ChatGPT
AI conversational assistant for answering questions, writing, and coding help
Khoj
Personal AI assistant that learns from your documents
Self-hostable AI assistant that chats using any LLM, searches your documents and the web, and creates custom agents with tunable personality and knowledge bases.
Overview
Your Personal AI Second Brain
Khoj is a personal AI application designed to extend your cognitive capabilities, scaling seamlessly from on-device deployment to cloud-scale enterprise solutions. It transforms how you interact with information by combining conversational AI with deep document understanding.
Flexible AI Integration
Connect with any local or online LLM including llama3, qwen, gemma, mistral, GPT, Claude, Gemini, and DeepSeek. Access Khoj from your browser, Obsidian, Emacs, desktop, phone, or WhatsApp, ensuring your AI assistant is always within reach.
Intelligent Knowledge Management
Khoj reads and understands PDFs, Markdown, Notion exports, Word documents, org-mode files, and images. Its advanced semantic search surfaces relevant information instantly, while custom agents can be configured with specific knowledge bases, personas, and tools to handle specialized tasks.
Automation & Research
Automate repetitive research workflows, receive personalized newsletters, and get smart notifications. The experimental research mode delivers comprehensive answers by synthesizing information from multiple sources. Generate images, use voice interaction, and enjoy multimodal capabilities.
Whether you self-host for complete privacy or use the cloud service, Khoj remains open-source under AGPL-3.0, giving you control over your AI infrastructure.
Highlights
Multi-LLM support with local and cloud models (GPT, Claude, Llama, Gemini, Mistral)
Advanced semantic search across PDFs, Markdown, Notion, Word, and org-mode files
Custom agents with configurable personality, knowledge bases, and tools
Cross-platform access via browser, Obsidian, Emacs, desktop, mobile, and WhatsApp
Pros
- Self-hostable for complete data privacy and control
- Works with both local and cloud LLMs for flexibility
- Extensive document format support including PDFs, Notion, and Markdown
- Active development with custom agents and research automation features
Considerations
- AGPL-3.0 license may restrict commercial derivative works
- Self-hosting requires technical setup and infrastructure management
- Advanced features like research mode are experimental
- Enterprise hybrid deployments require vendor engagement
Managed products teams compare with
When teams consider Khoj, these hosted platforms usually appear on the same shortlist.
Looking for a hosted option? These are the services engineering teams benchmark against before choosing open source.
Fit guide
Great for
- Knowledge workers managing large document collections
- Privacy-conscious users wanting self-hosted AI assistants
- Teams needing custom AI agents with specialized knowledge
- Researchers automating literature review and synthesis workflows
Not ideal when
- Users seeking plug-and-play solutions without any configuration
- Organizations requiring proprietary licensing terms
- Teams without technical resources for self-hosting setup
- Use cases demanding real-time collaborative editing
How teams use it
Personal Knowledge Base Search
Instantly find relevant information across years of PDFs, notes, and documents using semantic search instead of keyword matching
Automated Research Synthesis
Use research mode to gather, analyze, and synthesize information from the web and personal documents into comprehensive reports
Custom Support Agent
Create a specialized agent trained on company documentation to answer team questions via WhatsApp or Obsidian
Privacy-First AI Assistant
Self-host Khoj with local LLMs to keep sensitive documents and conversations completely private on your infrastructure
Tech snapshot
Python51%
TypeScript37%
CSS3%
HTML3%
Emacs Lisp3%
JavaScript2%
Frequently asked questions
What document formats does Khoj support?
Khoj reads PDFs, Markdown, Notion exports, Word documents, org-mode files, images, and more. It uses semantic search to understand content across all formats.
Which LLMs work with Khoj?
Khoj supports both local models (llama3, qwen, gemma, mistral) and cloud APIs (GPT, Claude, Gemini, DeepSeek), giving you flexibility in deployment and cost.
How do I create custom agents?
Custom agents can be configured with specific knowledge bases, personalities, chat models, and tools. Check the Khoj blog for step-by-step guides on agent creation.
Is Khoj suitable for enterprise deployment?
Yes, Khoj offers cloud, on-premises, and hybrid enterprise solutions. Visit khoj.dev to learn about enterprise licensing and deployment options.
Project at a glance
Active- Stars
- 32,232
- Watchers
- 32,232
- Forks
- 1,931
LicenseAGPL-3.0
Repo age4 years old
Last commit2 weeks ago
Self-hostingSupported
Primary languagePython
Last synced 10 hours ago