A self-hostable personal AI application that connects with local or online LLMs, gets answers from the web and your documents, builds custom agents, schedules automations, does deep research, and turns any online or local LLM into your personal, autonomous AI assistant.
One-Minute Overview#
Khoj is your personal AI second brain that can scale from an on-device personal AI to a cloud-scale enterprise AI. It supports chatting with any local or online LLM (like llama3, qwen, gemma, mistral, gpt, claude, gemini), gets answers from the internet and your documents, and is accessible through Browser, Obsidian, Emacs, Desktop, Phone or WhatsApp. You can create agents with custom knowledge, personality, and tools to take on any role.
Core Value: Integrates various LLM capabilities into a self-hostable personal AI system, enhancing information retrieval and knowledge management capabilities.
Quick Start#
Installation Difficulty: Medium - Requires basic knowledge of Python, Docker, and AI/LLM concepts, supports Docker container deployment
# Quick installation using Docker
docker run -d --name khoj -p 42110:42110 khojdev/khoj
Is this suitable for my needs?
- ✅ Knowledge Workers: Need to manage large document collections with intelligent search
- ✅ Researchers: Need to quickly retrieve information from multiple sources and conduct in-depth research
- ✅ Personal Knowledge Management: Want to build a personal AI assistant to organize and retrieve information
- ❌ Simple Information Queries: Only need basic search engine functionality
- ❌ Privacy-Conscious Users: Don't want to send data to third-party services
Core Capabilities#
1. Multi-LLM Support - Flexible AI Model Selection#
- Supports both local and online LLMs including GPT, Claude, Gemini, Llama, Qwen, Mistral, and more Actual Value: Users can choose the most suitable AI model based on their needs, balancing performance, cost, and privacy
2. Intelligent Document Retrieval - Get Information from Multiple Sources#
- Supports retrieving information from various formats including PDF, Markdown, Notion, Word documents, org-mode files, and more Actual Value: No need to manually organize documents; AI can quickly find relevant information from your personal knowledge base
3. Multi-Platform Access - Use Anywhere, Anytime#
- Access through Browser, Obsidian, Emacs, Desktop application, Phone or WhatsApp Actual Value: Seamless integration into your existing work environment without switching between applications
4. Custom Agents - Create Specialized AI Assistants#
- Agent system with custom knowledge, personality, and tools Actual Value: Create specialized AI assistants for different professional domains to improve productivity
5. Research Mode - Automated In-Depth Research#
- New experimental research mode that can automatically conduct deep research Actual Value: Automates repetitive research tasks, saving time and effort
Technology Stack & Integration#
Development Languages: Python, TypeScript, JavaScript Main Dependencies: FastAPI, React, OpenAI, LangChain, Sentence Transformers, PostgreSQL/SQLite, Redis Integration Method: API / SDK / Library
Maintenance Status#
- Development Activity: Highly active, with multiple commits per week
- Recent Updates: Recent new features and performance improvements
- Community Response: Active community support with good user engagement on Discord
Commercial & Licensing#
License: GPL-3.0
- ✅ Commercial Use: Allowed with attribution required
- ✅ Modification: Allowed
- ⚠️ Restrictions: Modified versions must be open source and include original license and copyright notices
Documentation & Learning Resources#
- Documentation Quality: Comprehensive
- Official Documentation: https://docs.khoj.dev/
- Sample Code: Rich examples and tutorials available