Khoj
Your AI second brain — self-hostable personal copilot that answers from the web and your docs.
About
Khoj is a personal AI app that scales from an on-device copilot to cloud-scale team AI. You chat with any local or online LLM, get answers grounded in your own documents (PDF, Markdown, Notion, Word, org-mode, images) and the live web, and reach it from the browser, desktop, phone, Obsidian, Emacs, or even WhatsApp. You can create custom agents with their own knowledge, persona, model, and tools, and automate recurring research into personal newsletters and smart notifications.
At 37,000+ stars it is one of the most-starred personal-AI projects on GitHub, and the "self-hostable, always" stance is real: AGPL-3.0 licensed, with a cloud app at app.khoj.dev for people who don't want to self-host. The team also ships Pipali, an open-source AI coworker that runs on your computer. The trade-offs: AGPL-3.0 is copyleft, so check it against commercial plans, and as a personal tool it competes with the built-in AI in every OS and chat app — its moat is openness and the document-grounding quality the team benchmarks publicly.
What it can do
- Chat with any local or online LLM from one interface
- Answers grounded in your docs plus live web search
- Custom agents with own knowledge, persona, model, and tools
- Research automation: personal newsletters and smart notifications
- Clients for browser, desktop, phone, Obsidian, Emacs, and WhatsApp
- Fully self-hostable and AGPL-3.0 open source
Setup
- Try it instantly at app.khoj.dev, or self-host:
pip install khoj(or Docker) per the setup docs. - Connect your model: local (llama3, qwen, gemma, mistral) or online (GPT, Claude, Gemini, DeepSeek).
- Index your documents and notes; Khoj builds its semantic search over them.
- Chat, generate images, and talk out loud from the web app, desktop, phone, Obsidian, Emacs, or WhatsApp.
- Create custom agents with their own knowledge, persona, and tools; schedule research automations.
Prerequisites
For self-hosting: Python or Docker and enough machine for your chosen model (or just a cloud API key); nothing for the cloud app.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: 37,689 GitHub stars (GitHub API), AGPL-3.0, pushed Aug 2 2026; the README documents the model flexibility, document support, client list, and custom agents; docs corroborate the self-host path. Strengths: enormous adoption, genuinely open and self-hostable, the team publishes retrieval benchmarks. Docked for: (1) AGPL-3.0 copyleft — check against commercial plans; (2) last push ~2.5 months ago; (3) competes with built-in OS/chat-app AI. License: AGPL-3.0.
★ 37,689 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 9.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · Adoption 9: starScore from 37,689 stars on khoj-ai/khoj
- · Maintenance 8: pushed 2026-08-02 (recency 8); releases est 8; 169 open issues on 37.7k stars (issues 9)
- · Openness 10: AGPL-3.0 OSI-approved (0.7*10) + official docs verified at docs.khoj.dev (0.3*10)
- · Manual 9: README documents model flexibility, document support, client list, and custom agents; team publishes retrieval benchmarks; docked for AGPL copyleft on commercial plans, ~2.5mo since last push
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.