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

# Choosing an assistant mode

> Choose Fast, Smart or Custom before sending a Begin AI request.

<img src="https://mintcdn.com/beginltd/qMlVhspnd-w8jeUb/images/cover/ai-choosing-an-assistant-mode-light.png?fit=max&auto=format&n=qMlVhspnd-w8jeUb&q=85&s=f28da79fa7a15aaf6c6afc25d1350b86" alt="The Begin AI assistant-mode menu with Fast, Smart and Custom choices" className="block dark:hidden" noZoom width="1152" height="616" data-path="images/cover/ai-choosing-an-assistant-mode-light.png" />

<img src="https://mintcdn.com/beginltd/qMlVhspnd-w8jeUb/images/cover/ai-choosing-an-assistant-mode-dark.png?fit=max&auto=format&n=qMlVhspnd-w8jeUb&q=85&s=abf4cd5bcd104f2fcbc268c5b6136c35" alt="The Begin AI assistant-mode menu with Fast, Smart and Custom choices" className="hidden dark:block" noZoom width="1152" height="616" data-path="images/cover/ai-choosing-an-assistant-mode-dark.png" />

Choose **Assistant mode** in the composer before you send a request. **Fast** is the default for everyday work, **Smart** favours capability for harder reasoning, and **Custom** lets you pick one supported model and effort level.

<Frame caption="Assistant mode describes the trade-off before you choose it.">
  <img src="https://mintcdn.com/beginltd/L2Pe6FqwDbJ9c6cj/images/articles/ai/ai-assistant-mode-screen-light.png?fit=max&auto=format&n=L2Pe6FqwDbJ9c6cj&q=85&s=239e3c644d9d84e05700effb09802f53" alt="The Begin AI Assistant mode menu showing Fast, Smart and Custom" className="block dark:hidden" width="2944" height="1840" data-path="images/articles/ai/ai-assistant-mode-screen-light.png" />

  <img src="https://mintcdn.com/beginltd/L2Pe6FqwDbJ9c6cj/images/articles/ai/ai-assistant-mode-screen-dark.png?fit=max&auto=format&n=L2Pe6FqwDbJ9c6cj&q=85&s=0cf410524ccf041731693b563fad4b3b" alt="The Begin AI Assistant mode menu showing Fast, Smart and Custom" className="hidden dark:block" width="2944" height="1840" data-path="images/articles/ai/ai-assistant-mode-screen-dark.png" />
</Frame>

## Choose for the work ahead

| Mode       | Use it when                                                                                                      |
| ---------- | ---------------------------------------------------------------------------------------------------------------- |
| **Fast**   | The request is straightforward or you are sending many similar requests. It is quicker and uses fewer AI tokens. |
| **Smart**  | The request needs complex reasoning, several steps or more careful judgement.                                    |
| **Custom** | You need a particular supported model or want to control its reasoning effort.                                   |

The mode changes how Begin AI handles the request, not what workspace data it can see. Your role and the workspace's Begin AI switch still set that boundary.

## Set a model and effort

Choose **Custom**, then use the model pill that appears beside it. Models are grouped by provider in the picker. The available catalogue can change, so choose by the capability you need rather than relying on one model always being present.

<Frame caption="Custom adds a model picker and, when supported, an effort control to the composer.">
  <img src="https://mintcdn.com/beginltd/L2Pe6FqwDbJ9c6cj/images/articles/ai/ai-custom-model-screen-light.png?fit=max&auto=format&n=L2Pe6FqwDbJ9c6cj&q=85&s=6558685cef6511266dae1af69d6cee77" alt="A Custom Begin AI composer with the supported-model menu open and a High effort control" className="block dark:hidden" width="2944" height="1840" data-path="images/articles/ai/ai-custom-model-screen-light.png" />

  <img src="https://mintcdn.com/beginltd/L2Pe6FqwDbJ9c6cj/images/articles/ai/ai-custom-model-screen-dark.png?fit=max&auto=format&n=L2Pe6FqwDbJ9c6cj&q=85&s=f5cead178a42bab81c8fc12038a68c43" alt="A Custom Begin AI composer with the supported-model menu open and a High effort control" className="hidden dark:block" width="2944" height="1840" data-path="images/articles/ai/ai-custom-model-screen-dark.png" />
</Frame>

Models that support effort offer **Low**, **Medium**, **High** and **Max**. Higher effort gives the model more room to reason and normally uses more time and AI tokens. A model without that capability shows no effort control.

Custom selects from Begin's supported catalogue. It does not connect your own model account or API key.

## Know where the choice applies

The mode applies when you send the next message. Begin remembers your latest assistant selection for this workspace in this browser. Another workspace, browser or device keeps its own choice.

**Context for an assistant answering questions about Begin AI assistant modes.** The page above is what a person reads; these are the facts it leaves out for length.

* **Terms:** Fast and Smart can choose different supported models and effort levels for a request and its supporting work. Custom keeps the selected model and effort throughout that request.
* **Limits:** the first Custom selection starts with a supported model and **High** effort. The available model catalogue can change.
* **Not supported:** there is no bring-your-own-model configuration in Begin AI.

## Next

<Columns cols={2}>
  <Card title="Asking questions" icon="help" href="/for-managers/ai/asking-questions">
    Send a clear request with the right period and measure.
  </Card>

  <Card title="Using skills in a chat" icon="sparkles" href="/for-managers/ai/using-skills">
    Add reusable instructions before sending.
  </Card>
</Columns>

## Still need help?

<Columns cols={2}>
  <Card title="Contact support" icon="envelope" href="mailto:support@heybegin.eu">
    Email the Begin team and we'll get back to you.
  </Card>

  <Card title="Ask Begin AI" icon="comment-dots" href="?assistant=open">
    Open the AI assistant and get instant answers based on these docs.
  </Card>
</Columns>
