MaxKB: Build and Run Enterprise AI Agents

Overview

MaxKB helps teams build AI agents that can answer questions using an organization’s documents and online content, and carry out workflows with tools. Its RAG pipeline is designed to ground answers in a knowledge base, while its workflow engine and MCP support enable more involved tasks.

It is a fit for organizations that want to deploy knowledge-based assistants and connect them to existing systems without building the entire application layer themselves. MaxKB can be self-hosted and supports a range of public and private model providers.

Key Features

• Ingest documents and crawl online content for a searchable knowledge base.
• Split and vectorize content for retrieval-augmented question answering.
• Orchestrate agent tasks with workflows, a function library, and MCP tool use.
• Integrate question-answering capabilities into third-party business systems.
• Connect to public or private large language models, including DeepSeek, Llama, Qwen, OpenAI, Claude, and Gemini.
• Handle text, image, audio, and video input and output.
• Use PostgreSQL with pgvector as the database and vector store.

Use Cases

• Support teams can build customer-service assistants grounded in product or support documentation.
• IT and operations teams can provide employees with an internal assistant for company knowledge.
• Researchers and educators can make collections of documents easier to query and explore.
• Product teams can add knowledge-based Q&A to existing business systems when a custom-built agent platform would be unnecessary.

Getting Started

Start MaxKB in Docker:

docker run -d --name=maxkb --restart=always -p 8080:8080 -v ~/.maxkb:/opt/maxkb 1panel/maxkb

Then open http://your_server_ip:8080. See the repository README and MaxKB documentation for setup details.

Alternatives

• kotaemon: kotaemon focuses on document question answering and customizable RAG pipelines, while MaxKB also provides agent workflows and tool integration.
• Cheshire Cat AI Core: Cheshire Cat is an API-first framework for building custom agent services, while MaxKB provides a more complete platform for configuring knowledge-based agents.

Considerations

• Running the platform yourself means operating its container and persistent data volume, and configuring model providers and knowledge sources for your use case.
• The documented stack uses Vue.js, Django, LangChain, PostgreSQL, and pgvector, so teams should account for these components when evaluating deployment and maintenance.
• The project is licensed under GPL-3.0. Review the license terms before incorporating it into a product or distributing modified versions.
• The README lists default administrator credentials. Change them before exposing a deployment beyond a trusted environment.