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

# Creating an Assistant

Create a customized assistant tailored to your requirements and needs. To complete the setup, you must select at least one AI model, enter a system prompt, and provide a name.

<Tip>
  Home page > Menu icon > All Assistants > New Assistant
</Tip>

<Steps>
  <Step title="Click “New Assistant.”">
    <Note>
      **Note** If you do not see the "New Assistant" button, you are missing the required permission. Ask an administrator to enable this feature for you.
    </Note>
  </Step>

  <Step title="Choose the settings that suit you." />

  <Step title="Click “Create Assistant.”" />
</Steps>

## Image and Name

Give your assistant a name and select a profile image.

<Steps>
  <Step title="Click into the form field and enter a meaningful name. To save the entry, click outside the form field." />

  <Step title="Optionally click Select Image to upload an image. A window will open where you can choose the desired image from your computer." />
</Steps>

## System prompt

Use the system prompt to define your assistant’s behavior, tone, and style.

<Steps>
  <Step title="Click the text field to view or edit your system prompt." />

  <Step title="Your changes are saved when you click outside the text field." />

  <Step title="Click the round arrow in the upper-right corner of the tile to reset the system prompt to its original version." />

  <Step title="A window opens with three options:">
    * **Cancel:** Close the window and keep the modified system prompt.
    * **Reset:** Revert the system prompt to the original version.
    * **Best Practice:** Choose a commonly used system prompt that typically yields good results.
  </Step>
</Steps>

## AI Model

Define which AI model your assistant should use.

<Steps>
  <Step title="Click the tile to select the desired AI model." />

  <Step title="You will be redirected to a subsequent page. Here you will see an overview of all AI models available to you." />

  <Step title="Make your selection by clicking a tile." />

  <Step title="The selected AI model is now highlighted in green with a border and a checkmark." />

  <Step title="Confirm your selection by clicking “Apply and back.”" />

  <Step title="The currently active AI model is now displayed directly on the tile." />

  <Step title="To select the original AI model, click the round arrow in the upper-right corner of the tile. Confirm by clicking “Reset” or choose “Cancel.”" />
</Steps>

## Connectors

Assign additional data sources—called connectors—to your assistant.

<Steps>
  <Step title="Click the tile to select one or more connectors." />

  <Step title="You will be redirected to a subsequent page. Here you will see an overview of all connectors available to you." />

  <Step title="Make your selection by clicking a tile." />

  <Step title="Confirm your selection by clicking “Apply and back.”" />

  <Step title="The tile will now display how many connectors are active." />

  <Step title="To reset the connector selection, click the round arrow in the upper-right corner of the tile and confirm the process in the following window." />
</Steps>

## Functions & Files

Configure file uploads and advanced processing options for your assistant.

<Steps>
  <Step title="Click the tile to go to the subsequent page." />

  <Step title="Allow file uploads in chat">
    Enable the checkbox **Allow file uploads in chat** so that users can send files to the assistant within the chat. The allowed file types are automatically determined based on the selected AI model.
  </Step>

  <Step title="Upload your own files">
    Use **Select files** to provide the assistant with your own files as a knowledge base. These files are permanently available to the assistant.
  </Step>

  <Step title="Finalize your adjustments by clicking the 'Apply and back' button." />
</Steps>

### Expert Settings

<Note>
  The **Expert Settings** section is only displayed if you have enabled the
  [Expert View](../account/expert-view) in your account settings.
</Note>

Under Expert Settings, you will find optional settings for document processing and tools:

* **Document Processing** – Defines how uploaded files are processed. By default, the method is chosen **automatically**. In the Expert Settings, you can manually switch between:
  * **via OpenAI API** – Files are processed through the OpenAI API.
  * **via configured Embedding Model** – Files are processed locally using the configured embedding model.
* **Code Interpreter always available** – Makes the Code Interpreter available on every call (not only for certain file types). The assistant can use it to execute code, e.g. for calculations or data analysis.

<Warning>
  Please note that files sent directly to the AI model may violate your
  company's data protection policies. Which file types are sent directly to the
  model is configured in the [Model Configuration](../../admin/models) via the
  **Allowed MIME Types** setting.
</Warning>

## Description & Examples

Optionally, enter a brief description for your assistant and, if desired, add sample task instructions.

<Steps>
  <Step title="Click the tile to go to the subsequent page." />

  <Step title="Click the text field to enter a description of your assistant. Click outside the field to save your entry. If you leave this field empty, the description will be generated automatically." />

  <Step title="Click the 'Add example' button and enter a sample task instruction for the assistant in the text field that appears." />

  <Step title="You can provide multiple examples." />

  <Step title="Click outside the text field to save the example." />

  <Step title="Click the trash can icon to delete an example." />

  <Step title="Select the 'Apply and back' button to continue." />

  <Step title="To reset your entries, click the round arrow in the upper-right corner of the tile and confirm the process in the following window." />
</Steps>

## Data protection

Decide whether the so-called GDPR Protection should encrypt your personal data before it is passed on to the AI model. You have three options:

* **Enabled**
* **Optional**
* **Disabled**

<Steps>
  <Step title="Make your selection by clicking one of the three buttons." />

  <Step title="Select 'Apply and back' to confirm your choice." />
</Steps>