# Prompt customization

> For the complete documentation index, see [llms.txt](/llms.txt)

> Configure how the Biel.ai assistant responds using the default prompt or a custom prompt.

Biel.ai offers two modes for controlling how the assistant responds: **Default prompt** (optimized for technical documentation) and **Custom prompt** (full control).

## Default prompt

The default prompt gives clear, step-by-step technical answers based on your documentation. It asks clarifying questions when a request is unclear and says when the documentation does not contain enough information instead of guessing.

You can extend the default prompt with additional instructions—for example, adding a signature to responses or modifying the tone. Prompt extensions are limited to `6,000` characters.

## Custom prompt

Write your own prompt for full control over the assistant's behavior. Custom prompts are limited to `8,000` characters.

:::warning
Custom prompts may disable built-in safeguards and advanced processing features.
:::

## Configure the prompt

:::important
Only **Administrator** or **Maintainer** roles can manage projects. See [Manage roles](https://docs.biel.ai/administration/roles.md).
:::

1. In the Biel.ai dashboard, select your project.
2. Click **Settings**.
3. Open **Assistant** in the left sidebar and, under **AI model and instructions** > **Define prompt**:

    ![The AI model and instructions section with the Define prompt options](./images/settings-chatbot-llm-settings-default-prompt.jpg)

   - Select **Use default prompt** and optionally add instructions in **Extend prompt**, or
   - Select **Custom prompt** and write your full instructions in **Extend prompt**.
4. Click **Save** on the AI model and instructions card.

### Examples

**Extending the default prompt:**

```
Always start responses with "Hi!"
```

**Custom prompt:**

```
You are a customer support assistant for [Product Name]. Provide clear, accurate answers related to product features, troubleshooting, and usage. Keep responses professional and focused on the product. Only respond based on the information you can extract from CONTEXT and the conversation HISTORY.
```