khoj

Your AI second brain. Self-hostable. Get answers from the web or your docs. Build custom agents, schedule automations, do deep research. Turn any online or local LLM into your personal, autonomous AI (gpt, claude, gemini, llama, qwen, mistral). Get started - free.

⭐ 34,383 stars on GitHub · 🍴 2,182 forks · 📜 License: agpl-3.0 · 💻 Language: Python

What is khoj?

If you want a self-hosted AI assistant that can work across your private documents, web research, and multiple LLM providers, Khoj is one of the most complete options available. Its differentiator is flexibility: you can run it locally, connect it to commercial models, or point it at open-weight models like Llama, Qwen, Mistral, Gemma, or DeepSeek.

Main components

• Chat interface for querying local documents, web results, and connected LLMs.
• Document ingestion for PDFs, Markdown, Word files, Notion exports, org-mode, images, and more.
• Semantic search to find relevant notes and files without exact keyword matching.
• Custom agents with their own knowledge, persona, model, and tool access.
• Automation features for scheduled research, newsletters, and smart notifications.
• Multi-client access through browser, desktop, mobile, Obsidian, Emacs, and WhatsApp.
• Support for voice interaction, message playback, and image generation.

Clear use cases

• Build a private AI knowledge base over internal docs, notes, PDFs, and project material.
• Run a personal research assistant that combines web search with your existing files.
• Create role-specific agents for engineering, writing, support, sales, or ops workflows.
• Use local LLMs for sensitive work while keeping the option to call GPT, Claude, Gemini, or other hosted models when needed.
• Give power users an AI assistant inside existing tools like Obsidian or Emacs instead of forcing everything into a new app.
• Schedule recurring research summaries, personal newsletters, or alerts around topics you track.

The biggest strength is model and knowledge-source flexibility — Khoj does not lock you into one AI vendor, one interface, or one document format. Compared with commercial “AI workspace” tools, it gives technical teams much more control over where data lives, which models are used, and how agents are shaped for specific tasks. That matters if you care about privacy, cost control, or avoiding another SaaS silo.

Khoj is not just a thin ChatGPT wrapper; it is closer to a self-hostable AI workbench for people who want retrieval, agents, automations, and multi-model support in one place. The tradeoff is that teams expecting a zero-maintenance consumer app may prefer the hosted version, while self-hosters should be comfortable managing LLM endpoints, indexing, and deployment details.

Best for developers, researchers, knowledge workers, and IT teams that want a private, customizable AI assistant over their own documents and workflows.

Topics: the project is tagged with popular topics:

• 🏷️ agent
• 🏷️ ai
• 🏷️ assistant
• 🏷️ chat
• 🏷️ chatgpt
• 🏷️ emacs
• 🏷️ image-generation
• 🏷️ llama3
• 🏷️ llamacpp
• 🏷️ llm

📸 Screenshots

Quick install

The project supports Docker Compose:

git clone https://github.com/khoj-ai/khoj.git
cd khoj
docker compose up -d

Check the README in the repo for required env variables.

Minimum system requirements

⚡ Deploy fast on VSIS

Use the VSIS VPS Standard 4GB RAM / 2 vCPU / 50GB SSD (~150k/tháng) plan from VSIS.NET — high-speed VN-based VPS, 24/7 support, ideal for running khoj smoothly.

🎯 Benefits:

• One-command docker compose up -d deploy in 2 minutes
• Dedicated IPv4, root access, unmetered domestic bandwidth
• Daily snapshot backup
• Free install assistance from the VSIS team

👉 See matching VPS plans at vsis.net

Resources

• 🔗 GitHub: khoj-ai/khoj
• 🌐 Homepage: https://khoj.dev
• 📚 Official docs: see README in the repo
• 💬 Community: GitHub Issues + Discussions

Article compiled from GitHub data on 05/05/2026. Star/fork counts may have changed — see live numbers via the GitHub link.