Create a RAG-powered AI Agent

Creating an RAG powered AI Agent is pretty straightforward. If you use the default settings, you can even create a chatbot with just a few clicks without typing anything!

First, navigate to the AI Agent list page following the steps below. And the click on "New AI Agent".

You should then see a "Create New AI Agent" modal popping up, similar to the one below. If you want to create a RAG chatbot with the default settings, just click the "Create RAG Chatbot" button. Or if you want some customization, you have the following options:

• LLM Endpoint: From the drop-down menu, you can choose either Llama 3.1 70B, Llama 3 8B, both hosted by Theta EdgeCloud, or GPT-4o-min. If you want to use your own LLM endpoints (e.g. GPT-4o or a finetuned LLM), you can setup a custom LLM endpoint (see also here).
• Chatbot Name (Optional): If you want, you can give this chatbot a name. The name can be changed anytime after the chatbot is created.
• Max Tokens and Temperature: These are LLM settings for OpenAI compatible LLM endpoints.

If you don't know which options to use, please just stick to the default settings. You can update all of the above settings after creating the RAG chatbot. Please check out Customize Your AI Agent and Update Your AI Agent's Knowledge Base for more information.

After clicking on "Create AI Agent", the next page lets you add documents to the Agent's knowledge base. You can either upload a file, copy over some text, or provide a URL of a webpage. For example, the screenshot below shows the "Webpage" tab where you can enter a URL, e.g. https://docs.thetatoken.org/docs/what-is-theta-network. Once you hit the "Add Document" button, the system will scrape the page automatically and add it to the knowledge base.

You can also skip this step and add documents to the knowledge base later from the details page of the AI Agent. You can add/delete documents to/from the knowledge base any time after the chatbot is created. The docs can be managed programmatically through API as well. With this feature you can update realtime information, such as the score of a basketball match, stock price actions, etc. for the Agent. Please check out Update Your AI Agent's Knowledge Base and Configure the AI Agent via APIs for more details.

Once the RAG-powered AI Agent is created, you will see it on the AI Agent list page. You can create multiple agents, each with a different knowledge base and settings.

Next, you can interact with the chatbot through the AI Agent Playground.