Why I Make My Agents Fight

The “agentic future” sounds like something a futurist says before showing you a slide with a spaceship on it.

Except… it’s already here. And I’m using it right now to save days of work.

Agent is an abused term. It’s just a small software program that has a mission, a set of knowledge, and tools.

A GPT that you can make in ChatGPT is a kind of agent. So is a project in Claude. They are just prompts that are saved so you don’t have to re-write them every time.

But what I saw last week showed me how much more an agent can do… right now and with no coding needed (at least by me).

Now I don’t have just one GPT. I have an army of them that can work together:

Business Strategy Agent, Research Agent, Finance Agent, Sales Agent… even a Skeptical Owner Agent whose sole purpose is to call BS on everyone else. On Thursday I built Alan Agent who has access to all my past newsletter posts.

Again I wrote zero lines of code. I just asked for what I wanted.

And here’s the fun part:

I make them fight.

Skeptical Owner Agent tells Operations Agent that the delivery plan is a fantasy.

Research Agent goes and finds new ways to do operations.

Alan Agent says that there are only 6 numbers that matter in business while Finance Agent wants to cover the breadth of finance frameworks out there.

The Orchestrator Agent forces them to defend every point until only the best ideas survive.

This is how I’m building the content for the Owner Rx tool.

It’s the result of hours of digital infighting and I’ve got something that might have taken a life time to build.

Here’s an actual discussion they had that I extracted afterwards. Again, I was just sitting there watching this go by at lightening speed:

WTF?

Where I learned about this:

A consultant I talked to last week starts every new project with organizational assessments. The old way? He’d spend days crunching survey data, then even more days building a project plan to address the gaps. That’s a full week before the client saw anything.

Now? He’s got agents:

• Assessment Agent: Analyzes the data and identifies the key gaps in the team’s knowledge.

• Project Plan Agent: Pulls winning approaches from his past work and tailors them to the gaps Assessment Agent identified.

• Deep Research Agent: Searches the web to plug any knowledge gaps in the project plan.

• Tools Agent: Matches the right resources to each step from the consultant’s library of methodologies.

• Orchestrator Agent: sits on top and makes sure each of the others does its job.

They collaborate, fact-check each other, and deliver a complete plan in 10-15 minutes.

Better quality in a fraction of the time and with zero drudgery.

This isn’t some vague “future of work” theory.

This is small business in the agentic future — and it’s available to you right now.

The How

We’re doing this with a simple tool anyone can learn to use: Claude Code. No one uses it because it looks scary. You need to use the Command Line Interface in the Terminal on your computer.

The first time I tried it, I thought, “Forget this.” But I had friend show me how it worked and within an hour I was flying. Now I’m an addict.

These tools are powerful and they are very accessible. You just need to learn.

That’s why I built AI for Small Business Owners: Going from 0-60 in 8 weeks — a focused mastermind sprint that takes you from zero practical AI knowledge to running your own specialized agent teams in just weeks. I’m offering a 30% discount to those who sign up today.

No fluff. No theory. Just the exact playbook I’m using to put the agentic future to work today.

Seats are limited, because we’ll be building your agents together.

Join here and turn AI hype into your unfair advantage.