# Usage in Athenic AI

#### AI Analyst (Chat)

Athenic AI can answer questions in two modes: **Metrics** (recommended) or **SQL**. Users can choose the mode from the dropdown next to the chat input.

<div data-with-frame="true"><figure><img src="https://2503620627-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FVqTQOKxSAPs7Xr7cb4Ms%2Fuploads%2FPAoX5hVgxUQccYdEm7ZM%2Fchat%20mode.png?alt=media&#x26;token=ebd90f43-029e-45cd-8209-affd25e673de" alt=""><figcaption></figcaption></figure></div>

The mode selection is on a per chat basis. A chat is locked to the mode you used for the first question:

* After you send your first question, the **SQL / Metrics dropdown is disabled** for that chat.
* To switch modes, start a **new chat** and choose the other option.

### **Metrics (default)**

Metrics mode uses the Governed Metrics configured by your team admin so results are computed using your defined semantic models, metrics, and dimensions.

When you ask a question, Athenic shows an **AI Interpretation** of the request as structured query parts, including:

* **Metrics**
* **Filters**
* **Group By**
* **Time Grain**
* **Date Range**
* **Order By** and **Limit** (when relevant)

<div data-with-frame="true"><figure><img src="https://2503620627-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FVqTQOKxSAPs7Xr7cb4Ms%2Fuploads%2FawVNv7UBWOEjkiHODnb0%2Fmetric%20proposal.png?alt=media&#x26;token=943dbd28-23db-403c-9c7a-bbee78405ecc" alt=""><figcaption><p>The image shows a proposed metric generated by Athenic based on the user's question.</p></figcaption></figure></div>

You can then:

* Click **Edit** to adjust the proposed metrics, filters, grouping, or time settings
* Click **Confirm** to run the query and **View Results**

This mode is best when you want consistent, governed definitions (e.g., revenue, units, active users) and reusable analysis patterns. For non-technical users, this is how they can ask questions and be confident they are getting what they asked for.

**AI Generated Metric / Advanced Dimension**

If Athenic AI can’t find an existing metric or dimension that matches your question, you can ask it to create a new one directly from the **Edit** panel.

**How it works**

1. Open **Edit** on the AI Interpretation.
2. Click on the AI Icon next to the Cancel button.
3. Toggle between **Metric** or **Dimension** depending on what you are trying to add.
4. Describe what you want in plain language (e.g., a metric for “Gross margin % = (revenue - cost) / revenue” or a dimension for “Customer region = US".
5. Athenic generates a **draft** metric or advanced dimension and will show you the description and generated logic.&#x20;
6. Once you save the draft, you can add it to your query via the editor.

<div data-with-frame="true"><figure><img src="https://2503620627-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FVqTQOKxSAPs7Xr7cb4Ms%2Fuploads%2F5q4vZdt1yPoUT2C7LmFN%2Fgenerate%20advanced%20dimension.png?alt=media&#x26;token=96b2df2e-53fc-4aeb-aa6f-ebd53e7a3118" alt=""><figcaption><p>Athenic will scan your semantic models to generate metrics or advanced dimensions based on user input. </p></figcaption></figure></div>

**Quarantined (Draft) Items**

* AI-generated metrics/dimensions appear **in orange** and are labeled **Quarantined**.
* Quarantined items require **admin approval** before they become part of your shared semantic layer.

<div data-with-frame="true"><figure><img src="https://2503620627-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FVqTQOKxSAPs7Xr7cb4Ms%2Fuploads%2F020GRPXd7jADsNqJtEWy%2Fquarantined%20advanced%20dimension.png?alt=media&#x26;token=9a12067c-7c63-42e7-9aa0-229830742aa5" alt=""><figcaption></figcaption></figure></div>

**After Approval**&#x20;

Once an admin approves the draft, it becomes a normal governed metric/dimension and will appear and behave like any other metric or dimension throughout Athenic.

### **SQL**

SQL mode generates SQL using your knowledge graph setup. It does not include the structured interpretation blocks (metrics/filters/group by) and you work directly in SQL.

Use SQL mode when you need full control over the query or exploratory custom logic that are not covered by Governed Metrics.