> For the complete documentation index, see [llms.txt](https://docs.mercuri.cx/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.mercuri.cx/features/ai-assistants/train-your-ai-assistant.md).

# Train Your AI Assistant

####

Before the assistant can respond accurately, Mercuri needs to index and process all the connected knowledge sources.

To start training:

* Click on **Start Training**
* Mercuri will process:
  * Website content
  * Sitemap pages
  * Added URLs
  * Connected knowledge sources

The training process usually takes **1–5 minutes** depending on the amount of content added.

<figure><img src="/files/P2Y4QwCNFQxQin02DQyw" alt=""><figcaption></figcaption></figure>

#### Deployment Modes

Once training starts, Mercuri provides different deployment modes for managing the assistant.

{% tabs %}
{% tab title="Testing Mode" %}
In Testing mode:

* Only added test numbers can interact with the assistant
* You can safely validate responses and workflows
* Customers will not have access yet

This mode is recommended before going live.
{% endtab %}
{% endtabs %}

{% tabs %}
{% tab title="Live Mode" %}
After successful testing, you can switch the assistant to **Live** mode.

In Live mode:

* Real customers can interact with the AI assistant
* The assistant becomes active on connected channels like WhatsApp or SMS
* Automations and AI replies start functioning publicly

Make sure testing is completed before enabling Live mode.
{% endtab %}

{% tab title="Back to Draft" %}
If you need to make changes later, you can move the assistant back to **Draft** mode.

This allows you to:

* Edit prompts or workflows
* Update knowledge sources
* Modify configurations
* Retrain the assistant after changes

Mercuri makes it easy to continuously improve the assistant even after deployment.
{% endtab %}
{% endtabs %}