Show HN: Rowboat – AI coworker that turns your work into a knowledge graph (OSS)
Rowboat is an open-source, local-first AI coworker designed to act as your personal work memory, building a 'living knowledge graph' from your emails and meeting notes. It leverages this persistent context to automate tasks, draft communications, and prepare for meetings, addressing the common challenge of providing robust, long-term memory to AI agents. Its emphasis on local storage, editable Markdown, and open-source nature strongly resonates with Hacker News's privacy-conscious and technically oriented audience.
The Lowdown
Rowboat is introduced as an innovative, open-source AI coworker that aims to solve the persistent context problem for AI agents by creating a dynamic, local knowledge graph from a user's digital workspace. Rather than relying on ephemeral, on-demand retrieval, Rowboat builds and maintains a comprehensive 'work memory' that continuously evolves with your professional interactions.
- Knowledge Graph Creation: It integrates with sources like Gmail and meeting note services (Granola, Fireflies) to extract and organize critical information such as decisions, commitments, deadlines, and relationships. This data is stored locally as linked and editable Markdown files, similar to Obsidian.
- Local AI Agent: A local assistant operates on top of this knowledge graph, offering capabilities like local shell access and Model Context Protocol (MCP) support for external tools. This allows it to perform tasks such as generating presentations, preparing meeting briefs, drafting emails, and creating voice notes.
- Differentiated Approach: Unlike typical AI tools that reconstruct context with each query, Rowboat's long-lived knowledge accumulates over time, making relationships explicit and user-inspectable. This 'memory that compounds' helps surface patterns and insights that might otherwise be missed.
- Privacy and Control: Designed as 'local-first,' all data remains on the user's machine, stored in a transparent, editable Markdown format. It's Apache-2.0 licensed and compatible with various LLMs, including local models, giving users full control over their data and AI processing.
- Background Agents: The system supports background agents for automated, routine tasks, such as drafting emails, generating daily summaries, or updating the knowledge graph proactively.
Rowboat distinguishes itself by offering a robust, transparent, and user-controlled solution to AI's context problem, making it a powerful tool for knowledge workers seeking an intelligent assistant that truly understands and remembers their unique work environment.