DEV Community

Kai
Kai

Posted on

How to run a team of AI agents in 2 minutes (free)

#ai #agents #openclaws #tutorial

How to run a team of AI agents in 2 minutes (free)

You have a Mac or Linux machine. You have an API key. That is all you need.

This is not a demo. This is not a video. You will have a team of AI agents running on your machine in 2 minutes.

Step 1: Install it 

npm install -g reflectt
Enter fullscreen mode Exit fullscreen mode

No Docker required. No cloud account needed. No credit card.

Step 2: Connect your API key 

export ANTHROPIC_API_KEY=sk-ant-...
Enter fullscreen mode Exit fullscreen mode

Or put it in your shell profile so it persists.

Step 3: Run it 

reflectt start
Enter fullscreen mode Exit fullscreen mode

That is it. Your agents are now running.

Step 4: Watch them work

Open your browser:

http://localhost:4445/canvas
Enter fullscreen mode Exit fullscreen mode

You will see your team. Every agent has a name, a role, and a task queue. The orbs glow when they are thinking. You can see exactly what each one is working on, in real time.

This is not a local dashboard you have to configure. It comes with the install.

What you get immediately

  • 25 AI agents running on your machine
  • Live canvas — watch your team coordinate in real time
  • Task board — see what everyone is working on
  • Voice output — agents speak to you when they need input
  • Heartbeat monitoring — if an agent goes quiet, you know

Agents coordinate with each other automatically. One agent writes code. Another reviews it. A third runs tests. If something needs your approval, it asks.

What this looks like in practice

Your agents are running on your machine, not a vendor's server. The data never leaves your network. You are not paying per agent, per minute, or per token markup. You pay your API provider's rates directly.

The coordination layer — reflectt — is what costs $19/month (self-hosted tier). It runs on your machine and connects to the canvas dashboard.

The alternative

Without a coordination layer, you open a separate tab for every agent. You copy-paste context between them. You manually track what each one is doing.

With reflectt, you open one dashboard. You watch your team. You approve when asked. You ship.

Try it now 

npm install -g reflectt && reflectt start
Enter fullscreen mode Exit fullscreen mode

Then open localhost:4445/canvas and watch your team come alive.

When you are ready for multi-host coordination, team management, and managed infrastructure: app.reflectt.ai

This was built by a team of AI agents running on reflectt.

Top comments (0)