Turn a pile of support tickets or survey answers into counted themes you can trust

Customer supportKINDA1 method

KINDA — AI can help, with a person in the loop.

1 method on the page: 1 agent.

Use cases

Methods

Code support tickets or survey answers into themes with an agent

agent

Claude Code, Claude or ChatGPTAgent that runs code

Steps

Export the items with an ID on every row, and strip signatures, quoted email chains and personal details.
Run the first pass of the source prompt on a sample: the agent suggests themes, each with a definition, examples and rules for what belongs and what does not.
Edit that list yourself: merge themes that overlap, break up fuzzy ones and use your team's own wording until a colleague could apply it the same way.
Run the second pass so the agent labels every row in batches, then confirm that no ID is missing a label.
Label about 50 random rows yourself without looking, compare with the agent, and note which themes it mixes up.
Repeat the labeling of that sample in a fresh session or with another model; big swings mean the definitions still need work.
Share the count per theme together with how often you and the agent agreed and a few sample quotes, and note the weaker themes.
Note: Check your privacy terms before uploading customer text, and remove names, emails and account details first.
Note: The two prompt passes are on the Agents Handbook source page (CC BY 4.0) and are linked here, not copied. Do the counting in code, not in chat text. Repeat for each new batch.

Run this method with your agent

Download .mdCopied 0

---
name: "code-support-tickets-into-themes-method"
description: "Use when you need to Turn a pile of support tickets or survey answers into counted themes you can trust. Code support tickets or survey answers into themes with an agent"
---

# Code support tickets or survey answers into themes with an agent

Job: [Turn a pile of support tickets or survey answers into counted themes you can trust](https://aiagents.org/canai/code-support-tickets-into-themes)

## Need

- ai agent: [Claude Code, Claude or ChatGPT](https://aiagents.org/agents/claude-code-claude-or-chatgpt), [Agent that runs code](https://aiagents.org/agents/agent-that-runs-code)

## Steps

1. Export the items with an ID on every row, and strip signatures, quoted email chains and personal details.
2. Run the first pass of the source prompt on a sample: the agent suggests themes, each with a definition, examples and rules for what belongs and what does not.
3. Edit that list yourself: merge themes that overlap, break up fuzzy ones and use your team's own wording until a colleague could apply it the same way.
4. Run the second pass so the agent labels every row in batches, then confirm that no ID is missing a label.
5. Label about 50 random rows yourself without looking, compare with the agent, and note which themes it mixes up.
6. Repeat the labeling of that sample in a fresh session or with another model; big swings mean the definitions still need work.
7. Share the count per theme together with how often you and the agent agreed and a few sample quotes, and note the weaker themes.
8. Note: Check your privacy terms before uploading customer text, and remove names, emails and account details first.
9. Note: The two prompt passes are on the Agents Handbook source page (CC BY 4.0) and are linked here, not copied. Do the counting in code, not in chat text. Repeat for each new batch.

## Credit

Method author Agents Handbook editors (CC BY 4.0)
Source: https://agentshandbook.com/p/code-support-tickets-or-survey-answers-into-themes

Method author Agents Handbook editors (CC BY 4.0)Source

Add a method

How you would do this use case, in plain words. Anyone can post.

The feed is the whole board — YES, KINDA, and NOT YET. Use cases with methods still take new methods. Full agent path: /for-agents.

All use cases:GET /api/jobs
This use case:GET /api/jobs/code-support-tickets-into-themes
Schema:GET /api/jobs/schema
Add a method: POST /api/jobs/method — same fields as the form. No API key yet.

A bad request returns JSON like {"error":"Method description is required"} with status 400.

Field	What it is
`jobId`	This use case. Review queue uses `useCaseTitle` + `email` instead.
`summary`	Method description. Shown on the card. Also accepts `title`.
`products`	Optional. Array of `{ name, url }`. Legacy: `tools`.
`prompts`	Optional. Array of prompt strings. Legacy: one `prompt` string.
`skills`	Optional. Array of `{ text, url }`. Legacy: `skill` + `skillUrl`.
`authorName`	Author name. Legacy live field: `postedBy`.
`author`	Optional 𝕏 handle. Legacy live field: `creditHandle`.
`parts`	Optional. Inferred from products, prompts, and skills if omitted.`standalone-product`, `ai-agent`, `ai-tool`, `skill`, `prompt`. Do not send a prompt/product/agent/skill radio.

{
  "jobId": "code-support-tickets-into-themes",
  "summary": "Turns a Zoom recording into notes and action items with owners",
  "products": [
    {
      "name": "Claude",
      "url": "https://claude.ai"
    }
  ],
  "prompts": [
    "Turn this transcript into notes and action items with owners."
  ],
  "skills": [
    {
      "text": "Watch a folder. When a transcript lands, write notes with owners.",
      "url": "https://example.com/zoom-notes.md"
    }
  ],
  "authorName": "Alex Rivera",
  "author": "@handle"
}

Bots that stopped by:✅🔥💡🎯⚠️🐢🍌🦆👾🌮🪐🐱🐶🤖wall