Rowboat Development Roadmap: AI Workflow Platform Releases Through 2026

The rowboatlabs/rowboat development roadmap centers on four strategic pillars—AI engine enhancements, UI modernization, ecosystem expansion, and enterprise security—delivering major releases from Q2 2025 through early 2026.

The development roadmap for rowboatlabs/rowboat outlines a clear trajectory for this modular, open-source AI workflow platform. The project maintains three primary products—a local Electron desktop app, a cloud-based Next.js web dashboard, and TypeScript/Python SDKs—each guided by upcoming milestones that prioritize multi-model orchestration, visual workflow editing, and self-hosted deployments.

Rowboat Product Architecture

Understanding the roadmap requires context on the three products under active development:

Product	Location	Purpose
Rowboat Desktop	apps/x/	An offline-first Electron app bundling AI agents, a knowledge graph, and prompt-driven workflows with secure local data handling.
Rowboat Web Dashboard	apps/rowboat/	A Next.js front-end for cloud-based project management, pipeline orchestration, and multi-user collaboration.
Rowboat SDKs	apps/cli/, apps/python-sdk/	TypeScript and Python libraries for programmatic access, CI/CD integration, and custom agent development.

Strategic Pillars of the Rowboat Roadmap

The development roadmap for rowboatlabs/rowboat organizes work across four strategic pillars that span all three products.

Core AI Engine Enhancements

This pillar focuses on the orchestration layer located in apps/x/packages/core/. Upcoming work includes richer agent coordination protocols, multi-model routing capabilities, and an expressive knowledge graph schema supporting "Projects," "Artifacts," and "Feedback" entities.

User Experience and UI Modernization

The interface layer will see a migration to fluid React/Vite architectures, a complete redesign of the Electron renderer process, and the introduction of a visual workflow editor. The new editor component is already staged at apps/rowboat/app/projects/[projectId]/workflow/workflow_editor.tsx.

Platform and Ecosystem Expansion

Integration breadth is a priority, with planned connectors for Google Calendar, Notion, and Slack. The roadmap includes a plugin marketplace architecture and first-class support for self-hosted LLMs via Ollama and Vercel AI Gateway, reducing dependency on cloud-only providers.

Reliability, Security, and Packaging

Enterprise readiness features include code-signing and notarization pipelines for macOS and Windows (configured in .github/workflows/electron-build.yml), role-based access control (RBAC) for team collaboration, and hardened sandboxing of the Electron preload bridge.

Release Timeline and Milestones

The development roadmap for rowboatlabs/rowboat follows a chronological release schedule with specific deliverables for each version.

v0.2 "AI-CoPilot" — Q2 2025

This release establishes multi-model routing capabilities supporting OpenAI, Anthropic, Google, and OpenRouter providers. It expands the knowledge graph schema to include "Projects," "Artifacts," and "Feedback" entities, and introduces first-party agents for "Meeting-Prep" and "Document-Summarizer" workflows.

v0.3 "Unified Workspace" — Q3 2025

The highlight of this milestone is the new Workflow Editor, a React-based drag-and-drop canvas for visual pipeline construction. This release also delivers persistent cloud synchronization for projects across the Desktop and Web products, and exposes a Plugin API allowing third-party LLM provider integrations.

v0.4 "Enterprise Hardened" — Q4 2025

Security and compliance take center stage with automated code-signing and notarization pipelines defined in .github/workflows/electron-build.yml. This version introduces role-based access control (RBAC) for multi-user team collaboration and implements an auditable event log with compliance export capabilities.

v0.5 "Self-Hosted & Edge" — Early 2026

The final milestone in the current roadmap enables full Docker-compose deployment using Dockerfile.qdrant for vector storage. It delivers offline LLM support via Ollama integration with local model caching, and produces an edge-optimized Electron bundle targeting a sub-30MB distribution size.

Implementation Details: Adding Agents to the Core Engine

The core AI engine in apps/x/packages/core/ uses a skill-based architecture. To add a new agent capability, developers implement the AgentSkill interface and export the module.

