---
title: AI Features
feature: true
featureGroup: ai
featureState: beta
---

## Overview

The Business Platform includes AI-powered capabilities that can be enabled or disabled on a per-project basis. Project administrators control AI feature availability through the project settings, allowing teams to opt in to AI functionality while maintaining data privacy awareness.

## Enabling AI Features

AI features are disabled by default for each project. To enable them:

1. Open your project and navigate to **Settings** in the sidebar.
2. Under the **Features** section, click the **AI Settings** tile.
3. Toggle the **AI Features** switch to the enabled position.
4. The setting is saved immediately — no additional confirmation is required.

To disable AI features, return to the same page and toggle the switch off.

> **Note:** Only users with project update permissions can change AI settings.

## Affected Features

When AI features are **disabled** for a project, the following capabilities are blocked and return an error if accessed:

| Feature | Behavior when disabled |
|---|---|
| **Documentation Chat** | Returns a 403 error with a message prompting the user to enable AI in project settings. |
| **AI Agents (Chat)** | Requests to AI agent chat endpoints are rejected with a 403 error. |
| **Smart Import (AI Extract)** | AI-powered data extraction is blocked with a 403 error. |
| **Action Summaries** | AI-generated action summaries are unavailable. |

When a user attempts to access a disabled AI feature from within the application UI, they are shown a prompt explaining that AI features are not enabled for the project, with a direct link to the AI Settings page.

## Data Privacy

Enabling AI features for a project means that certain data may be processed by third-party AI providers. The following data types can be involved:

- **Documentation Chat** — User questions and relevant documentation excerpts.
- **AI Agents** — Conversation messages and agent-configured context.
- **Action Summaries** — Action metadata and associated entity data.
- **Smart Import** — File content submitted for data extraction.

### AI Providers

The Business Platform integrates with the following external AI providers when AI features are enabled:

- **Mistral AI** — Used for selected language model tasks. See [Mistral AI Privacy Policy](https://mistral.ai/privacy-policy/).
- **OpenAI** — Used for embedding generation and language model tasks. See [OpenAI Privacy Policy](https://openai.com/privacy/).

No data is sent to AI providers while AI features are disabled for a project.

### Data Handling Considerations

- Only data relevant to the specific AI operation is transmitted to providers (e.g., the user's message and retrieved documentation context for documentation chat).
- Sensitive project data should be reviewed before enabling AI features in environments where data residency or confidentiality requirements apply.
- Consult your organization's data protection officer (DPO) if you are unsure whether enabling AI features is appropriate for your use case.

## API Behavior

All AI-related API endpoints check the `aiEnabled` flag on the project before processing requests.

| Endpoint | Description |
|---|---|
| `POST /api/projects/[projectId]/ai-agents/[agentId]/chat` | AI agent chat |
| `POST /api/projects/[projectId]/types/[typeId]/ai-extract` | Smart import / AI extraction |
| `POST /api/help/chat` | Documentation chat |

If `aiEnabled` is `false`, these endpoints return:

```
HTTP 403 Forbidden
{
  "error": "AI features are disabled for this project. Please enable AI features in project settings."
}
```

The AI settings themselves are managed via the following endpoint:

```
PATCH /api/projects/{projectId}/ai-settings
Content-Type: application/json

{ "aiEnabled": true }
```

This endpoint requires the authenticated user to have update permission on the project.