# AI Coworkers

An AI Coworker is a configurable AI colleague that operates across channels, chooses the right skill for each request at run time, and remembers context across sessions through long-term memory. Unlike an agent you build in a single modality, a Coworker is assembled from skills — including agents you have already built — and acts autonomously within the boundaries you define.

<details class="watch">
<summary><span class="watch-thumb"><img src="https://i.ytimg.com/vi/Yb0arnh_iVw/mqdefault.jpg" alt=""><span class="video-length">25 min</span></span><span class="watch-text"><span class="watch-label">Watch the series: Neo: Coworker in depth</span><span class="watch-meta">9 episodes · about 25 min</span></span></summary>
<div class="video"><iframe src="https://www.youtube-nocookie.com/embed/videoseries?list=PLdwdGIspgPs0&rel=0&autoplay=1" title="Neo: Coworker in depth, all episodes" allow="accelerometer; clipboard-write; encrypted-media; gyroscope; picture-in-picture" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen loading="lazy"></iframe></div>
<p class="video-caption">Use the playlist menu in the player to jump to an episode. <a href="../videos/coworkers.html">Episode list</a>.</p>
</details>

You create one by describing who it is, where it works, what it can do, what it remembers, and how it should behave. OneByZero Neo turns that into a working agent you can publish to your channels and monitor.

**Use an AI Coworker when:**

- You want an always-available assistant on several channels at once — NeoGPT, Microsoft Teams, email, or phone.
- It should decide which capability to use for each request, instead of following a fixed script.
- It should handle many kinds of task by drawing on different skills, including agents you have already built.
- It should remember context between conversations, or check in on a schedule.
- Sensitive actions should wait for a person to approve them.

New to Coworkers? [Build Your First Coworker](../3_getting_started/for_coworkers.html) walks you through creating one in about ten minutes.

Building and running a Coworker falls into a few stages, each covered in its own group of pages:

| Stage | What it covers |
| --- | --- |
| [Getting Started](getting_started.html) | Roles and permissions, the Coworker list, and the creation wizard. |
| [Configure](configure.html) | The seven builder tabs that define who the Coworker is, what it can do, and how it is governed. |
| [Operate](operate.html) | Publish it, see how it runs, then monitor and evaluate it once it is live. |
| [Build a Coworker](worked_example.html) | A complete, worked example — an IT Helpdesk assistant — that ties every tab together. |

> **Note**
>
> A Coworker builds on other parts of the platform: its skills, guardrails, and evaluations come from the [Foundry](../12_foundry/index.html); it runs on the [Channels](../13_channels/index.html) you set up; and you monitor it much like any agent in the [Command Center](../11_command_center/index.html).