Skip to content
Beta · We're looking for testers

Your AI agent team for every project — in one cockpit.

Agent Manager controls several specialized AI agents. Claude Code and OpenAI Codex as a coordinated team — per client project, instead of from dozens of terminal windows. Locally on your Mac, via your existing subscription.

No API key required; runs through your Claude/OpenAI subscription; no cloud intermediary.

Agent-Manager-Cockpit: Übersicht aller Projekte mit Agenten-Status
The problem

AI agents are powerful — but chaotic in everyday life.

Multiple terminal windows, several client projects running concurrently, constant context switching, all manually coordinated — and ultimately tied to a single provider. The overview? Gone.

The solution

One cockpit for all your agents

Two providers, one interface

Claude Code and OpenAI Codex — selectable per agent, switchable mid-work with context handover. No vendor lock-in.

About your subscription

Login via OAuth, which you perform yourself. No API key, no additional token costs.

Local & data-saving

Runs on your Mac, no cloud intermediary, no trackers. Your customer data stays with you.

Agents as a team

Each project has one coordinator + specialist agents (WEB, MARKETING, SEO …) who collaborate via a shared mailbox.

Built-in automation

Routines (real cron, runs even when the app is closed) and AI-generated, checkable workflows.

Full overview

Portfolio of all projects, live sessions, "waiting for you" status, mailbox — all at a glance.

Here's how it works

Ready to go in five steps

  1. 1

    Create project

    Folder structure, roles & boot prompts are created automatically.

  2. 2

    Choose a team

    Coordinator + specialist agents from the catalog, supplier per project.

  3. 3

    Working in the chat

    Submit orders, attach files, mode & model per agent.

  4. 4

    Coordinate

    Agents coordinate via the mailbox; you keep track of everything.

  5. 5

    Automate

    Check off workflows, let routines run — day after day.

Multi-provider

Claude Code & OpenAI Codex — together

Leverage the strengths of both providers. Configure which tool is used per agent or per project — and switch between them mid-conversation without losing the thread.

  • Provider freely selectable per agent/project
  • Switch with automatic context transfer
  • Both logged in simultaneously, status in view
Kommunikations-Zentrale: Chat mit einem Agenten, Anbieter- und Modellwahl
automation

Workflows & routines that anticipate your needs

Let the AI build a checklist-based workflow from a goal, assign steps to agents — and set up routines that automatically complete recurring tasks, even when the app is closed.

  • AI-powered workflows from a single goal
  • Steps: Assign agents & check off
  • Routines as a true system cron
Routinen-Seite: automatisierte, wiederkehrende Aufgaben pro Agent
Projects & Knowledge

Each project has its own team and memory.

Create projects with matching agent sets, keep rules and reusable knowledge centralized — read by all agents in all projects.

  • Project Templates & Custom Agents
  • Knowledge database & rule sets
  • Export/Import for Backup & Migration
Neues Projekt anlegen mit Agenten-Auswahl und Anbieter
Security & Data Protection

Your data stays with you.

Agent Manager runs locally on your Mac — no cloud intermediary, no trackers, fonts are self-hosted. Ideal for agencies that work with customer data.

Local instead of cloud

The server runs on your computer. There's no third-party intermediary between you and the AI.

About your subscription

Login via OAuth by yourself — the app does not store API keys and never performs a login itself.

No external calls

No CDNs, no trackers, no third-party services in the app. Data-efficient & GDPR-compliant.

For whom

Made for everyone who wants to achieve a lot with AI

Agencies
Many customer projects running in parallel — each with its own team of agents, all in one cockpit.
Freelancer
Work alone, but like a whole team — with multiple specialist agents by your side.
Small Teams & Studios
Handovers and coordination take place via mailboxes, workflows, and shared knowledge.
Consultants & Power Users
Automate recurring processes, use two providers without terminal chaos.
Beta program

Become a tester for Agent Manager

Early access, direct line to the creators, influence on the roadmap. We are looking for agencies, freelancers, and teams working with Claude Code and/or OpenAI Codex on Mac.

Pre-register for the beta now

macOS · active Claude and/or OpenAI subscription · no costs, no payment details.

FAQ

Frequently Asked Questions

Do I need an API key?

No. Agent Manager runs on your existing Claude/OpenAI subscription. You log in once via OAuth — the app doesn't store any keys and doesn't incur any additional token costs.

What are the requirements?

macOS and an active subscription to Claude and/or OpenAI (ChatGPT). Both providers can be used simultaneously.

Where is my data stored?

Everything runs locally on your Mac. There's no cloud intermediary and no external trackers — your project and client data stays with you.

Can I mix Claude and Codex?

Yes. You choose the provider per agent or per project and can switch mid-conversation — the current status is handed over as a handover.

How much does Agent Manager cost?

You can test for free in the beta program. We will inform testers about the terms and conditions after the beta period in due course.