> ## Documentation Index
> Fetch the complete documentation index at: https://docs.atako.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Creating and managing agents

> Every field in the agent creation form, plus lifecycle, slots, and permissions.

An agent is a persistent, always-on AI worker — not a one-off chat session. Creating one takes a minute; the choices you make below shape how it behaves for its entire lifetime.

## Who can create agents

Company admins can always create agents. Other members need the **Allow creating agents** permission, granted per member by an admin from **Settings → Users**. Trying to create an agent without it shows: *"You don't have permission to create agents. Ask your admin to enable it for you in settings."*

<Note>
  If your company has no active subscription, or has reached its active agent limit, agent creation is blocked with a message pointing you to the [subscription page](/guides/subscription) instead.
</Note>

## The creation form

<Steps>
  <Step title="Agent name">
    A short, descriptive name — shown everywhere the agent appears, including to the people who chat with it.
  </Step>

  <Step title="Image">
    Pick from a set of built-in avatars. Click the image to shuffle to another one at random.
  </Step>

  <Step title="Instructions (optional)">
    A system prompt describing the agent's role, tone, and how it should behave — sent to the agent on startup.

    <Tip>
      Be concrete about scope: what the agent owns, what it should never do without asking, and how it should sound. Vague instructions ("be helpful") do less than a short list of do's and don'ts.
    </Tip>
  </Step>

  <Step title="My business context (optional)">
    Free text describing your business, the tools you use, and your working style. The agent works without it, but context meaningfully improves results — the more specific, the better.

    This is different from your company's shared context (set once under **Settings → Context**), which every agent receives automatically. Use the agent's own context field for anything specific to that agent's role, not facts every agent should already know.
  </Step>

  <Step title="AI model">
    Choose a model from a curated list, or leave it on the recommended default. If [BYOK](/concepts/credits#byok-agents) is enabled for your company, you can instead run the agent on your own LLM provider key — those agents never draw from your shared credit balance.
  </Step>

  <Step title="Execution engine (advanced)">
    For companies where Atako has enabled model selection, an advanced option lets you pick the agent's underlying execution engine.

    <Warning>
      This choice is final. It can't be changed after the agent is created — leave it on the default unless you have a specific reason to change it.
    </Warning>
  </Step>
</Steps>

Channels aren't part of the creation form — you turn them on afterward, from the agent's own page. Integrations already connected at company level, however, can be granted directly in the creation form; you can add more, or connect new tools, later from the agent's page. See [Integrations overview](/integrations/overview) and [Channels overview](/channels/overview).

## Lifecycle

| Status           | Meaning                                                                                                                                           |
| ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Provisioning** | The agent's environment is being set up — this only takes a moment.                                                                               |
| **Active**       | The agent is running and occupies one of your company's agent slots.                                                                              |
| **Paused**       | Compute is stopped, but everything the agent has built up — memory, cron tasks, files — is kept. Resuming restarts the same agent, not a new one. |
| **Completed**    | The agent has ended. Its slot is freed and its resources are deleted.                                                                             |

Agents are paused automatically in some cases — for example, when a free trial ends without a subscription in place — and can be resumed once the underlying issue (like billing) is resolved.

<Warning>
  Ending an agent is permanent: it can't be resumed, and its slot frees immediately. Its conversation history and files remain viewable. There is no manual pause — pausing only happens automatically (for example when a trial expires), and a paused agent can then be resumed.
</Warning>

## Agent slots

Every active agent occupies one **agent slot** from your company's shared pool. The number of slots equals what you've purchased on your subscription — a trial includes 3. Slots aren't tied to human users: your team can have any number of members regardless of how many slots you own. Ending an agent frees its slot immediately for a new one. See [Subscription](/guides/subscription) for pricing per slot.

## Related

<CardGroup cols={2}>
  <Card title="Agents" icon="cpu" href="/concepts/agents">
    The concept behind agents, slots, and sub-agents.
  </Card>

  <Card title="Chat" icon="messages" href="/guides/chat">
    Working with an agent day to day.
  </Card>

  <Card title="Team" icon="users" href="/guides/team">
    Roles, permissions, and inviting teammates.
  </Card>

  <Card title="Credits" icon="coins" href="/concepts/credits">
    How model usage and BYOK are metered.
  </Card>
</CardGroup>