Train Your AI

A step-by-step guide to train your AI. Follow these simple steps to get started. If you get stuck anywhere, please reach out to our team on discord


Objective and contents

This documentation provides detailed instructions on how to train your AI for the chat support system using hyperlinks and crawled website content.

Estimated time to complete this guide: ~10 min

What is Train Your AI?

The AI-powered chat support system is designed to provide efficient and effective responses to customer inquiries. To train the AI model and enhance its capabilities, the system leverages the information available on the dapp's website.

This documentation outlines the process of crawling through the website, extracting relevant content, and training the bot to generate responses that cater to customer needs.

By following these steps, you can enhance the AI's knowledge base and improve its ability to respond effectively to customer inquiries.

Step 1 - Create a free account on hashmail dapp console

👋🏼If you have already setup your dapp on the hashmail console, please skip to step 2

Get your access to the hashmail dapp console for free here

Step 2 - Accessing the Training Page

  1. Navigate to the chat support page of admin console. This page serves as the central hub for managing chat support functionalities.
  2. Locate the "Preferences" section within the chat support page and find the "Train Your AI" tab. Click on it to access the training page specifically designed for training the AI bot.

Step 3: Adding Hyperlinks

  1. On the "Train Your AI" page, you will see a button labeled "Add Hyperlink." Click on this button to initiate the process of adding a hyperlink to the training data.
  2. A modal window will open, providing fields to enter the necessary details for the hyperlink. Fill in the name or label for the hyperlink and the corresponding URL it should point to.
  3. Once you have entered the required information, click on the "Add Hyperlink" button within the modal to add the hyperlink to the training data.

Step 4: Crawling and Reviewing Links

  1. After adding the hyperlink, a new modal window will appear. This modal is responsible for crawling the website associated with the provided URL, extracting relevant information, and preparing it for training the AI model.
  2. In the modal, you can review the crawled links. Take a closer look at the extracted links and evaluate their relevance to ensure they contribute valuable information to the AI model.
  3. If you come across any irrelevant links, you have the option to remove them. Simply click on the delete button next to each link to exclude it from the training data.

Step 5: Initiating the Training Process

  1. Once you have reviewed and finalized the links, it's time to start the training process. Click on the "Train" button on the modal to initiate the training of the AI model.
  2. As the training progresses, you can monitor the progress by observing the green tick mark next to each link. The tick mark indicates that the respective link has been successfully processed and included in the training data.

By following these detailed steps, you can effectively train your AI for the chat support system by adding hyperlinks, crawling website content, and leveraging the extracted information to enhance the AI model's capabilities.

Previewing Bot Responses

  1. After completing the training process, you can access the widget designed for previewing the bot's responses.

  2. Locate the preview widget, which is typically available within the chat support page or a dedicated preview section.

  3. The widget will display a chat interface where you can enter test queries or simulate customer interactions to observe the bot's responses.

  4. Enter a query or message in the chat interface and submit it to the bot. The trained AI model will analyze the input and generate a response based on its learned knowledge.

  5. The bot's response will appear in the chat interface, allowing you to evaluate the accuracy, relevance, and overall quality of the generated response.

  6. Continue to test the bot by submitting different queries or messages, and observe how the bot handles various scenarios and inquiries.