Jan
Jan is an open-source personal AI platform that gives users full control over their own intelligence, offering a self-hosted agent and backend infrastructure. It supports a wide range of open and proprietary models, allowing users to run, route, and govern AI interactions on their own infrastructure. The platform emphasizes privacy and customization, with features like persistent memory and a modular architecture for building personalized AI workflows.
- Artificial Intelligence
- AI Agents
- Developer Tools
- Software Only
Funding
Founders
Product
Problem
Users seeking personalized AI assistance often face a trade-off between convenience and control, as mainstream AI services are typically closed, cloud-based, and store user data on external servers. This raises concerns about data privacy, limits customization, and prevents users from fully owning their AI experience and its underlying models.
Solution
Jan provides an open-source, personal AI platform that operates on the user's own infrastructure, ensuring data sovereignty and privacy. The platform includes Jan Agent, a core agent that can be run on a personal VM or container, and Tokamak, a self-hosted backend that provides routing, a fusion model, and governance/audit capabilities. This architecture allows users to connect to and switch between a wide array of open and proprietary models, including Llama, Mistral, and DeepSeek, while maintaining full control. Jan is built in public, fostering a community-driven development approach and ensuring transparency. The platform is designed to be extensible, with features like persistent memory that learns user preferences over time, creating a highly personalized and private AI experience.
Target Audience
Jan targets individual developers, privacy-conscious professionals, and enterprises seeking a customizable, self-hosted AI solution that ensures data privacy and offers model flexibility.
Features
- Self-hosted architecture with Jan Agent and Tokamak backend for complete data control and privacy.
- Support for a wide range of models, including open-source options like Llama, Mistral, Qwen, and DeepSeek, as well as commercial APIs.
- Persistent memory feature that carries context and user preferences across sessions for a personalized experience.
- Governance and audit capabilities through the Tokamak backend, providing oversight of AI interactions.
- Open-source and built in public, allowing for community contributions and transparent development.