Anyone remember the Thigh Master?

They sold 15 million of those things. Fifteen million! Why? Because they weren't selling a thigh exerciser. They were selling a shortcut to a promise. Buy this. You will look like this.

That's what I kept thinking when I saw the news about OpenAI and Anthropic launching massive new consulting companies to help enterprises "adopt" AI.

Except it's not adoption. It's commoditization.

I'm not saying it won't be helpful. It will! But it's like how in the late '90s you had to get your company online. You needed a website. You needed email. You needed somebody's nephew who "knew computers" to set it all up.

Having a website didn't mean you were going to be Amazon. It meant you were keeping up. Table stakes.

That's what this is. Table stakes dressed up in a $5.5 billion tuxedo.

Because here's the thing nobody is talking about: There are Two AIs.

The first is embedding AI into your systems. That drives productivity. The second is changing how your people think and work. That drives innovation. This news only addresses the first. And companies will mistakenly think it addresses both.

I know this with every fiber of my being. Because we want the easy answer. We want to get in shape, so we hire somebody to build us the best gym possible.

Say it with me: That doesn't change behavior.

WHAT HAPPENED

OpenAI launched an entirely new company - the OpenAI Deployment Company, "DeployCo" - with $4 billion in backing from TPG, Goldman Sachs, SoftBank, Bain Capital. McKinsey and Capgemini signed on as consulting partners. They even bought a whole consulting firm called Tomoro just to get 150 engineers on day one. Nothing says "we're in a hurry" like acquiring an entire company for its employees.

A week earlier, Anthropic announced its own $1.5 billion version backed by Blackstone, Goldman Sachs, and Hellman & Friedman. Blackstone alone owns hundreds of portfolio companies. That's not a customer pipeline. That's a customer fire hose.

The pitch from both: embed engineers directly inside your company to wire AI into your infrastructure and workflows.

Companies need this! But this is the Thigh Master. It feels like you're doing the thing. But you're not doing the thing.

THE FIRST AI vs. THE SECOND AI

Here's what's about to happen.

Companies are going to write enormous checks. They're going to get AI beautifully installed into every corner of their operation. Dashboards gleaming. Chef's kiss.

And then six months later some CEO is going to be standing in front of the board like "So...why is everyone still using this thing to summarize emails?"

Because you built the world's most incredible gym. You flew in architects from Milan. You got equipment designed by NASA. The towels are woven from the chest hair of Greek gods.

And then nobody exercised.

THAT'S NOT A GYM PROBLEM, FRIENDS. THAT IS A BEHAVIOR PROBLEM.

Goldman Sachs' Marc Nachmann - one of the Anthropic venture partners - told CNBC: "Having the model alone doesn't change your workflows or how you operate. You need people who can combine the technology with what's actually happening in the business."

He's SO close. He's like a guy who says "The problem isn't owning a treadmill - it's that you need someone to show you how the buttons work."

No, Marc! (Can I call you Marc? I'm calling you Marc.)

The problem isn't the buttons. The problem is that it's 6am and your bed is warm and your brain is whispering sweet nothings about going back to sleep and Doritos are in the pantry. That's a behavioral change problem. No amount of Forward Deployed Engineers - great name, by the way, sounds like a Navy SEAL unit for nerds - will fix that.

WE HAVE ACTUAL DATA ON THIS

Microsoft Research ran a field experiment with us at Gap Inc. 388 employees. Controlled study. Published on arXiv.

Employees who got cognitive training produced work rated twice as high as those who got standard feature training. Same tools. Same access. Different thinking.

The bottleneck was never the technology. It never is.

WHAT TO DO ABOUT IT

Both OpenAI and Anthropic are sprinting toward IPOs this year. They need fat enterprise revenue numbers. These ventures are brilliant distribution plays for them.

But for you - the company writing the check - AI infrastructure is going to be a commodity. OpenAI's plumbing will look like Anthropic's plumbing will look like Google's plumbing. The tools converge. That's what tools do.

What won't be a commodity? How your people use it.

So get DeployCo. Get your tech stack wired up. Absolutely.

But invest just as hard - harder, honestly - in how your people think and work with AI. That's where the actual differentiation lives.

Because the companies that win won't be the ones with the best AI plumbing. They'll be the ones whose people know what to do with it.

Don't be the company that bought 15 million Thigh Masters and wondered why nobody looks like Suzanne Somers.

We got this, friends.

AI News of the Week

SAP Unveils the "Autonomous Enterprise" at Sapphire

SAP just dropped its Autonomous Enterprise platform this week. They’re touting 50+ Joule Assistants, partnerships with Anthropic, AWS, Google, Microsoft, NVIDIA and Palantir, plus a €100 million fund to push it all into customer hands.

MIT Open Learning Launches "Universal AI"

MIT quietly launched something called Universal AI this week. It’s a free, modular program designed to take anyone from "no idea" to AI fluency. Notice the word. Not "access." Not "tools." Fluency. The vice provost specifically called it "a pathway to AI fluency that's approachable to anyone, anywhere." Ever heard anyone say something like that before?

Google Pushes Gemini to the Center of Android

Google is shoving Gemini into Android as a full operating layer this week. Think app automation, multi-step task execution, Android Auto rebuilt around it. Two hundred and fifty million cars are about to have Gemini in the dashboard. Huge.