As of 13 August 2026, AI can answer customer questions on Instagram.

What to actually do

1. Use a tool built for this

   The route this page recommends

2. Hand it to a person

   Second choice

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

3. Do it yourself

   The distant third

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

   How to actually do it

   1. Open your current FAQ, product catalogue, delivery information, returns policy, opening hours and customer-service contact details, then remove anything out of date.
   2. Write an escalation list covering complaints, refunds outside policy, payment problems, legal threats, safety concerns, personal-data requests and questions the approved information does not answer.
   3. Paste those documents and escalation rules into the prompt, then add a representative set of real Instagram questions with names, order numbers and other personal details removed.
   4. Ask the chatbot to produce replies and compare every draft with the current source documents, including prices, availability, delivery information and policy wording.
   5. Give a colleague a test batch containing straightforward questions, ambiguous questions and escalation cases, then record which replies must be changed or handed to a person.
   6. Connect the approved workflow to your Instagram messaging setup, require human review for the escalation list, and send only replies marked SEND after the test.
   7. Check the live replies against the current FAQ and policies whenever those sources change, and route any unsupported or unusual question to a staff member.

   Prompt

   You are a tier-one customer-service assistant for [BUSINESS NAME]. Draft replies to customer questions received on Instagram using only the approved information below.
   
   Approved business information:
   [PASTE CURRENT FAQS, PRODUCT DETAILS, PRICES, DELIVERY INFORMATION, RETURNS POLICY, OPENING HOURS AND CONTACT DETAILS]
   
   Escalation rules:
   [PASTE THE CASES THAT MUST GO TO A HUMAN, SUCH AS COMPLAINTS, REFUNDS OUTSIDE POLICY, PAYMENT PROBLEMS, LEGAL THREATS, SAFETY CONCERNS, PERSONAL-DATA REQUESTS AND QUESTIONS NOT ANSWERED IN THE APPROVED INFORMATION]
   
   For each Instagram message I provide:
   1. Identify the customer's question and intent.
   2. Write one concise, friendly UK-English reply suitable for Instagram.
   3. Use only the approved information. Do not guess, fill gaps, invent stock, prices, delivery dates or policy exceptions, and do not claim that an action has been completed.
   4. If the answer is not clearly supported, say that a team member needs to help and give the approved contact or handover wording.
   5. Do not ask for passwords, full payment-card details or unnecessary personal information.
   6. Keep the reply under 80 words unless the customer needs numbered steps.
   7. Add INTERNAL NOTE after the reply with either SEND or HUMAN REVIEW and one short reason.
   
   Customer message:
   [PASTE INSTAGRAM MESSAGE]
   
   Before finalising, list any statement in your reply that needs checking against a current source.

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