Skip to main content
Rowboat Hero

Open-source AI coworker with persistent memory

Rowboat connects to your email and meeting notes, builds a long-lived knowledge graph, and uses that context to help you get work done - privately, on your machine. All your data is stored locally as plain Markdown in an Obsidian-compatible vault. No proprietary formats, no hosted lock-in.

Context that compounds

Most AI tools reconstruct context by searching transcripts. Rowboat maintains long-lived knowledge instead - relationships are explicit, notes accumulate over time, and memory compounds rather than starting cold.

Local-first by design

Everything lives on your machine as plain Markdown. Inspect, edit, back up, or delete your data at any time. Compatible with Obsidian and any text editor.

Real artifacts, not just chat

Generate meeting briefs, draft emails grounded in history, create PDF slide decks from context, and produce real work artifacts - not just conversational responses.

Background agents

Spin up agents to handle repeatable work automatically: draft email replies, generate daily voice notes, create recurring project updates, and keep your knowledge graph current.

What you can do with Rowboat

Meeting prep

Pull past decisions, open questions, and relevant threads into crisp briefs before meetings.

Build decks

Generate PDF slides using context from your knowledge graph.

Draft emails

Write replies grounded in your history and commitments.

Voice memos

Record notes that automatically capture and update key takeaways in the graph.

Follow-ups

Track decisions, action items, and owners so nothing gets dropped.

Query your graph

Ask questions about people, projects, and context from your work.

How it’s different

Rowboat maintains an Obsidian-compatible vault of plain Markdown notes with backlinks - a transparent “working memory” you can inspect and edit.
Your knowledge graph lives in ~/.rowboat/knowledge/ and follows entity-based organization:
  • People/ - Contacts with roles, organizations, and interaction history
  • Projects/ - Active work with status, decisions, and commitments
  • Organizations/ - Companies and teams you work with
  • Topics/ - Domain knowledge and recurring themes
  • Voice Memos/ - Recorded notes organized by date
The result is memory that accumulates over time, not retrieval that starts cold every session.

Integrations

Rowboat builds memory from the work you already do:
  • Gmail - Email threads and communication history
  • Google Calendar - Meeting schedules and context
  • Google Drive - Document access
  • Granola - Meeting notes
  • Fireflies - Meeting transcripts

Bring your own model

Rowboat works with the model setup you prefer:
  • Local models via Ollama or LM Studio
  • Hosted models - OpenAI, Anthropic, Google (bring your own API key)
  • Swap models anytime - your data stays in your local Markdown vault
Configuration lives in ~/.rowboat/config/models.json.

Extend with tools (MCP)

Connect to external tools and services via Model Context Protocol (MCP). Plug in search, databases, CRMs, support tools, and automations.
Examples: Exa (web search), Twitter/X, ElevenLabs (voice), Slack, Linear, Jira, GitHub, and custom internal tools.

Open source and transparent

Rowboat is built in the open. Inspect the code, contribute features, or fork and customize for your workflow.

View on GitHub

Star the repo and follow development

Join Discord

Get help and share feedback with the community

Ready to get started?

Quickstart Guide

Download Rowboat and connect your first data source in under 5 minutes

Build docs developers (and LLMs) love