Platform Overview

Learn about the Rowboat platform architecture and its key components

What is Rowboat?

Rowboat is an open-source AI coworker that turns your work into a knowledge graph and acts on it. Unlike traditional AI assistants that reconstruct context on demand, Rowboat maintains long-lived knowledge that accumulates over time.

Local-First Design

All data stored locally as plain Markdown with no proprietary formats or hosted lock-in

Knowledge Graph

Builds an Obsidian-compatible vault with backlinks for transparent working memory

Multiple Interfaces

Desktop app, CLI, Python SDK, and web platform for flexible workflows

Extensible

Supports MCP (Model Context Protocol) for connecting external tools and services

Key Capabilities

Memory & Context

Rowboat connects to your email and meeting notes, building a knowledge graph that:

• Remembers important context you don’t want to re-explain (people, projects, decisions)
• Understands what’s relevant right now (before meetings, during email replies)
• Accumulates knowledge over time with explicit, inspectable relationships
• Stores everything as editable Markdown files on your machine

AI-Powered Actions

Help you get work done by:

• Generating meeting prep from past decisions and open questions
• Drafting emails grounded in history and commitments
• Creating docs and PDF slide decks from your ongoing context
• Tracking follow-ups, decisions, and action items automatically

Background Agents

Spin up background agents for repeatable work:

• Draft email replies in the background using your context
• Generate daily voice notes with agenda and priorities
• Create recurring project updates from latest emails/notes
• Keep your knowledge graph up to date as new information comes in

Bring Your Own Model

Works with the model setup you prefer:

• Local models via Ollama or LM Studio
• Hosted models with your own API keys (OpenAI, Anthropic, Google, OpenRouter)
• Swap models anytime - your data stays in your local Markdown vault

Monorepo Structure

Rowboat is organized as a monorepo containing multiple applications and SDKs:

rowboat/
├── apps/
│   ├── x/                 # Electron desktop app
│   ├── rowboat/           # Next.js web dashboard
│   ├── rowboatx/          # Next.js frontend
│   ├── cli/               # CLI tool
│   ├── python-sdk/        # Python SDK
│   └── docs/              # Documentation site
├── CLAUDE.md              # AI coding agent context
└── README.md              # User-facing readme

Each application can be run independently, but they all work together to provide a unified experience for managing your knowledge graph and AI workflows.

Integrations

Rowboat builds memory from the work you already do:

• Gmail - Email history and threads
• Google Calendar - Meeting schedules
• Google Drive - Document access
• Granola - Meeting notes
• Fireflies - Meeting transcriptions

Optional Services

• Deepgram - Voice note transcription
• Brave Search - Web search capabilities
• Exa - Research search
• MCP Tools - Extend with Twitter/X, ElevenLabs, Slack, Linear, Jira, GitHub, and more

What Makes Rowboat Different?

Most AI tools search transcripts or documents on demand. Rowboat is different:

Context Accumulates

Knowledge builds over time instead of starting cold every time

Relationships are Explicit

Connections between concepts are visible and inspectable

Human Editable

Notes are stored as Markdown, not hidden inside a model

On Your Machine

Everything lives locally with no cloud dependencies

Next Steps

Architecture

Dive into the technical architecture and component design

Quickstart

Get started with Rowboat in minutes

Development

Learn how to contribute to the platform

API Reference

Explore the API documentation