# AI Training

**AI Training** is the process of teaching the AI model how to respond accurately and efficiently to user queries by feeding it different types of data. This training ensures that the AI understands both the content it needs to provide and how to interact with users.&#x20;

<mark style="color:blue;">Home page > WebAssist > AI Training > Active Data</mark>

### Active Data

Active Data is often updated regularly and directly influences how the AI responds to user queries. It allows you to craft specific responses to common questions, making the AI's interactions more accurate and tailored. Active data is organized and formatted for easy searching, processing, and analysis via APIs. It's stored in databases or business applications like CRM, ERP, and HelpDesk.&#x20;

This section tackles the AI training for the previously mentioned List Menu, which appears on the right side in the chat widget.

There are multiple ways to train Active Data:

#### Integration

Integrate with your existing platforms by clicking on Integration in the menu on the left side, or use our data models for creating and managing your details. Contact us for specific integrations or data structures.&#x20;

Within Integrations you will see all the available integrated apps such as WhatsApp, Shopify and many more.

#### Use CleanDesk AI’s DB

This allows you to utilise CleanDesk’s resources and database, to create Products, Campaigns, Services and so on.

Let us take an example, say, 'Products'.

To train a new product, click the **New Product** button and fill in the required fields, such as:

\- Product Name: Enter the name of your product.

\- Product Category: Select the appropriate category for your product.

\- Product Description: Provide a clear and concise description of the product, detailing its features and benefits.

\- Price Range: Specify the price range for the product.

\- Discount: Mention any applicable discounts to attract customers.

\- Product Image: Upload an image that best represents the product.

Additionally, you can add a detailed description to provide more in-depth information and set a call-to-action (CTA) button (such as “Buy Now”) linked to a specific URL or webpage. Once all the details are filled in, click **Submit** to save your changes.

View the following video for better understanding:

{% embed url="<https://youtu.be/Hbt56uppnaI>" %}
How to train AI with Active Data - create a new Product
{% endembed %}

Similar to the above example of 'Products', you can also add various other data sets to enhance your platform such as:

* Campaigns: Promote your offers or schemes. Click the **Create New Campaign** button, then fill in details such as the Campaign Title, Applicable Dates, and upload a relevant image. You can also provide a detailed description and set a CTA button linked to a specific URL. Click **Publish The Content** to save the campaign. Multiple campaigns can be created and managed easily.

This video shows how to add a Campaign:

{% embed url="<https://youtu.be/C2caQJcfbs4>" %}
How to train AI with Active Data - train a new Campaign
{% endembed %}

* Services: Provide detailed descriptions of the services your company offers. Use the same format as products to add services.
* Teams: Share information about your team members, including their roles, expertise, and profiles.
* Portfolio: Highlight key projects or company achievements by adding a portfolio section. Include descriptions, images, and any relevant links to showcase your work.
* Testimonials: Display customer feedback and reviews to build trust with potential clients. Add customer testimonials with names, feedback, and images if available.
* Contact: Provide essential contact information, including phone numbers, email, physical address, or any other relevant ways for customers to reach out to your company.

Each of these sections can be customised with images, descriptions, and action buttons, providing a comprehensive overview of your business and offerings.&#x20;

#### Generate with AI

You can also add products using the Generate with AI feature by clicking the **Generate with AI** button. This option offers three convenient methods to choose from:

1\. Upload Document: Upload a document that contains detailed information about the product.

2\. Fetch from URL: Provide a URL link to a webpage that has all the relevant product details.

3\. Train Custom Data: Enter a set of custom questions and answers to train the AI on the specific product information.<br>

After selecting one of these methods, click **Generate** to generate and save the details.

\
The following video shows how to add multiple team members under Teams, using the Generate with AI option:

{% embed url="<https://youtu.be/Pwm0VRnMQAY>" %}
How to train AI with Active Data - train multiple team members using Generate with AI
{% endembed %}

### Passive Data

<mark style="color:blue;">Home page > WebAssist > AI Training > Passive Data</mark>

Passive data refers to the Assistant’s abilities gained by training it with documents, URLs, or text. By assembling a diverse collection of documents and utilizing CleanDesk AI’s pre-built index displayed in the left menu, you'll enable your assistant to understand nuanced information and provide precise, contextually accurate responses.&#x20;

As an example, let’s look at the Home page under Passive Data. This is where you will train the AI for the main landing page of your web site.

First click on the **Train** button, next use any of these 4 available options for AI training:

\
1\. Upload Document: Upload a document that contains detailed information about the data you wish to train.

This video shows how to train Passive Data by uploading a document, while training a Package (for instance):

{% embed url="<https://youtu.be/oRsapELX8_0>" %}
How to train AI with Passive Data - train a Package using a Document Upload
{% endembed %}

2\. Fetch from URL: Provide a URL link to a webpage that has all the relevant details of the data you wish to train.

Here is an example of training Passive Data using a URL:

{% embed url="<https://youtu.be/UlCoOOE77tg>" %}
How to train AI with Passive Data - train a Home page using a URL or web site
{% endembed %}

3\. Create Custom Text: Provide the test you would like to use, to train the AI.

The following video explains how to generate Passive Data by creating custom text:

{% embed url="<https://youtu.be/9MAxH5OSyN8>" %}
How to train AI with Passive Data - train Service related information using custom text
{% endembed %}

4\. Create Q\&A: Enter a set of custom questions and answers to train the AI on the specific product information.

Here is how you can generate Passive Data by using Q\&A, for AI training:

Click on **Upload** to proceed, and then click on the **Published** toggle button. To verify that your document has been published, click on Sample Queries under Action on the right hand side. In case the Sample Queries are not available, try clicking on the **Published** toggle button (off and then on again) once more. Check Sample Queries again, to verify.

\[Sample Queries are the potential intentions that users might be inquiring about, and the generated responses are based on this training.]

To edit any document, for that particular document, first click on the **Published** toggle button to unpublished. Changes can only be made after a document is unpublished. Make the necessary changes in the document, then upload and publish it as mentioned above.

To delete any document, for that particular document, first click on the **Published** toggle button to unpublished. A document can be deleted only after it is unpublished.

\
\
\ <br>


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.cleandesk.ai/ai-training.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
