We use cookies: By continuing to browse our website, you agree to the use of cookies, which help us analyze traffic to improve your experience. For more information, please read our privacy policy.

GPT Interface

Quick Guide

for creating and configuring a ChatBot in GPT Interface

This guide provides the essential steps and recommendations for preparing, configuring, and improving a ChatBot using the GPT Interface platform.

Preparation for Implementation Before Proceeding to the Builder

Before you start building your ChatBot, it is important to prepare and answer a few key questions. These questions will help you define the goals and functionalities of your ChatBot, as well as gather the necessary baseline information for its successful operation. For more in-depth preparation, we suggest filling out the ChatBot Passport. If you encounter difficulties, you can refer to our guide on filling out the ChatBot Passport. Additionally, our ChatBot Jeronimo is available to assist by answering questions about the installation and settings of the ChatBot.

Why do I need a ChatBot?

The first question you need to answer is why you need a ChatBot. Possible reasons may include improving customer interaction, automating responses to frequently asked questions, providing real-time information, and reducing the workload of your support team. By clearly defining your goals, you will be able to create a ChatBot that meets the specific needs of your business.

What do I want it to do for my clients?

Once you have determined why you need a ChatBot, the next step is to define what you want it to do for your clients. A ChatBot can provide technical support, answer questions, direct clients to appropriate products or services, and gather feedback. Defining the specific functions and tasks of the ChatBot will help you configure it correctly and ensure maximum benefit for your clients.

How, with what information will it do this?

The last question you need to answer is how and with what information the ChatBot will perform its tasks. The information should be structured and organized in logical sequences. This may include official websites, email correspondence, user reviews, company documents, and API documentation. Collecting and organizing this information will ensure accurate and useful responses from the ChatBot.

Configuration

Quick Start with Assistant and Uploaded Files from the Extractor

To quickly implement your ChatBot, you can use an assistant with the OpenAI API. The main information you want to include may relate to your company, such as contacts, services, prices, etc. It is important to set up the ChatBot with rewriting disabled and semantic extraction disabled.

Step 1:

Information

Define the necessary URLs that you will use.

    1. Gather the URLs of the pages you will use for information.
    2. Ensure that the pages contain up-to-date and relevant information about your company, such as contacts, services, prices, etc.

Step 2:

Registrations

    1. Registration in GPT Interface:

      – Log into the GPT Interface platform and create a new account if you do not have one.
      – Confirm your registration via the email you will receive.

    2. Profile Settings:

      – Log into your profile and fill in the necessary information such as name, contacts, and other details.

    3. Adding API:

      – Register on the OpenAI website for access to their API and fill in the necessary information.
      – Add funds to your balance.
      – Copy and save the generated key with the organization name.
      – Copy and save the API key that will be generated for you.
      – Log into GPT Interface.
      – Invoke the main menu from the “hamburger” button at the top right of the screen.
      – Select the “Settings” menu and from the navigation bar select “API” (the second icon).
      – Add the generated OpenAI organization and API keys.
      – Save.

    4. Allowing Models from the “Settings” Menu:

      – Invoke the main menu from the “hamburger” button at the top right of the screen.
      – From the menu, select “Settings”.
      – You will see a list of all models you have access to. In the list of models, check the box for allowing (at the beginning of the row) and the checkbox for chat mode (the first from the group on the right). You can define a name to recognize the model more easily, as well as edit the description it has.

Step 3:

Saving the Necessary Information

    1. In the Vector Lab Engine module, select the “Extractor” tab.
    2. Enter the URL from which you will process information.
    3. Download the information from the address.
    4. Edit the downloaded information so that it contains only the necessary data.
    5. Define a name for the file, for example “about-us” (in Latin, without spaces).
    6. Save the file in the storage – it will be saved in the shared file storage.
    7. Repeat the steps for the other URLs you defined in step 1.

Step 4:

Vector Storage

Create a new vector storage and associate the file that is in the shared storage, generated by the extractor.

    1. Again in the Vector Lab Engine module, select the “File Storage” tab.
    2. Under the list of uploaded files, in the Vector Storage section, add a name for the storage and from the “+Add” button create a new vector storage.
    3. Enter edit mode for the new storage.
    4. From the dropdown menu, select the file you saved with the extractor and associate it with this vector storage.

Step 5:

Configuring the Assistant

Again in the Vector Lab Engine module, select the “Assistants” tab.

    1. Select the option to create a new assistant.
    2. Enter a name and description for the assistant.
    3. Select a model (e.g., GPT-4 Omni).
    4. Configure parameters such as creativity and colorfulness or leave them as set by default.
    5. Save the assistant.

Step 6:

Associating the Vector Storage with the Assistant

    1. Enter edit mode for the assistant you created.
    2. Enable the File search option.
    3. From the dropdown menu, select the created vector storage and associate it with the assistant.

Step 7:

Allowing the Assistant in Chat Mode

    1. Invoke the main menu from the “hamburger” button at the top right of the screen.
    2. From the menu, select “Settings”.
    3. Find the new assistant in the list of models and check the box for allowing and the checkbox for chat mode.

Step 8:

Configuring the ChatBot

From the main menu, select “Home”. Select the “Widget ChatBot Builder” module.

    1. Enable the ChatBot from the toggle button (ON/OFF).
    2. Specify a name by which the ChatBot will be identified.
    3. Enter the URL of the site where the ChatBot will operate.
    4. Disable rewriting: In the ChatBot settings, disable the rewriting option.
    5. Disable semantic extraction: In the ChatBot settings, disable the semantic extraction option.
    6. Select a model for generating responses: Choose the assistant you created.
    7. Add a main instruction to the ChatBot, which may include instructions for its purpose, tone, and style of responses, etc.
    8. Define the depth of context from previous iterations (questions and answers) to be included as context for the question that will be sent. Keep in mind that a larger number will significantly increase token costs.
    9. Define a token limit for the response that the model should generate; if it is too small, the responses may not be complete. The model will stop generating a response once it reaches the defined limit.
    10. Press the save button.

Step 9:

Integration

    1. Again in the “Widget ChatBot Builder” module, select the “Integration” tab.
    2. Copy the provided code.
    3. Paste the code into your site.

Step 10:

Testing and Optimization

    1. Test the ChatBot: After configuring the assistant, test the ChatBot with various queries to ensure it responds correctly and provides accurate information.
    2. Collect feedback: Gather feedback from users and analyze the results. This will help you identify potential issues and areas for improvement.
    3. Optimization: Make necessary adjustments in the assistant settings and files to improve the performance and accuracy of the ChatBot.

Analysis and Improvement

Monitoring and analyzing the performance of the ChatBot are essential for its successful operation and continuous improvement. By collecting and analyzing data, you can identify the strengths and weaknesses of the ChatBot, understand how users interact with it, and make necessary adjustments. This includes analyzing frequently asked questions, response times, and customer satisfaction. Continuous improvement of the ChatBot will ensure that it remains a useful and effective tool for your business.

Conclusion

By following these steps, you will be able to quickly and easily launch your own ChatBot. This will allow you to familiarize yourself with the system, the principles of operation, and the various methods of organizing information. Analyze and experiment with different approaches. Track the needs of your clients and the questions they ask. Gradually, over time, you will develop the information array and be able to optimize your token costs by transferring some of the information to the vector database.