# AI

> Build with Kustomer AI and build Kustomer integrations with AI. The AI and automations API, the kusty CLI and MCP server for coding agents, and docs published in formats models can read.

Source: https://developer.kustomer.com/docs/ai

There are two sides to AI on this site: APIs for working **with Kustomer AI** (summaries, AI agents and their evaluations), and tools that let **your AI assistants work with Kustomer** (the kusty MCP server and machine-readable docs).

- [AI and automations API](https://developer.kustomer.com/docs/ai/ai-api): Read AI summaries, list your AI agents, and test them with evaluations from scripts and CI.
- [kusty CLI](https://developer.kustomer.com/docs/ai/kusty): The whole Kustomer API from the command line, with JSON output and exit codes built for agents.
- [kusty MCP server](https://developer.kustomer.com/docs/ai/kusty-mcp): Give Claude Code, Cursor, Codex or Gemini CLI typed tools for every Kustomer endpoint.
- [Docs for AI](https://developer.kustomer.com/docs/ai/docs-for-ai): llms.txt, Markdown for every page and operation, and a docs MCP server to feed your models.

## Which tool for which job

| You want to                                                            | Use                                                                                                                                                                      |
| ---------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Show AI summaries of conversations in your CRM, BI tool or Slack       | [Summaries API](https://developer.kustomer.com/docs/ai/ai-api#summaries)                                                                                                 |
| Catch regressions in an AI agent before you turn on a change           | [Evaluations API](https://developer.kustomer.com/docs/ai/ai-api#evaluations), run from [CI](https://developer.kustomer.com/docs/ai/ai-api#gate-deploys-on-agent-quality) |
| Ask your coding assistant "which customers wrote in about order 4417?" | [kusty MCP server](https://developer.kustomer.com/docs/ai/kusty-mcp)                                                                                                     |
| Script one-off data fixes and exports with JSON you can pipe to `jq`   | [kusty CLI](https://developer.kustomer.com/docs/ai/kusty)                                                                                                                |
| Have an assistant write integration code against Kustomer              | [Docs for AI](https://developer.kustomer.com/docs/ai/docs-for-ai): point it at `llms.txt` or a page's `.md`                                                              |

## Building AI features on Kustomer events

The rest of the platform works well with your own models too:

- [Webhooks](https://developer.kustomer.com/docs/events/outbound-webhooks) stream conversation and message events to your service as they happen.
- [Workflows](https://developer.kustomer.com/docs/workflows) can call your AI service with a REST action and write the answer back to a conversation.
- The [server-side chat API](https://developer.kustomer.com/docs/chat/server-chat-api) lets your own bot answer chats, with an AI typing indicator.
- [Apps](https://developer.kustomer.com/docs/apps) package all of this so other orgs can install it.

> **For AI agents reading this:** Start from [`/llms.txt`](https://developer.kustomer.com/llms.txt). Every page is available as Markdown at its URL plus `.md`, and every API operation has its own Markdown page under `/api/<spec>/<operation>.md`.