Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for user support specialists

Support Knowledge Repair Agent

Help articles answer real customer questions correctly, verified against resolved tickets.

Quality controlAutomationCustomer support
See how it worksGet access
Support Knowledge Repair Agent: what goes in, what the agent does and what you get

What it does

When a help article is outdated, support agents keep correcting it in tickets instead of fixing the article. Customers who self-serve still get the wrong steps. This agent runs weekly, finds repeated corrections in resolved tickets that are backed by verified resolutions, and picks the article section with the most corrections. It drafts a revised section and tests it against past cases: does the new text answer the resolved tickets correctly, and do answers that were already right stay right? If any answer gets worse, it revises or restores that part. When the product changelog and the ticket resolutions disagree, it asks the product owner. The knowledge-base owner reviews and publishes the revision. Edge case: an article serving two product versions gets a version note rather than losing the older steps.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueApprovedNo 1 STARTS WHEN Weekly run or product change 2 USES A TOOL Find repeated corrections in resolved tickets 3 DOES Pick the article section with the most corrections 4 USES A TOOL Draft the revised section 5 USES A TOOL Test retrieval against resolved cases before andafter 6 CHECKS THE RESULT Do answers improve without breaking correct ones? If not: revise or restore the section. Back to step 4. 7 YOU APPROVE Knowledge-base owner reviews and publishes 8 RESULT Tested article revision ready
Read the steps as a list
  1. Weekly run or product change
  2. Find repeated corrections in resolved tickets
  3. Pick the article section with the most corrections
  4. Draft the revised section
  5. Test retrieval against resolved cases before and after
  6. Do answers improve without breaking correct ones?If not: revise or restore the section. Back to step 4.
  7. Knowledge-base owner reviews and publishesThe agent waits here for your OK.
  8. Tested article revision ready

How it decides

It ranks articles by how often agents corrected them, edits the most-corrected section first, and accepts a draft only when retrieval tests improve without breaking previously correct answers.

  • Which article first: most corrections.
  • Accept a draft: only when the test set improves overall and no answer gets worse.
  • Ask the product owner: when the changelog and resolutions disagree.

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Minimum repeated corrections before an article is picked (default 5 in 30 days)
  • Size of the test set of resolved cases (default up to 50 per article)
  • Run schedule (default weekly, plus after each product release)
  • Who is asked when the changelog and tickets disagree (default the product owner)
  • Who publishes (default the knowledge-base owner)

What keeps you in control

It always asks you first

  • Publishing
  • Statements not supported by product documentation

Hard limits

  • Drafts only; never publishes.

It stops when

  • Done: draft passes tests.
  • Needs a human: documentation contradicts resolutions.

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensIn the run of September 8 the agent finds that agents corrected the 'export to CSV' article 23 times because the button moved. The draft passes 22 of 23 test cases, but an older Excel question now gets a wrong answer, so the check fails. The agent restores that paragraph and adds a version note. All 23 pass. Knowledge-base owner Iris Novak publishes it.

More agents for user support specialists

Quality control · Customer support

Support Resolution Verification Agent

Every 'solved' ticket has evidence that the customer's own symptom is fixed.

8 steps1 check3 jobs
Customer service · Customer support

Support Attachment Interpretation Agent

A clear, evidence-based next diagnostic step for every ticket with attachments.

8 steps1 check4 jobs
Customer service · Customer support

Helpdesk Translation Ambiguity Agent

Support agents understand exactly what the customer means, whatever their language.

7 steps1 check4 jobs
Customer service · Customer support

Customer Export Troubleshooting Agent

A tested procedure that exports all the customer's data correctly.

8 steps1 check2 jobs