Skip to main content

What is Macroa?

Macroa is an AI agent operating system designed around a core principle:
AI is used only where no deterministic process can do the job. Everything else is infrastructure.
While most AI agent frameworks focus on making LLMs easier to use, Macroa focuses on making LLMs unnecessary for as much of the agent lifecycle as possible—so that when AI is finally invoked, it’s for the task it’s uniquely suited to: reasoning about ambiguity.

Pulse’s Role in Macroa

Pulse is the proactive cognition subsystem of Macroa. It is the first component being built because it’s what makes Macroa fundamentally different from reactive agent frameworks like LangChain, AutoGen, or CrewAI.

Why Build Pulse First?

Every other agent framework is reactive. An agent acts because something triggered it—a user message, a webhook, a cron job. Remove the trigger, and the agent does nothing. Pulse eliminates this limitation by enabling agents to notice when something is worth attention and act on it autonomously. Without Pulse, Macroa would just be another reactive framework. With Pulse, it becomes proactive.

The Macroa Stack

Macroa is composed of several interconnected components, with Pulse as the foundation:

1. Pulse (Proactive Cognition)

The hierarchical signal perception system that monitors the environment and produces scoped questions for the agent. Runs entirely on local compute with zero LLM calls in normal operation. Status: Active development (Proof of Concept)

2. Kernel (Agent Runtime)

The execution environment that manages agent lifecycle, capability-based security, and module coordination. The kernel receives scoped questions from Pulse and wakes the appropriate agent context. Status: Pre-implementation specification

3. SDK (Module Development)

The developer interface for building Macroa modules—reusable agent capabilities that register with Pulse and declare their signal fingerprints. Status: Pre-implementation specification

4. Shell (User Interface)

The command-line and natural language interface for interacting with agents, viewing Pulse activations, and providing feedback to improve model accuracy. Status: Pre-implementation specification

5. Memory (Persistent State)

The structured knowledge store that agents use to persist facts, user preferences, and task state. Changes to memory namespaces are signal sources for Pulse. Status: Pre-implementation specification

The Complete Vision

When fully built, Macroa will enable:

For Users

  • Agents that notice when you need help without being asked
  • Privacy-first design—all learning happens on your machine
  • Performance that improves over time as the system learns your patterns
  • No subscription costs for basic agent functionality

For Developers

  • A module system for building reusable agent capabilities
  • Capability-based security that prevents modules from accessing resources they shouldn’t
  • Deterministic infrastructure that removes the need to use LLMs for routine coordination
  • A platform that works entirely offline (optional network access for specific capabilities)

For Researchers

  • A reference implementation of hierarchical signal perception for AI agents
  • Open-source architecture for studying proactive agent behavior
  • Transparent training loops and model weights for reproducible research

Why “Operating System”?

Macroa is called an operating system because it provides the same fundamental services for AI agents that traditional operating systems provide for programs:
Traditional OSMacroa
Process schedulingAgent activation via Pulse
Memory managementStructured memory namespaces
Capability-based securityModule permissions and sandboxing
Inter-process communicationSignal bus and event coordination
Device driversSignal source adapters (filesystem, network, etc.)
Just as you don’t reimplement process scheduling every time you write a program, with Macroa you don’t reimplement proactive behavior every time you build an agent module.

Project Status

Macroa is in active early development. Pulse is the first component being built as a proof of concept. The architecture documents for all components are complete, and implementation is proceeding in the order specified in the build documents. This is a long-term project with a clear vision but no artificial deadlines.

Open Source Philosophy

Macroa is released under the AGPLv3 license:
  • Free for everyone to use, study, and modify
  • Any modifications must be published under the same license
  • Commercial use is permitted, but source code must remain open
This ensures that the platform remains transparent and that improvements benefit the entire community.

Learn More

Design Paper

The technical rationale behind Pulse’s architecture

GitHub Repository

Source code and complete documentation

Contributing

Macroa is an open-source project welcoming contributions. The project is in early stages, and the best way to contribute right now is to:
  1. Read the architecture documents to understand the vision
  2. Experiment with Pulse once the proof of concept is released
  3. Provide feedback on the design and implementation approach
  4. Build modules that demonstrate novel use cases for proactive agents
Visit the GitHub repository for more information on how to contribute.

Build docs developers (and LLMs) love