AI Chatbot Training
The AI chatbot can answer a variety of user questions and also perform tasks on behalf of the user. 
However, the key feature of the AI chatbot is its ability to be trained. This involves providing a resource or rich text to the AI and teaching it how to use that information to respond to user queries.
You can train the chatbot to perform the following tasks:
- Send a resource card Provide answers to user questions by delivering a resource (e.g., a file, post, report, no-code app, table, chart, diagram, etc.).
- Send a formatted text Respond to user questions with richly formatted text (e.g., incorporating different fonts, images, URLs, etc.).
Depending on your goals, you can train your chatbot accordingly.
Send a resource card
The chatbot can understand user questions and respond by providing a resource, such as a report, as illustrated in the example below:
Send a formatted text
The chatbot can also respond with answers formatted as rich text, as demonstrated below:
Training AI Model: Sending Resource Card
The diagram below illustrates the process of training an AI model for sending a resource card.
This feature is valuable if you want to incorporate your resource card into the AI and train it to respond to user queries using that card.
You can add any resource card from the ESN by simply clicking on your card menu and selecting the following option:
As shown in the picture above, you can include chats, tables, and rich text from your BI reports.
You will then be directed to a screen where you need to enter the details for your AI model training.
You can either create your own questions or have the system generate them automatically for you.
It’s recommended to include at least five questions.
All questions should be distinct.
To auto-generate questions, enter your description and click the button for the system to create questions for you.
You can modify your questions and then save the form.
You must always select your language, as the system supports multiple languages, each requiring its own training.
Once you’re finished, save the form, and the system will automatically train the AI chatbot.
Training AI Model: Sending Formatted Text
The diagram below shows the process of training an AI model for sending formatted text.
You can add formatted text enhanced with URLs, images, tables, and more to the AI chatbot. This feature is perfect for creating a knowledge base, sharing corporate guidelines, or displaying different types of informational content.
Click the following menu to begin editing the AI chatbot's rich text models.
To begin entering new text for the AI model, click the following button:
You will then be directed to a screen where you need to enter the details for your AI model training.
You'll come across an editor interface allowing you to input your text.
You're free to either copy and paste content from elsewhere or make direct edits within the interface.
Furthermore, a variety of rich text editor options are available, empowering you to style and format your text according to your preferences.
You can either create your own questions or have the system generate them automatically for you.
It’s recommended to include at least five questions.
All questions should be distinct.
To auto-generate questions, enter your description and click the button for the system to create questions for you.
You can modify your questions and then save the form.
You must always select your language, as the system supports multiple languages, each requiring its own training.
You can enter your text, insert images or URLs as needed, and format it to your preference.
Once you’re finished, save the form, and the system will automatically train the AI chatbot.
You can view all of your AI model texts, as well as edit or delete them.
Testing
To test, open the AI chatbot by selecting the following menu option:
Enter your question. It should be similar to the questions you provided during the AI model training.
The AI algorithms will identify the best match for your question and present you with a resource card.
For example, the illustration below demonstrates that rich text from a BI report was added to the AI chatbot, enabling the system to respond with this rich text whenever someone asks a relevant question.
Questions Examples
Training your model with appropriate questions is essential.
Attempting to use the same question to train two different resources will be ineffective, as the system will detect this and prompt you to rewrite the question.
To create the right question set, consider what queries the end user might have, and then test those questions.
For instance, if the goal is to train the AI to identify questions about the sales report, sample questions could include:
- What's the revenue?
- What is the total income?
- How much money did we make?
- What are our earnings?
- Could you provide the revenue figures?
- What is the gross revenue?
Was this article helpful?