Skip to content

Answer tickets from your knowledge base

An assistant finds the right help article and drafts an answer inside your ticketing tool.

Recommended: InnovatorDocumentsAutomation

Best for

  • Support teams with a large help centre
  • BPO agents handling several client products
  • Teams onboarding new agents often

What you need

  • An up-to-date help centre or knowledge base
  • A ticketing tool with API access
  • A list of topics that must go to a senior agent

Starter prompt

Copy it, fill in the parts in square brackets, and paste it into the AI assistant your organisation allows.

You are a support assistant for [product]. Read the ticket below and search our knowledge base. Draft a reply using only the retrieved articles, and list the article titles you used. If no article clearly answers the question, say 'No matching article' and suggest a tag for the ticket. If the ticket mentions [sensitive topics such as payment failure, legal threat or data deletion], do not draft a reply; flag it for a senior agent. Ticket: [paste ticket].

How to use it

  1. 1.Fix the knowledge base

  2. 2.Test on old tickets

  3. 3.Connect the ticket tool

  4. 4.Close the loop

  1. 1

    Fix the knowledge base

    Remove outdated articles and merge duplicates. The assistant can only be as correct as these pages.

  2. 2

    Test on old tickets

    Run fifty resolved tickets through and compare drafts with the real answers. Note where it chose the wrong article and why.

  3. 3

    Connect the ticket tool

    Let the assistant read tickets and post drafts as internal notes, not as replies to customers.

  4. 4

    Close the loop

    Track 'No matching article' flags weekly and write new articles for the common ones.

Watch out. A confident answer from the wrong article frustrates customers. Keep drafts as internal notes that an agent reviews, and keep personal data inside your approved tools.

What you end up with

Faster, sourced first drafts inside your ticketing tool and a knowledge base that improves weekly.

Read why AI matters in customer support

Related use cases