Rowboat
Open-source AI coworker, with memory
Last verified:
What is Rowboat?
Rowboat is an open‑source, local‑first AI coworker that turns your work data—such as email, meeting notes, and documents—into a living knowledge graph on your own machine. It uses that graph to help you draft emails, prep meetings, track projects, and produce other work artifacts while staying privacy‑focused and on‑device by default.
The tool integrates with common work sources like Gmail and meeting‑transcript providers, automatically extracting people, projects, decisions, and commitments into an Obsidian‑compatible vault of Markdown notes with backlinks. This vault acts as Rowboat’s transparent, inspectable memory, letting you review and edit the context it uses to answer questions and take actions.
Rowboat is designed for individual knowledge workers, small teams, and creators who want a private, persistent AI assistant that understands their personal context without shipping sensitive data off‑device. It also supports integration with external tools and APIs, including MCP servers and web‑search providers, so you can extend it for workflows like research, customer‑facing assistants, or internal‑tool automation.
Rowboat pricing
Pricing model: Freemium
Rowboat is open‑source and self‑hostable for free, with no mandatory subscription for the core desktop app or local‑first features. For those using the hosted version or advanced integrations, Rowboat connects to external services such as MCP servers, Brave Search, and Exa Search, each of which has its own pricing tiers including free tiers and paid plans. The company also offers enterprise‑style arrangements for bigger deployments, often negotiated per organization rather than a fixed public SaaS plan, giving teams control over where data stays and how much they pay for external API usage.
Rowboat pros
- Local‑first architecture keeps sensitive data on your machine
- Open‑source client you can inspect and modify
- Builds an Obsidian‑compatible knowledge graph from your work
- Automatically connects to email and meeting‑note sources
- Remembers people, projects, decisions, and commitments over time
- Helps draft emails and other work artifacts in context
- Assists with meeting prep and project tracking
- Transparent memory you can browse and edit as plain Markdown
- Supports MCP servers for custom tool integrations
- Integrates web‑search providers like Brave and Exa
- Can run with local or cloud‑hosted LLM backends
- Cross‑platform desktop app for Mac, Windows, and Linux
- Easily shareable vault structure for collaboration
- Lightweight, low‑code setup via Docker and scripts
- Community‑driven templates and workflows in the open‑source repo
Rowboat cons
- Requires local setup and Docker knowledge for self‑hosted use
- Not all features are available without external API keys
- Knowledge‑graph quality depends on quality and volume of your work data
- Limited out‑of‑the‑box support for non‑English or niche tools
- Desktop‑only; no standalone mobile app yet
- Performance can be slower on lower‑end machines when syncing large inboxes or notes
- Some advanced features require manual configuration instead of guided UI
- Smaller third‑party ecosystem compared to big‑platform AI assistants
Frequently asked questions about Rowboat
What is Rowboat and what does it do?
Rowboat is an open‑source, local‑first AI coworker that connects to your email and meeting‑note sources, builds a long‑lived knowledge graph from your work, and uses that graph to help you draft emails, prepare for meetings, track projects, and create other documents on your machine. It exposes its memory as an Obsidian‑compatible Markdown vault so you can inspect and edit the context it uses to interact with you.
Is Rowboat free to use?
The Rowboat desktop app and its core open‑source code are free to download, run, and self‑host. You can run it locally with Docker without paying Rowboat Labs directly, though you may incur costs from external services such as OpenAI keys, MCP providers, or search APIs if you choose to integrate them.
How does Rowboat keep my data private?
Rowboat is designed local‑first, meaning your emails, meeting notes, and the resulting knowledge graph are stored on your own machine by default. The app exposes your data as local Markdown files and runs inference either locally or via your chosen cloud provider, so Rowboat Labs does not host or routinely access your raw work data unless you opt into a hosted setup.
Can Rowboat integrate with my email and meeting tools?
Yes, Rowboat can connect to providers such as Gmail, Granola, and Fireflies to ingest email and meeting transcripts, extracting entities like people, projects, and decisions into its knowledge graph. These integrations allow the assistant to draft replies and prep agendas based on your actual conversation history and commitments.
What is the Obsidian‑compatible vault in Rowboat?
The Obsidian‑compatible vault is a folder of plain‑text Markdown notes, each representing a node in Rowboat’s knowledge graph. Notes include backlinks and relationships that mirror how Rowboat ‘thinks’ about your work, and you can open this vault in Obsidian or any Markdown editor to browse, search, and edit your AI’s memory.
Does Rowboat work without an internet connection?
Rowboat can operate in a mostly offline mode because its note vault and memory live locally, but certain features such as web searches, external API calls, and cloud‑based LLMs require an internet connection. You can still view and edit your knowledge graph and draft content offline, as long as the specific tool or model you use supports local inference.
How does Rowboat handle MCP servers and custom tools?
Rowboat lets you connect MCP servers and import their tools into the app, so your AI assistant can take actions like querying databases, triggering workflows, or updating records in your internal systems. You configure these tools in Rowboat’s settings, then refer to them by name in natural language to build multi‑step, tool‑augmented workflows.
Can multiple people collaborate using Rowboat?
Rowboat is primarily designed for individual use on a single machine, but teams can collaborate by sharing the local vault folder or syncing it via a shared drive or Git, then each person running their own instance. For more advanced shared‑workspace scenarios, teams often layer their own sync or version‑control setup on top of Rowboat’s Markdown vault.
What are the system requirements to run Rowboat?
To run Rowboat you need Docker installed plus a reasonably modern machine capable of handling local LLM inference or at least the web UI and vault sync. Exact needs depend on whether you run small local models or connect to cloud APIs, but the official docs provide a minimal setup script that spins up the app at http://localhost:3000 and works across Mac, Windows, and Linux.
How does Rowboat’s pricing differ from hosted support assistants?
Unlike most hosted customer‑support assistants that charge per‑conversation or per‑seat, Rowboat is primarily self‑hosted and open‑source, so you pay only for your own cloud services or hosting costs. Rowboat Labs may offer additional hosted or managed plans or enterprise support, but the core product is designed to be cost‑transparent and avoid per‑conversation fees by keeping your data and control on‑device.