AingDesk: Run Local AI Models and Build a Private Knowledge Base

AingDesk is a direct and efficient AI assistant designed to put powerful tools within easy reach. It handles local AI models and major APIs through a simplified one-click setup. Users can build and manage private knowledge bases, configure intelligent agents, and share their entire setup with teammates. The platform includes native web search and supports server-side deployment for those who need a more robust infrastructure.

Model Management: AingDesk features one-click installation for local models and mainstream cloud APIs. The current lineup includes Gemma 3 in four sizes (1b, 4b, 12b, 27b), as well as qwq:32b, deepseek-r1:1.5b, and others. Because different models offer specific strengths—such as computer vision or tool-calling capabilities—you can easily select the one best suited for your current task.

Knowledge Base Construction: Users can create dedicated knowledge bases by uploading PDFs and other documents. This allows the AI to reference your specific files to provide more accurate, context-aware responses.

Web Search: This built-in feature allows the assistant to pull real-time information from the internet directly into your conversations, ensuring your data is always current.

Agent Creation: The platform includes several preset templates that are ready for immediate use, including a translation assistant, a literary translator, a social media writer, and a business writing expert. Alternatively, you can build your own custom agents from scratch to handle unique workflows.

Sharing: AingDesk allows you to share model access with friends or colleagues. For security, the connection service automatically shuts down once the share list is empty, ensuring no outside access persists when you don't want it.

Third-Party API Configuration: You can easily plug in your own keys for services like SiliconFlow, DeepSeek, Bailian (Tongyi Qwen), Baidu Qianfan, Hunyuan, Kimi, Baidu PaddlePaddle, Volcano Engine (Doubao), and more. Simply enter the API key and endpoint, then hit save to integrate them.

Installation and Use

Desktop Clients: Dedicated applications are available for macOS and Windows. These can be downloaded from the official site, CNB, or GitHub.

Server Deployment: There are two primary Docker options for setting up AingDesk on a server.

Docker Run:

docker run -d \
  --name node \
  -v $(pwd)/data:/data \
  -v $(pwd)/uploads:/uploads \
  -v $(pwd)/logs:/logs \
  -v $(pwd)/bin:/aingdesk/bin \
  -v $(pwd)/sys_data:/sys_data \
  -p 7071:7071 \
  -w /aingdesk \
  aingdesk/aingdesk

Docker Compose:

mkdir -p aingdesk
cd aingdesk
wget https://cnb.cool/aingdesk/AingDesk/-/git/raw/server/docker-compose.yml
# Then launch:
docker compose up -d
# or
docker-compose up -d

Build from Source:

git clone https://github.com/aingdesk/AingDesk.git
cd AingDesk
# macOS users: remove `@rollup/rollup-win32-x64-msvc` from frontend/package.json
cd frontend
yarn
cd ..
yarn
yarn dev

AingDesk provides a clean, local-first AI workspace with built-in sharing capabilities. By removing complicated setup rituals, it allows you to install the software and start working immediately.