Khoj

Khoj is an open-source 'AI second brain' that lets you build personal agents, schedule automations, and run research across your own documents alongside live web data. It indexes formats like PDF, plaintext, Markdown, org-mode, and Notion pages, then answers questions and runs multi-step tasks over that corpus, with a cloud instance at app.khoj.dev and a self-host path for users who want to keep their data on their own hardware.

The pitch is transparency and ownership: the core is on GitHub, you can swap in your own model providers, and the same agent framework runs locally or in their hosted app. It's aimed at researchers, knowledge workers, and tinkerers who want a ChatGPT-style assistant rooted in their personal corpus rather than a generic chatbot. Khoj the company also ships adjacent tools (Pipali, a desktop AI co-worker, and Open Paper, an academic-paper organizer), which signal an agents-and-research focus rather than a single chatbot product.

Pricing isn't surfaced clearly on the marketing site; the standard pattern here is a free self-hosted tier plus a paid cloud subscription, but you'll need to check app.khoj.dev for current numbers. The biggest caveat is that this is a smaller, community-driven project rather than a polished enterprise SaaS, so expect rough edges in the UI and setup compared to commercial agent platforms.