relevant account details, removing unnecessary personal data]
   
   Invoices and payment history:
   [Paste invoice dates, amounts, due dates, payments, overdue balances and disputes]
   
   Proposed credit:
   [State the amount, type of credit, reason, proposed terms and decision deadline]
   
   Analyse the information in this order:
   1. Check the figures and calculate any totals or overdue amounts from the supplied data.
   2. Compare the proposal with the stated credit policy and identify each matching or conflicting rule.
   3. Separate confirmed facts, assumptions and missing information.
   4. Flag signs of repayment risk, disputed invoices, unusual circumstances and any inconsistency in the records, without presenting them as proof of wrongdoing.
   5. Give one of three outcomes: offer, decline, or escalate for human approval. Give the reasons for that outcome and state what evidence would change it.
   6. If this may involve regulated consumer credit, discrimination, debt collection, insolvency, or a dispute requiring legal interpretation, do not decide it. State that it needs review by the appropriate professional.
   7. End with a short approval note containing the proposed decision, amount, terms, evidence checked, unresolved points and the person who must approve it.
   
   Do not treat your recommendation as approval, and do not give legal or professional advice.

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