YES

As of 13 August 2026, AI can draft answers to common customer questions.

This still needs a person who signs their name to it.

Can you do it?

5 minutesto a draft.

30 minutesto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ita colleague

What the alternative costsA purpose-built support product such as DocsBot can create a chatbot from your documentation and embed it for customers.

If this goes wrong, customers receive a confident answer that is out of date or unsuitable for their situation, and your team has to correct it and repair the resulting confusion.

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 help pages, product documentation, price list, delivery information, refund policy and escalation rules that govern customer replies.
    2. Collect the common customer questions from support tickets, email, chat or your existing FAQ, removing duplicate wording while keeping materially different situations.
    3. Paste the questions and the approved source material into the prompt, then set the intended answer length and tone in the bracketed fields.
    4. Ask the chatbot to produce the table and the list of claims needing live-system or policy checks.
    5. Compare every draft answer with the current source document it cites, checking product names, prices, dates, eligibility, steps and promised outcomes.
    6. Send answers marked for escalation, and any answer involving an exception or missing information, to a support colleague for a decision.
    7. Publish the approved answers in the help centre or response library, and keep the source documents and approval record with them.

    Prompt

    You are drafting customer-service answers for [BUSINESS NAME]. Use only the information in the source material below. Do not invent features, prices, delivery times, guarantees, legal rights, technical steps or exceptions. If the sources do not answer a question, write "Needs human confirmation" and state what information is missing.

Write one answer for each question in the list. Keep each answer to [LENGTH] in plain British English. Answer the question directly in the first sentence. Use the approved tone: [TONE]. Include a clear next step where useful. Do not make promises on behalf of the business. Do not request sensitive personal information in the public answer. Mark any question involving a complaint, refund exception, account access, safety issue, personal data, legal rights or an unusual customer situation as "Escalate to a person".

Return a table with these columns: Customer question, Draft answer, Source used, Escalate to a person. After the table, list every claim that needs checking against a live system or current policy.

Customer questions:
[PASTE QUESTIONS HERE]

Approved source material:
[PASTE CURRENT HELP PAGES, PRODUCT INFORMATION, POLICIES, PRICE LISTS AND ESCALATION RULES HERE]

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

What it gets wrong

Even on a YES, the friction has a name: context depth, judgement under ambiguity and legal accountability.

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
Inputs2
Verification2
Liability1
Effort delta2
Total9 / 10

The methodology and its thresholds are published in full.

FAQ

Can ChatGPT write customer service answers?
Yes. Give it the questions, current approved information and escalation rules, then check every draft against those sources before a colleague approves it.
Can AI answer customer questions without a human?
It can handle straightforward questions when the answer is covered by current, approved content. A person still needs to handle exceptions, complaints, personal data, account-specific issues and anything the sources do not settle.
How do I make AI customer service answers accurate?
Give the model current source documents and instruct it not to invent information or fill gaps. Compare each answer with the relevant policy, product details and live systems, then have a colleague approve answers before publishing.
What customer questions should AI not answer?
Do not let it decide complaints, unusual refunds, account-access cases, safety issues, personal-data requests or questions involving unclear rights or exceptions. Route those conversations to a trained person with access to the relevant records and authority to decide.

Nearby answers

All Help docs & FAQs rulings

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.