Product Introduction
Definition: Maestri is a high-performance, native macOS spatial orchestration platform and infinite canvas designed specifically for autonomous coding agents. It functions as a specialized runtime environment that replaces traditional tabbed terminal emulators with a visual, node-based interface, enabling developers to manage multiple AI agents—such as Claude Code, Codex, and OpenCode—within a single, GPU-accelerated workspace.
Core Value Proposition: Maestri eliminates the "coder bottleneck" caused by terminal fragmentation and context switching. By providing a spatial UI for multi-agent workflows, it allows developers to visualize agent interactions, delegate tasks through PTY (Pseudo-Terminal) orchestration, and maintain project context using on-device AI. Its core pillars are performance (Swift/Metal), privacy (zero cloud/telemetry), and seamless Apple Intelligence integration.
Main Features
1. Spatial Canvas & Custom Metal Engine: Maestri features a canvas engine built from the ground up using Swift, SwiftUI, and the Metal framework. This GPU-accelerated architecture ensures buttery-smooth panning and zooming across an infinite workspace, even when rendering dozens of active terminal nodes. Unlike Electron-based tools, Maestri maintains a minimal RAM footprint, ensuring that system resources remain dedicated to the AI agents and local compilers rather than the UI layer.
2. PTY Orchestration & Agent Interconnectivity: The app utilizes advanced PTY (Pseudo-Terminal) orchestration to host any CLI-based coding agent. Users can physically connect terminal nodes by dragging lines between them, creating a logic flow where agents can prompt each other. For example, a "Claude Code" node can be configured to delegate sub-tasks to a "Gemini" or "OpenCode" node, facilitating a multi-agent collaborative ecosystem without manual copy-pasting of outputs.
3. Ombro On-Device AI Companion: Ombro is a dedicated AI monitor powered by Apple Intelligence and local Foundation Models. It operates entirely on-device to observe terminal activity in the background. When an agent completes a long-running task—such as a full-repo refactor or test suite execution—Ombro generates a summary of the outcomes and suggests next steps via a floating "Liquid Glass" window, allowing the developer to stay productive in other applications without losing track of agent progress.
4. Integrated Sketching and Markdown Notes: The canvas supports non-textual data through hand-drawn architecture diagrams and sticky notes. These nodes are not just static elements; they can be linked to terminal PTYs. Agents can be programmed to write progress updates, architecture decisions, or summaries directly into Markdown-based sticky notes, creating a living document of the development process that persists locally.
Problems Solved
1. Terminal Tab Fatigue and Context Loss: Traditional workflows involve "alt-tabbing" through numerous terminal windows or tabs, which leads to cognitive overload. Maestri solves this by placing every active process on a single spatial plane, allowing the human brain to use spatial memory to track different parts of a project.
2. The AI Agent Monitoring Bottleneck: As coding agents become faster, the developer becomes a passive observer. Maestri's Ombro assistant solves the "watch-and-wait" problem by providing asynchronous summaries, ensuring the developer is only interrupted when human intervention or architectural decisions are required.
3. Data Privacy and Corporate Compliance: Many AI tools rely on cloud-based telemetry and analytics. Maestri addresses security concerns by implementing a strict "Zero Cloud" policy. Workspaces are stored as local JSON files, notes are Markdown, and no usage data is ever collected, making it suitable for developers working on proprietary or sensitive codebases.
Target Audience:
- Software Architects: Designing complex systems while agents implement components in parallel.
- AI Engineers: Benchmarking and orchestrating multiple LLM agents simultaneously.
- Full-Stack Developers: Managing frontend, backend, and database agents in a single visual context.
- Privacy-Conscious Developers: Users requiring native macOS performance without third-party tracking.
Use Cases:
- Multi-Agent Debugging: Running a specialized debugging agent alongside a testing agent, with both reporting to a central "Conductor" agent.
- Live Architecture Mapping: Drawing a system flow on the canvas and positioning live terminals next to the corresponding architectural components.
- Asynchronous Code Generation: Launching long-running refactoring tasks and relying on Ombro to provide a high-level summary once the agents finish.
Unique Advantages
1. Native macOS Excellence (No Electron): Maestri is 100% native, built with SwiftUI and AppKit. This results in pixel-perfect TUI (Text User Interface) rendering, lower latency, and better battery efficiency compared to web-tech-based IDEs. It adheres to macOS system conventions, including Dark Mode, native shortcuts, and Liquid Glass design principles.
2. Zero-Latency On-Device Intelligence: By leveraging Apple Foundation Models via Apple Intelligence, the Ombro companion provides intelligent summaries without API costs or network latency. This ensures the orchestration layer remains functional even in offline environments.
3. Tool Agnosticism: Maestri does not lock users into a specific AI provider. It functions as a horizontal runtime for any CLI tool. Developers can switch between Claude, GPT-based tools, or local Ollama instances seamlessly within the same canvas.
Frequently Asked Questions (FAQ)
1. Does Maestri send my code to its own servers? No. Maestri is built with a zero-telemetry, zero-cloud architecture. All workspace configurations, canvas layouts, and notes are stored locally on your Mac. While the coding agents you run (like Claude Code) may connect to their respective providers, Maestri itself never intercepts, collects, or uploads your data.
2. What are the system requirements for running Maestri? Maestri requires macOS 26.2 or later and is optimized specifically for Apple Silicon (M1, M2, M3, M4 chips). It utilizes Metal GPU acceleration and Apple Foundation Models, which perform best on Mac hardware with Unified Memory.
3. Can I use Maestri with my existing IDE like VS Code or Cursor? Yes. Maestri is designed to complement your favorite editor. It includes "one-click" integration to jump from a terminal node directly into the project folder within VS Code, Zed, or Xcode, allowing you to switch from agent orchestration to manual coding instantly.