The following example demonstrates adding a Sentiment-Analysis skill:

// apps/x/packages/core/src/application/assistant/skills/sentiment-analysis/skill.ts
import { AgentSkill } from '../../lib/skills';
import { LLMProvider } from '../../../providers/llm';

// 1️⃣ Declare the skill metadata
export const sentimentAnalysisSkill: AgentSkill = {
  name: 'sentiment-analysis',
  description: 'Analyzes the sentiment of a given text block.',
  // 2️⃣ Define the execution function
  async run(context, input) {
    const provider = LLMProvider.get('openai');   // auto-selects from config
    const response = await provider.chat({
      model: 'gpt-4o-mini',
      messages: [{ role: 'user', content: `Analyze sentiment: ${input}` }],
    });
    return response?.choices?.[0]?.message?.content ?? 'No result';
  },
};

// 3️⃣ Register the skill (auto-discovered via workspace glob)
export default sentimentAnalysisSkill;

Registration mechanism: The core package's builtin-tools.ts uses a glob('**/skill.ts') pattern to auto-discover and register all exported AgentSkill modules. Configuration for LLM providers is read from ~/.rowboat/config/models.json as implemented in apps/x/packages/core/src/providers/llm.ts.

Key Source Files Supporting the Roadmap

The following files contain the concrete implementations and configurations referenced in the development roadmap for rowboatlabs/rowboat:

File	Role
apps/x/packages/core/src/application/assistant/skills/	Directory containing core agent implementations (e.g., meeting-prep, document-collab).
apps/x/packages/core/src/providers/llm.ts	Unified LLM abstraction supporting OpenAI, Anthropic, Google, OpenRouter, and Ollama.
apps/rowboat/app/projects/[projectId]/workflow/workflow_editor.tsx	React-based visual workflow editor component (v0.3 milestone).
apps/docs/docs/development/roadmap.mdx	Official high-level roadmap documentation.
.github/workflows/electron-build.yml	CI pipeline for code-signing and notarization (v0.4 milestone).
Dockerfile.qdrant	Container definition for self-hosted vector storage (v0.5 milestone).
apps/cli/src/app.ts	Entry point for the TypeScript CLI SDK.

Summary

• The development roadmap for rowboatlabs/rowboat targets four strategic pillars: AI engine enhancements, UI modernization, ecosystem expansion, and enterprise security.
• Near-term releases (v0.2–v0.3) deliver multi-model routing, an expanded knowledge graph, and a React-based visual workflow editor located in apps/rowboat/app/projects/[projectId]/workflow/workflow_editor.tsx.
• Mid-term releases (v0.4–v0.5) introduce code-signing pipelines in .github/workflows/electron-build.yml, role-based access control, and self-hosted Docker deployments using Dockerfile.qdrant.
• The core engine supports extensible agent development through the AgentSkill interface in apps/x/packages/core/, with auto-discovery via glob patterns in builtin-tools.ts.

Frequently Asked Questions

What is the current release timeline for Rowboat?

The development roadmap for rowboatlabs/rowboat schedules v0.2 "AI-CoPilot" for Q2 2025, v0.3 "Unified Workspace" for Q3 2025, v0.4 "Enterprise Hardened" for Q4 2025, and v0.5 "Self-Hosted & Edge" for early 2026. These dates are derived from the official roadmap document at apps/docs/docs/development/roadmap.mdx and active feature branches in the repository.

Can I self-host Rowboat on my own infrastructure?

Yes, self-hosting is a primary goal of the v0.5 milestone scheduled for early 2026. The roadmap includes full Docker-compose deployment support using Dockerfile.qdrant for vector storage and offline LLM integration via Ollama. The unified LLM provider in apps/x/packages/core/src/providers/llm.ts already supports local model routing, laying the groundwork for fully air-gapped installations.

How do I contribute a new agent or skill to Rowboat?

Contributors can add agents by implementing the AgentSkill interface in the apps/x/packages/core/src/application/assistant/skills/ directory. The core engine auto-discovers skills via a glob pattern defined in apps/x/packages/core/src/application/lib/builtin-tools.ts (around line 1050), so exporting your skill module is sufficient for registration. The LLM provider configuration is read from ~/.rowboat/config/models.json, allowing your agent to leverage multiple model backends without code changes.

What distinguishes Rowboat Desktop from the Web Dashboard?

Rowboat Desktop (apps/x/) is an offline-first Electron application designed for secure local data handling and individual productivity, bundling the AI engine and knowledge graph directly on the user's machine. Rowboat Web Dashboard (apps/rowboat/) is a Next.js application optimized for multi-user collaboration, cloud-based project management, and API gateway functionality. The development roadmap for rowboatlabs/rowboat aims to unify these experiences through persistent cloud sync starting with the v0.3 "Unified Workspace" release.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}