Khoj
Your AI second brain — self-hostable personal AI over your docs and the web.
About
Khoj is a personal AI app that works as a second brain. It chats with any local or online LLM (llama, qwen, mistral, gpt, claude, gemini), searches your own documents (PDF, markdown, org-mode, Notion) with semantic search, and answers from the web.
You can build custom agents with their own knowledge, persona and tools, and set up automations like personal newsletters. It ships clients for browser, desktop, phone, Obsidian, Emacs and WhatsApp. Fully open-source and self-hostable under AGPL-3.0, with a hosted option if you don't want to run it yourself.
What it can do
- Chat with local or cloud LLMs
- Semantic search over personal docs
- Custom agents with tools and personas
- Scheduled automations and newsletters
- Image generation and voice I/O
- Obsidian, Emacs, WhatsApp and mobile clients
Setup
- Install via pip (
pip install khoj) or pull the Docker image (ghcr.io/khoj-ai/khoj). - Start the server and open the web app.
- Connect a model — a local one through Ollama or an API key for a cloud provider.
- Point Khoj at your notes and documents to index them.
- Chat, search, and build agents from the dashboard.
Prerequisites
Python 3.10+ or Docker; a local model via Ollama or an API key for cloud models.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: 37.6k stars — strong community adoption; AGPL-3.0 — strong copyleft, slightly less permissive for embedding. Manual review: personal AI second brain; semantic search plus agents and automations; Obsidian/Emacs/WhatsApp clients show real user traction. License: AGPL-3.0.
★ 37,665 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 7.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · 37.6k stars — strong community adoption
- · AGPL-3.0 — strong copyleft, slightly less permissive for embedding
- · manual review: personal AI second brain; semantic search plus agents and automations; Obsidian/Emacs/WhatsApp clients show real user traction
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.