Introducing Garagent

I started building Garagent with a simple idea: what if working with AI agents felt more like managing a small team?

I wanted a place where each agent had a clear role, the tools it needed, and a team to work with. That idea became a visual office, with robots representing agents and rooms organizing their work.

You can create an Engineering room, build a Marketing team, or start with one agent helping you with everyday tasks. Each agent has a name, a role, and an AI model you choose. Click a robot to talk through an idea, assign work, or check its progress.

From a conversation to a project

You can work directly with one agent or give a project to a lead agent and its teammates. For example, a content project might involve research, a first draft, and a review. You choose the agents and their capabilities, then follow the work as it progresses.

The office gives that work a familiar setting. Your agents become a team you can organize and return to, with responsibilities you define.

What you can do today

Garagent already lets you:

• Create agents with different roles and AI models.
• Choose the tools and capabilities available to each agent.
• Chat with agents and assign individual tasks.
• Give projects to a lead agent and its teammates.
• Save reusable skills and review suggested memories before approving them.
• Schedule tasks and projects.
• Track token usage and costs.

Help shape the next version

Garagent is currently in invite-only beta. I’m opening it to a small group of testers to learn which workflows are useful, where things get confusing, and what needs improving.

If you have a task you’d like to try with an AI team, request beta access. Already invited? Sign in.

You can also reach me at hello@getgaragent.com. I’d like to hear what you want your agents to help you accomplish.