rowboat: a desktop AI coworker with a persistent local knowledge graph and integrated work surfaces

What it solves

Rowboat is a desktop AI coworker that provides a persistent, long-lived memory of your work context. Unlike tools that perform cold retrieval on demand, Rowboat accumulates context over time in a local knowledge graph, allowing the AI to act on your data across various work surfaces.

How it works

Rowboat indexes email, meetings, Slack conversations, and assistant interactions into a backlinked knowledge graph stored as plain Markdown files on your machine. It provides integrated work surfaces—including an email client, a browser, a meeting note-taker, and a code mode—that use this context to perform tasks. It supports local models via Ollama or LM Studio, hosted models via API keys, and extends its capabilities via the Model Context Protocol (MCP) to connect to external tools like GitHub, Slack, and Jira.

Who it’s for

Professionals and developers who want an AI assistant with a persistent memory of their professional life and local-first data ownership of their work context.

Highlights

• Living Knowledge Graph: Indexes work data into an Obsidian-style backlinked graph for compounding memory.
• Built-in Work Surfaces: Includes a dedicated email client, browser, meeting note-taker, and coding environment.
• Background Agents: Event-driven or scheduled agents that can search the web, use the browser, and write code.
• Local-first Design: All data is stored locally as plain Markdown, avoiding proprietary lock-in.
• Extensible: Supports MCP servers for third-party tool integration and custom app building within the platform.