A self-hosted personal assistant that uses tools, memory, and agentic execution to automate digital workflows.

Leon is an open-source, self-hosted web app and server that functions as a personal AI assistant. It is designed to automate tasks and manage digital workflows while prioritizing user privacy and data ownership. The software is deployed as a server application with a companion web interface, running on Linux, macOS, and Windows via Node.js.
Users can interact with the assistant through different execution modes depending on the complexity of the task. It supports both local and remote AI providers, allowing for a balance between strict data control and high processing capability. The system is built to be extensible, utilizing a modular architecture of skills and toolkits to interact with the host environment and execute real world tasks instead of only replying with plain text.
The system architecture consists of a main runtime for routing and memory, a web application, and a set of bridges that connect the core to Python services and system binaries. It utilizes a specific hierarchy where skills lead to actions, which then trigger tools and functions. This structure allows Leon to access system utilities, media workflows, and coding assistance. The assistant maintains a compact self-model to ensure its behavior remains consistent over time without flooding the AI with unnecessary context.
This software is intended for users who want a grounded AI that can interact with their specific machine setup and local files rather than relying on general model responses from a cloud provider. It provides a framework for those who wish to build a personalized digital ecosystem with a high degree of transparency.
Leon is a privacy-focused assistant for users who require a customizable, agentic AI capable of running on their own hardware.
A self-hosted personal AI assistant that integrates with multiple messaging apps to perform tasks and automate workflows.
Manage research through AI summarization, multi-format content ingestion, and multi-speaker podcast generation with local data control.
Join our newsletter to get shiny new open source software delivered to your inbox. Unsubscribe anytime.