Khoj is a personal AI app ("your AI second brain") that extends your capabilities, scaling from an on-device personal AI to a cloud-scale enterprise AI.
Khoj is an open-source, self-hostable personal AI application that lets you chat with any local or online LLM and get answers from the internet and your documents (including image, pdf, markdown, org-mode, word, and notion files). It can be accessed from your browser, Obsidian, Emacs, Desktop, Phone, or WhatsApp, and supports creating custom agents with their own knowledge, persona, chat model, and tools. It also automates repetitive research, delivers personal newsletters and smart notifications, provides advanced semantic search, and can generate images and play messages aloud. It can be run privately on your own computer or used via a cloud app.
A personal AI "second brain" app to extend your capabilities — chatting with LLMs, getting answers from your documents and the internet, creating custom agents, automating research, semantic search, and generating images, available self-hosted or as a cloud service.
Although the entrypoint code shows no explicit autonomy signals (subprocess, exec, etc.), the README describes Khoj as a personal AI app that goes well beyond pure suggestion. It can "create agents with custom knowledge, persona, chat model and tools to take on any role," "automate away repetitive research," deliver newsletters and smart notifications to your inbox, fetch answers from the internet and your docs, generate images, and talk out loud. These are multi-step agentic workflows (research automation, scheduled notifications, tool-using agents) that execute without per-step human confirmation. This places it at Tier 4 (acts autonomously) rather than a simple advisory/suggestion tool. There is no described mandatory human-in-the-loop confirmation before agent actions like sending scheduled emails or running automations.