This domestic open source RAG project ChatWiki is amazing. It can build an enterprise-level AI knowledge base and WeChat robot

Written by

Audrey Miles

Updated on:July-01st-2025

ChatWiki is a domestic open source knowledge base AI question-answering system. The system is built on the Large Language Model (LLM), Retrieval Enhanced Generation (RAG) and GraphRAG knowledge graph, providing out-of-the-box data processing, model calling and other capabilities.

1. Enterprise Private Knowledge Base

It covers all enterprise knowledge scenarios. For example, private enterprise documents uploaded by industries such as education, finance, law, healthcare, government departments, etc. ChatWiki is responsible for segmenting and cleaning the documents, and large AI models such as DeepSeek are responsible for quickly giving accurate answers based on the existing content in the knowledge base.

Currently it supports more than 20 mainstream models around the world, including DeepSeek R1, doubao pro, qwen max, Openai, and Claude.

For example, if you ask DeepSeek "How do I apply for xx rural planning license?", DeepSeek's answer may have the risk of "hallucination", but by learning knowledge in specific fields on ChatWiki, it can give an accurate answer.

2. Support access to DeepSeek ChatWiki supports access tomore than 20 mainstream models around the world, including DeepSeek R1, DeepSeek V3,
doubao pro, qwen max, Openai, Claude, etc. You only need to configurethe API Key to access DeepSeek.

3. Support access to WeChat ecosystem

Through the API interface, you can seamlessly access the official account and WeChat customer service to create a dedicated artificial AI intelligent chat robot;
It also supports embedding in websites, desktop clients, WebApps, WeChat applets, Douyin enterprise accounts, Kuaishou accounts, video accounts and API calls, etc., fully covering the needs of enterprise multi-terminal business scenarios.
Support real-time viewing of the conversation content between users and AI robots;
Human-machine collaboration supports keyword conversion to manual customer service.

ChatWiki access to WeChat customer service demonstration

4. GraphRAG Knowledge Graph Construction

ChatWiki supports knowledge graph retrieval based on vector retrieval, full-text retrieval, and hybrid retrieval .

Compared with traditional RAG systems, GraphRAG can handle complex queries more efficiently and provide more accurate and relevant answers, especially for datasets involving a large number of entities and relationships.

Simply put, GraphRAG = Knowledge Graph + Retrieval Enhanced Generation RAG.

5. More features

① External documents can be set

It supports the rapid publication of knowledge base content as a publicly accessible documentation site, provides SEO optimization, multi-visit statistics and other functions, and makes it easy to create a branded customer support portal.

② Import documents in multiple formats

Supports importing documents in various formats including OFD, Word, Excel, PPT, PDF, markdown, etc.

③Support building AI workflow

In ChatWiki, you can quickly build workflows by dragging and dropping nodes. For example, you can freely choose multiple atomic capabilities such as AI dialogue, question classification, knowledge base, Http request, judgment branch, etc., and quickly build business processes by arranging and combining them in a visual drag-and-drop manner.

④Connect with third-party business data

It can be seamlessly embedded in business systems and integrate third-party data sources (such as sales, inventory, and logistics data) in real time, breaking down data silos and significantly shortening manual processing time.

⑤Permission hierarchical setting support

Provides enterprise-level multi-level permission control and supports role assignment (administrator/editor/read-only member) to meet the needs of sensitive data management and team collaboration.

⑥Large model semantic segmentation

RAG segmentation determines the accuracy of AI answers and is very important. ChatWiki not only supports common segmentation, but also supports large-model semantic segmentation. It performs segmentation based on sentence vector similarity to prevent the loss of key semantic information between segments.

⑦Support downloading desktop version

The local deployment version supports desktop clients and is ready to use after downloading.

6. 6 unique advantages

7. Support multiple deployments

ChatWiki supports multiple deployment methods:

Docker deployment;
Offline docker deployment;
No Docker deployment required, completely local deployment, source code installation.

The docker deployment and installation process is as follows:

(1) Install Docker (skip this step if already installed)

sudo curl -sSL https://get.docker.com/ | CHANNEL=stable sh

(2) Clone or download the chatwiki project code

git clone https://github.com/zhimaAi/chatwiki.git

(3) Modify Docker environment variables as needed (optional)

vim ./chatwiki/docker/.env

(4) Modify project configuration parameters as needed (optional)

vim ./chatwiki/configs/chatwiki/config_pro.ini

(5) Use Docker Compose to build and start the project

cd chatwiki/docker
docker-compose up -d

(6) Use load balancing or nginx to configure the domain name to point to the corresponding service (optional)

(7) Access the management backend via IP+port (need to open the specified port ${CHAT_SERVICE_PORT}, default is 18080) or domain name

8. ChatWiki open source address