> For the complete documentation index, see [llms.txt](https://docs.gaiodataos.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.gaiodataos.com/settings/ai.md).

# AI

The **AI** section in Settings lets administrators manage AI providers, define data-sharing policies and control how contextual information is sent to AI models during analyses, conversations and agent runs.

<figure><img src="https://1671598980-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0muDd0LkZG6CmTQvGQ4D%2Fuploads%2FVYyQVsGtceNqVrRI9eXu%2Fimage.png?alt=media&amp;token=abf417fc-7a8e-4fd6-a572-d1190bdbf6be" alt=""><figcaption></figcaption></figure>

## Data Sharing Permissions

At the top of the page, you can configure which types of data can be shared with AI providers.

These settings are **global** and define the privacy and governance boundaries for AI use within Gaio DataOS.

***

### 1. Metadata

Shares only basic structural information about the tables.

#### Includes:

* Column name
* Column title
* Data type

#### Excludes:

* Sensitive values

#### Purpose:

Lets the AI understand the schema structure without accessing real data.

***

### 2. Table Sample Data

Shares a limited subset of the table data.

#### Includes:

* First 10 rows
* Sensitive data (if enabled)

#### Purpose:

Improves the AI's contextual understanding for more accurate answers in data analysis and interpretation.

***

### 3. Analytics

Shares analytical information derived from the tables.

#### Includes:

* Frequency distributions
* Metrics
* Analytical summaries
* Sensitive data (if enabled)

#### Purpose:

Lets the AI reason about patterns, trends and aggregations, beyond the raw data.

***

### 4. Results Sample

Shares a sample of the results of executed queries.

#### Includes:

* First 10 query results
* Sensitive data (if enabled)

#### Purpose:

Provides execution context for follow-up questions and more complex analyses.

***

## Creating or Editing an AI Configuration

1. Go to **Settings → AI**.
2. Click **New** or the pencil of an existing configuration. The **Ai Manager** window opens.
3. Fill in the fields below.
4. Click **Save**. The trash icon deletes the configuration.

### Fields

* **Name**: internal identifier of the AI configuration.
* **Use type**: **General** or **RAG**.
* **Supplier**: DeepSeek, OpenAI, Anthropic, Grok, Mistral or Azure.
* **Model**: the available models depend on the supplier. For OpenAI, for example: GPT-5, GPT-5 Nano, GPT-5 Mini, GPT-4.1, GPT-4.1 Mini, GPT-4.1 Nano, GPT-4o, GPT-4o Mini, GPT-4o Audio Preview, GPT-4 Turbo, GPT-4, O3, O3 Mini and O4 Mini.
* **API Key**: key provided by the supplier. It is stored securely and used only at runtime.

***

## Preferred Model

You can mark a configuration as **Preferential**.

The preferred model is used by default in:

* Dashboards
* Agents
* AI-powered features

Other models remain available for manual selection or advanced cases.

***

## Configuration list

The AI configuration list shows the columns **ID**, **Name**, **Supplier**, **Model**, **Preferential**, **Define** and **Provider**.

* **Preferential**: selector that marks the preferred model.
* **Define**: button **Apply in all apps**.
* **Provider**: **Gaio** or **Custom**.
* The pencil at the end of each row opens the configuration for editing.

***

## Governance and Security

* Sharing of sensitive data is explicitly controlled by scope.
* API keys are never exposed to end users.
* AI providers only receive the data you enable.
* You can use multiple providers in the same environment (e.g. OpenAI + Anthropic).

***

## Best Practices

* Start by enabling only **Metadata**, and expand as needed.
* Use **Table Sample Data** for exploration and schema understanding.
* Enable **Analytics** only when deeper statistical reasoning is needed.
* Set a **Preferential** model to keep behavior consistent.
* Use different providers to compare quality, cost and performance.

{% hint style="info" %}
This configuration layer ensures Gaio's AI features are powerful, transparent and governance-compliant, while keeping full control over how intelligence is applied to data.
{% endhint %}
