# Tutorials

> Short videos that teach you to build agents on Mindset and run them in production, in playlists you can watch in order.

## Build your first agent

Start here. Connect Claude, build your first agent, hook it up to a real system and switch it on.

1. How Mindset fits together: A quick tour of how Mindset works. You build agents from your own AI agent, like Claude Code, Codex or Copilot, and they run in Mindset for your people and systems.
2. Set up Claude Code with Mindset: Connect Claude to your Mindset org in a couple of minutes, then check it's working.
3. Your first agent: Describe the agent you want. Claude builds it, then you check it and switch it on.
4. Connect a system: Hook your agent up to a real system, like a course catalogue, without ever giving Claude your keys.
5. See it, check it, activate it: See everything Claude has built and decide when a new version goes live, or go back to an earlier one in one click.

## Ship it like software

Run your agents like real software: keep versions, test them, go to Production and put them in your product.

1. Keep your agents in your repo: Keep a copy of every agent in your own code repository, so changes get reviewed like any other code.
2. Test before it goes live: Check your agent behaves the way you expect before anyone else meets it.
3. Move it to Production: Move a finished agent from your safe Sandbox into Production, where real people use it. (coming soon)
4. Start work without a chat: Have an agent start work on its own: on a schedule, from your app, or when something happens in another tool. (coming soon)
5. Put an agent on your page: Put your agent inside your own website or product, styled to fit right in. (coming soon)

## Make it do more

Go further with step-by-step flows, approvals, charts, messages and a clear view of what happened.

1. Make it follow steps: Give an agent steps it must follow in order, with checks it can't skip. (coming soon)
2. Ask a person partway through: Pause a task until a person says yes, then carry on where it left off. (coming soon)
3. Functions: Work through a whole list of data, doing exactly the same thing for every row. (coming soon)
4. Show things: Show charts, tables and forms inside the conversation. (coming soon)
5. Post to Slack or send email: Let your agent post in Slack or send an email, with a person's approval where you need it. (coming soon)
6. See what happened: See what your agents did, what it cost, and why one went wrong, then fix it with Claude. (coming soon)
7. Bring an agent from another framework: Bring an agent, skill or spec you already have in another tool over to Mindset. (coming soon)
8. Switch it off: Stop one action, or a whole connection, the moment something looks wrong. Claude can switch it off for you, but only a person switches it back on. (coming soon)
9. Who did what: See who changed what, and when, in a record nobody can edit. Then keep learners' personal details away from the AI provider. (coming soon)

## For builders

Everything a builder needs to take one agent from nothing to live, in order.

1. How Mindset fits together: A quick tour of how Mindset works. You build agents from your own AI agent, like Claude Code, Codex or Copilot, and they run in Mindset for your people and systems.
2. Set up Claude Code with Mindset: Connect Claude to your Mindset org in a couple of minutes, then check it's working.
3. Your first agent: Describe the agent you want. Claude builds it, then you check it and switch it on.
4. See it, check it, activate it: See everything Claude has built and decide when a new version goes live, or go back to an earlier one in one click.
5. Connect a system: Hook your agent up to a real system, like a course catalogue, without ever giving Claude your keys.
6. Test before it goes live: Check your agent behaves the way you expect before anyone else meets it.
7. Move it to Production: Move a finished agent from your safe Sandbox into Production, where real people use it. (coming soon)
8. Put an agent on your page: Put your agent inside your own website or product, styled to fit right in. (coming soon)
9. See what happened: See what your agents did, what it cost, and why one went wrong, then fix it with Claude. (coming soon)

## For operators

Keep agents running well once they're live: check them, change them safely, and stay in control.

1. How Mindset fits together: A quick tour of how Mindset works. You build agents from your own AI agent, like Claude Code, Codex or Copilot, and they run in Mindset for your people and systems.
2. See it, check it, activate it: See everything Claude has built and decide when a new version goes live, or go back to an earlier one in one click.
3. Test before it goes live: Check your agent behaves the way you expect before anyone else meets it.
4. Move it to Production: Move a finished agent from your safe Sandbox into Production, where real people use it. (coming soon)
5. See what happened: See what your agents did, what it cost, and why one went wrong, then fix it with Claude. (coming soon)
6. Switch it off: Stop one action, or a whole connection, the moment something looks wrong. Claude can switch it off for you, but only a person switches it back on. (coming soon)
7. Who did what: See who changed what, and when, in a record nobody can edit. Then keep learners' personal details away from the AI provider. (coming soon)
