Development environment · These builds are intended for testing.
MAH App iconMAH
Checking account…

MAH Product System

One Project. Many Machines. Continuous Evolution.

MAH brings your machines and AI Agents from different providers together to keep your project improving. Evolution discovers opportunities, advances the work, checks results, and continues into the next round. Explore the direction with MAH, and stay involved in the decisions that matter.

MAH Workbench · Simplified interface with example content.

MAH Workbench illustration

MAH Workbench layout: project and tasks on the left, the selected task and Agent conversation in the center, events and project documents on the right.

Simplified interface with example content.

Multi-Agent Collaboration

Different Agents. One Team.

Combine AI Agents from different providers. Bring their different strengths to the same project: thoughtful planning, working software, deeper research, and clear documentation.

MAH connects what each Agent contributes, so ideas become improvements and new findings guide the next round of Evolution.

One Project. Shared Context.

Plan

Claude Code

Work through the approach

Build

OpenAI Codex

Implement and verify changes

Research & Docs

Google Antigravity

Turn sources into clear knowledge

An example team. Match each Agent’s strengths to your project.
  • OpenAI Codex
  • Claude Code
  • Google Antigravity
  • Cursor Agent
  • GitHub Copilot

Agent accounts, subscriptions, and provider limits are separate from MAH.

Think with MAH

Find the Direction. Together.

You do not need to arrive with every answer. Bring an idea, a goal, or a problem. MAH explores possibilities, spots opportunities, and proposes where to go next.

Shape the direction together, then let Evolution move it forward. As the project develops, new findings can lead to better ideas. You stay involved in the choices that matter.

From an idea to a direction

You · An ambition

“Help more people get value from this product.”

MAH · A proposal

“Let’s explore where new users get stuck. A simpler first experience could be our next improvement.”

Together · The next direction

Understand the friction. Improve the experience. Learn from the result.

Illustrative conversation. New directions can start with you or MAH.

A stable division of work

One System, Three Clear Responsibilities

Perspective and judgment

You

Bring your goals and perspective. Explore proposals with MAH, shape the boundaries, and make the decisions that need you.

Proposals and progress

MAH

Discover opportunities, propose directions, and turn shared goals into ongoing work. Keep the project moving and bring back results to learn from.

Execution and checking

AI Agents

Research, analyze, write, implement, operate, and verify through the providers configured for the project.

What MAH keeps connected

The Work Survives the Chat Window

Keep files and instructions close to the work. Use task history to follow progress, add follow-ups, and inspect results without reconstructing the project for every step.

A local project connects files and notes, rules and skills, task history, and results for review by you and your configured Agents.

Files and notes · 01

Give the work a starting point.

Keep the brief, source material, and working files with the project. Your next task can refer back to them.

  • The project brief
  • Research and source notes
  • The working draft
Project context illustration. Connected Agent services process the information sent to them under their own terms.

A Project That Remembers

Keep the original outcome, local files, project knowledge, task history, and follow-ups connected instead of rebuilding context in each tool.

Different Agents, Connected Work

Combine different Agents for planning, execution, and review. MAH keeps their work connected to the same project without making you coordinate every handoff.

Keep Going Through Interruptions

Keep work going beyond a single conversation. MAH tracks progress, handles recoverable interruptions, and shows you what needs attention.

See What Changed and Why

See the results, the checks behind them, and what MAH suggests next. Open the details when you want to understand how it got there.

A clear starting point

What You Need to Start

Apple Silicon Mac

Development builds are intended for testing. Download availability follows the current release policy. macOS 12.0 or later.

MAH Account

Sign in to connect the app, package resources, account machines, and any shared team seats to one account.

At Least One Agent

Configure a supported Agent integration on the Mac. Provider accounts, subscriptions, and limits remain separate from your MAH package.

A Project Folder

Choose the local folder that contains the work. MAH keeps the outcome, tasks, knowledge, and review trail attached to it.

Supported Agent Integrations

OpenAI CodexClaude CodeGoogle AntigravityCursor AgentGitHub Copilot

Configure at least one supported Agent to begin. Provider subscriptions and usage limits are separate from MAH; you do not need to connect every integration.

The ongoing work loop

From an Idea to Ongoing Progress

01

Explore the Direction Together

Bring an idea, a goal, or a problem. MAH proposes possibilities; together you shape what to pursue and the boundaries for the work.

02

Let MAH Plan and Route

MAH turns the direction into steps and coordinates your configured Agents across participating machines.

03

Keep Work Moving

Agents carry out the work and check the results. MAH keeps the context, follows progress, and handles recoverable interruptions.

04

Review, Decide, and Continue

Use the results and MAH’s suggestions to refine what comes next. Evolution continues, with you involved where your judgment matters.

Your MAH workspace

One App. Three Connected Views.

Use Workbench for hands-on project work, Evolution for ongoing improvement, and Fleet to connect the people and machines behind it.

Daily project work

MAH Workbench

Sign in, connect Agents, open projects, run tasks, schedule follow-up work, and review results from the desktop project center.

Automatic project improvement

MAH Evolution

Run continuous rounds of discovery, implementation, review, refinement, maintenance, and repair across participating machines.

Account and team operations

MAH Fleet

Manage team members and machines, and see which machines are advancing your shared projects.

Human Control Stays Visible

Stay in Control of the Work

See what changed, why it matters, and what MAH suggests next. Explore the reasoning, refine the direction, or pause the work whenever you need to. The details remain available for a closer look.

Desktop-First Data Boundary

Project Work Stays Anchored on Your Mac

MAH coordinates around the local project and the Agents configured on the machine. Information you choose to send to a third-party Agent is handled under that provider's terms. See the current service and account boundaries before you start.

Review Security and Data Boundaries

Simple package choice

Same Capabilities. Choose the Resources You Need.

Every package includes the same MAH capabilities. Packages differ only by seats, account machines, and monthly Agent time.

Start with Free. Add resources as your project grows.

Development Environment

Start Your Next Project with MAH

Install MAH, connect your AI Agents, and start your first project.

Download MAH for macOS

Apple silicon · macOS 12.0 or later