Khoj is a self-hosted AI assistant that lets you build a personal knowledge system powered by local or cloud LLMs.
The tool solves the problem of accessing information scattered across the web and your personal documents by providing a unified interface for semantic search and retrieval-augmented generation. It works by indexing your documents and web content, then using an LLM to answer questions grounded in that indexed material. You can connect it to multiple LLM providers including open-source models like Llama and Mistral, or commercial services like OpenAI and Claude. The system supports building custom agents that can perform autonomous tasks and scheduling automations to run on a schedule.
Khoj suits developers and knowledge workers who want to maintain privacy by self-hosting their AI infrastructure, or who need to work with proprietary documents that shouldn't be sent to external APIs. It works well for deep research workflows where you need to synthesize information from multiple sources, and for building domain-specific assistants trained on your own data. The tool integrates with existing workflows through plugins for Emacs and Obsidian, and supports interaction via WhatsApp. If you prefer a fully managed solution or don't need local deployment, a cloud-hosted version is available, though the self-hosted option is the project's primary focus.
The project shows consistent development activity with regular updates across its core components. The codebase maintains active work on the backend infrastructure and agent capabilities. Documentation receives ongoing attention with examples and guides being refined. The project accepts community contributions and maintains responsiveness to issues and feature requests. Development spans multiple areas including the web interface, plugin ecosystems, and LLM integration layers, indicating a broad maintenance effort rather than focus on a single component.
