Ask the AI to prepare a meeting brief by pulling together past emails, decisions, and open questions about a specific person.
Draft a slide deck or document by asking the assistant about your project roadmap, it draws from your stored notes and emails.
Store your work context as plain Markdown files you can open in Obsidian, edit yourself, or back up without any vendor lock-in.
Connect Slack, calendar, and web search via Model Context Protocol to give the AI access to your live work context.
Requires connecting at least one data source such as email before the knowledge graph is useful, some features need their own API keys.
Rowboat is a free, open-source desktop application that acts like an AI assistant for your work. You download it for Mac, Windows, or Linux, and it connects to things you already use, such as your email and your meeting notes. From all of that material it builds what the project calls a knowledge graph: a long-lived store of who you work with, what projects are running, what decisions have been made, and what you have committed to do. The main idea is that the assistant keeps remembering context over time instead of starting from scratch on every request. Because of that memory, you can ask it to do practical things in plain language. The README gives examples like asking it to build a slide deck about next quarter's roadmap and getting back a PDF, or asking it to prepare you for a meeting with a specific person and getting a short brief that pulls together past decisions, open questions, and relevant email threads. A key design choice is that everything runs locally on your own machine. Your data is stored as plain text Markdown files in a folder you can open, read, edit, back up, or delete yourself. The notes are compatible with Obsidian, a popular note-taking app, so nothing is locked inside a hidden system. You can also see and change the knowledge graph directly, since it is just Markdown. Rowboat is flexible about which AI model does the thinking. You can run local models through tools like Ollama or LM Studio, or connect a hosted model by supplying your own API key, and you can switch between them at any time. It can also connect to outside services, such as web search, Slack, calendars, and issue trackers, through a standard called Model Context Protocol. The project is made by Rowboat Labs, a Y Combinator company. Some features, including voice input, voice output, and certain integrations, are optional and require their own API keys.
← rowboatlabs on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.