Khoj is an open-source, self-hostable personal AI application that functions as a second brain, scaling from on-device deployment to enterprise cloud infrastructure. With 32,800 stars on GitHub and 5,144 commits from 62 contributors, Khoj has matured into one of the most comprehensive personal AI assistants available. The platform supports any LLM backend including Llama3, Qwen, Gemma, Mistral, GPT, Claude, Gemini, and DeepSeek, letting users switch between local and cloud models seamlessly. Khoj provides semantic search across personal documents (PDF, Markdown, Notion, Word, Org-mode, images), custom agent creation with tunable personalities and tools, and a dedicated research mode via the /research command for deep investigation. Multi-platform access spans browser, Obsidian plugin, Emacs integration, desktop apps, mobile, and WhatsApp. Additional capabilities include image generation, text-to-speech, newsletter automation, and RAG-based knowledge integration. The latest release (v2.0.0-beta.25) introduced performance improvements and expanded agent capabilities. Deployment options include the free cloud app at app.khoj.dev, Docker containers for self-hosting, and enterprise on-premises solutions. Licensed under AGPL-3.0, Khoj is built primarily in Python and TypeScript.