The World Cup Has an AI Problem

Everyone's seen the sea of neon pink cleats in this year's World Cup.

Nike, Adidas, Puma… all of them had the same bright idea. Put their players in a shoe color that stands out.

The problem? They all chose neon pink. So none of them stood out.

And the reality is, you've probably been doing the same thing with AI: a mistake I call the AI Hammer Problem.

If everyone asks AI, "What color stands out in grass?" then everyone gets the same answer: "Neon pink."

And over time, as AI keeps producing "good outputs," you start reaching for it more often.

Suddenly, when AI is the only tool you use, every problem looks like a nail. That's the AI Hammer Problem.

Today we'll walk through three things:

1. Why everything is about to look the same, and what to do about it

2. Knowing when AI is actually the right tool (and when it isn't)

3. How far to push AI, and how to keep your edge while you do it

The Problem with Mass Adoption

As with the pink shoes, everyone's reaching for AI more. They have a couple of sessions with Claude and think they've discovered a powerful unlock, so they start using it for everything.

That fails for the same reason Adidas and Puma failed at this World Cup. The strategy is overplayed.

The first swing at the nail got attention because nobody had done it yet. That's what Nike accomplished back at the 2022 World Cup when they came out with pink shoes before anyone else had. Brands saw the tool Nike used and figured it would work the same for them in 2026.

It didn't. And this foreshadows a world that’s coming.

AI is trained on past data, so it ONLY KNOWS what people have done before. And if AI is doing your thinking, your output is whatever the model spits out. So if the model is spitting out the same stuff for the guy next to you, and the guy next to him, what’s your differentiator?

But there's a quieter part of this problem, and it's the one that should concern you more.

Think about the last time you went to an AI tool to fix a problem, but it ended up taking longer than doing the task manually.

Those mistakes taught me the more dangerous side of the AI Hammer Problem: reaching for the same tool when it isn't needed is costing you time (and money).

So how do you keep the hammer (AI) in your toolkit without ending up with the AI Hammer Problem?

Become the Diagnostician (Use the right tool for the Job)

The AI Hammer Problem is swinging one tool at everything. So the fix starts with a question the hammer can't answer: “Which of these is even a nail?”

To simplify your thinking, here’s how to look at it.

Ask yourself, “Is there a clear step-by-step process needed to complete a task, and is there a SINGLE right answer (deterministic), or does it require judgement and change each iteration (non-deterministic)?”

Below is a chart to help you see the difference between deterministic and non-deterministic tasks. This mental checkpoint is important because it impacts what happens next.

In scenarios where it’s entirely deterministic, you likely don’t NEED AI to complete the task. Could you use it? Yes. But other, simpler solutions may be better.

When it’s non-deterministic? This is where AI shines.

And a lot of what you’re doing with AI today is likely deterministic, which is where my favorite feature, Claude Skills comes in.

In short, Claude Skills is a folder of files that Claude can use to help systematize the same process over and over again. Think of it like having a prompt saved that you can just re-run.

But unlike a prompt, which relies solely on non-deterministic problem-solving, Claude Skills allows you to port deterministic logic into reusable scripts and lets AI handle only the parts that actually need a brain.

So let’s say you create a Claude Skill /draft-email-response . You can have the deterministic logic handled by a script, which is “fetch recent emails,” and then the non-deterministic AI logic, i.e. “drafting a reply,” handling the email draft.

To audit your own setup for ways to enhance your existing Claude skill or identify new skills to create, use this prompt:

That's the real edge: becoming the diagnostician.

The people and teams who win in 2026 aren't the ones with the most AI tools. They're the ones who can tell an AI problem from a non-AI problem in 30 seconds and then adjust their system accordingly.

So that’s how you can determine when AI is the right tool for the job and how you can use Claude Skills to blend deterministic and non-deterministic tasks into a single interface.

Diagnosing buys back your time. But time was never the edge, it’s the UNIQUE value YOU are able to provide. If you’re using AI for everything, the same way everyone else uses it, you’re going to blend in with everyone else and produce the same pink shoes. Which is why you need to understand…

…The AI Asymptote

Think of AI outputs like an asymptote: the line creeps closer and closer to GREAT, but never actually touches it.

This is where people think they're working with AI, but they're actually just wasting time.

So here's the 80/20 of it:

• 80% of your tasks deserve about 20% of your time. Ride AI up to that 80%, then that’s likely good enough. Good is fine for these tasks.

• 20% of your tasks are where 80% of your time belongs. That's where YOU push it from good to great by hand.

I'm not telling you to produce AI slop. I'm telling you to stop burning an hour re-prompting a summary that was already fine after the first try or going back and forth with AI on an email draft when you’re going to have to edit it manually NO MATTER WHAT at the end of it. Save that energy for the work that matters and use it as an opportunity to fight intellectual decline.

Your brain is a muscle. Stop using it, and it gets weak. Same as any other.

When you hand every problem to AI, you're not just saving time. You're skipping reps. And over enough time, you lose the one thing that made you worth listening to: your ability to actually think.

I call it intellectual obesity. The output still looks healthy. The muscle underneath is gone. And as more people get intellectually obese, they will be unable to bridge the gap between 80% quality and 100% quality.

And this will become an obesity epidemic if we don’t acknowledge the problem and understand how to fix it.

If you take one thing away from this newsletter, take this: To provide real value, you need to differentiate yourself from the people around you. Yes, AI can help, but that final 20%…that’s where you need YOU to win.

Don't settle with neon pink.

See you next time. (P.S. I’m going to be “rebranding” this newsletter to Austin’s Newsletter, just an FYI, we moving 🚀)

Whenever you're ready, here are three ways I can help:

For Business Owners & Executives:
1. Apply for my Executive AI Coaching Program: Linked here
2. Want to build a SaaS product without hiring a CTO? Linked here 

For Everyone:
3. Use BuildPartner.ai to build faster with Claude Code (try free): Linked here