MaxKB

Turn your company documents into an AI assistant that answers questions for staff or customers, in a web app you host yourself.

Turn your company documents into an AI assistant that answers questions for staff or customers, in a web app you host yourself.

The gist

MaxKB is an open-source platform for building AI assistants on top of your own documents. You upload files or point it at web pages, and it answers questions based on that content. It is used for customer service, internal company knowledge bases, research and education.

What it does

  • Takes uploaded documents, or collects pages from websites automatically
  • Splits and indexes that content so the AI can look up the right parts
  • Answers questions using your own material, which cuts down on made-up answers
  • Lets you build step-by-step workflows that can use extra tools
  • Works with many AI models, both private ones like DeepSeek, Llama and Qwen and public ones like OpenAI, Claude and Gemini
  • Handles text, images, audio and video, and can be added to other business systems without coding

How it works

Architecture Diagram
MaxKB Architecture Diagram

You load your documents into MaxKB, and it breaks them into small pieces it can search. When someone asks a question, MaxKB finds the most relevant pieces. It sends them to the AI model you chose, together with the question. The answer comes back in a web chat or inside another system you connected it to.

Best for

  • Customer service bots that answer from your own help pages and documents
  • An internal company knowledge base staff can ask in plain language
  • Teams in research or education who want Q&A over a set of documents

Watch out for

  • Setup uses Docker (a tool for running packaged apps) on a computer or server you control
  • It ships with a default admin username and password; change them straight away
  • You still need an AI model, either a paid public one or a private one you run yourself
  • Some documentation is in Chinese first, for example the offline install guide

What you need: Install it on your own computer. Cost: free, open source.