Lemon AI
A self-hosted AI agent that runs local models through Ollama or vLLM, with Docker-based code execution and an editor for generated HTML.
Lemon AI is a local AI agent for people who want research, coding and data analysis tools running on their own hardware. It combines web browsing and task execution with an editor for refining generated HTML pages, reports and charts without regenerating the whole page.
The agent can plan tasks, take actions and review its results. It keeps conversation memories and has an experience repository for learning from previous work and adapting to business needs. Its tools cover web research, document processing, content creation, code generation and execution.
Local models run through Ollama or vLLM, with support for DeepSeek, Kimi, Qwen, Llama, Gemma and GPTOSS. This setup doesn't depend on cloud model services. Web searches and browsing still involve online resources. You can also connect Claude, GPT, Gemini or Grok through their APIs; those requests go to cloud providers. A hosted subscription offering is available alongside the client and self-hosted deployment options.
Code runs inside a local Docker-based VM sandbox that isolates execution from the host's files and operating system. The HTML editor lets you select a page element and ask the agent to change that section, insert content or adjust formatting. You can also edit text directly.
Lemon AI runs on macOS and Linux, and on Windows through WSL. Docker Desktop is part of the local setup. The application recommends a modern processor and at least 4 GB of RAM.
