MaxKB: A Software Engineer's Guide to Building AI Agents

1Panel-dev/MaxKB

MaxKB provides a robust foundation for building intelligent agents, which saves you a ton of time and effort. Instead of building a chatbot or knowledge base from scratch—dealing with data ingestion, vector databases, and agent logic—MaxKB handles a lot of that heavy lifting for you.

Here's why it's so useful

Rapid Development
You can quickly create and deploy agents without needing deep expertise in every component of an AI pipeline. This lets you focus on the business logic and user experience rather than the underlying infrastructure.

Knowledge Management
It’s built to work with your internal data. You can feed it various documents (PDFs, Markdown files, etc.) to create a searchable knowledge base. This is perfect for building customer support bots, internal help desks, or even specialized research assistants.

Agent Customization
The platform is flexible. You can integrate different large language models (LLMs) and tools to create agents that perform specific tasks, like answering questions, summarizing documents, or even executing commands.

Open Source
Being open source means you have full control. You can inspect the code, customize it to your needs, and integrate it seamlessly with your existing tech stack.

Getting MaxKB up and running is straightforward, thanks to its use of Docker. This makes the setup process consistent and portable across different environments. The simplest way to get started is by using Docker Compose.

First, make sure you have Docker and Docker Compose installed on your system.

Open your terminal and clone the MaxKB GitHub repository.

git clone https://github.com/1Panel-dev/MaxKB.git
cd MaxKB

MaxKB includes a docker-compose.yml file that defines all the necessary services, including the MaxKB application, a database, and a vector store. Just run the following command to start everything.

docker-compose up -d

The -d flag runs the containers in the background, so your terminal is free.

Once the containers are up and running, you can access the MaxKB web interface. It's typically available at http://localhost:8080.

When you first log in, you'll be prompted to set up an admin account. After that, you're ready to start building!

Let's walk through a simple example of how you might use MaxKB to build an agent that can answer questions based on a set of internal documents.

In the MaxKB web interface, you'll first create a new "Knowledge Base." This is where you'll upload your documents. Let's say you have a few .md files detailing your company's HR policies. You'd upload them here. MaxKB will automatically process these documents and store them in its vector database.

Next, you'll create a new "Agent." When configuring the agent, you'll select the knowledge base you just created as its primary data source. You'll also choose the large language model (LLM) you want the agent to use. You can configure it to use a local model or an API from services like OpenAI.

Once the agent is created, you can interact with it directly within the MaxKB interface. Try asking a question like, "What is the policy for requesting vacation time?" The agent will look up the relevant information from the HR policy documents and provide a concise, accurate answer.

The real power comes from integration. MaxKB provides an API that lets you interact with your agents programmatically. Here's a quick example of how you might use Python to query the agent you just built.

import requests
import json

# Replace with your MaxKB API endpoint and API key
API_URL = "http://localhost:8080/api/v1/agent/chat"
API_KEY = "your_api_key_here"  # Get this from your MaxKB settings

headers = {
    "Content-Type": "application/json",
    "Authorization": f"Bearer {API_KEY}"
}

# The question you want to ask the agent
query_data = {
    "question": "What is the policy for requesting vacation time?"
}

try:
    response = requests.post(API_URL, headers=headers, data=json.dumps(query_data))
    response.raise_for_status()  # Raise an exception for bad status codes

    result = response.json()
    print("Agent's Answer:")
    print(result.get("answer"))

except requests.exceptions.RequestException as e:
    print(f"An error occurred: {e}")

This simple code snippet shows you how you can integrate your custom-built agent into a web application, a Slack bot, or any other system. You're not just limited to the web UI; you have the flexibility to use your agents wherever they're needed.