For the past few years, artificial intelligence has been something you could see.

Open ChatGPT and there it was. Type in a question, ask it to write an email, create an image, or summarize a document, and within seconds AI produced something that felt almost magical if not… serviceable.

That was millions of people’s introduction to generative AI, and it helped turn artificial intelligence into one of the biggest investment stories I’ve seen in my career.

But we’re now entering a much more important phase. AI is going to start disappearing.

Not because the technology is going away. Quite the opposite. AI is going to become so deeply embedded inside the products and services we already use that eventually we won’t even think about it as AI.

That’s when this revolution really gets interesting.

After the Magic Trick

Corporate America has spent the last few years experimenting with AI. Executives launched pilot programs, employees played around with ChatGPT, and virtually every company found some way to mention AI on an earnings call.

Some of it was legitimate. Some of it was probably designed to keep investors happy.

But experimentation was always going to be the first step. The real opportunity begins when companies figure out how to use AI to save money, generate more revenue, or fundamentally improve their products.

We’re beginning to cross that line now.

Think about how you use the internet today. You don’t sit down in the morning and announce, “I’m going to use the internet.” You check your bank account, order something from Amazon, watch Netflix, request an Uber, and read your email.

The internet disappeared into the products.

AI is headed in the same direction.

Your Definition of ‘AI Stock’ Is Too Small

And that’s why investors need to expand their definition of an “AI stock.”

Yes, Nvidia (NVDA) remains incredibly important. So do the giant cloud providers and companies building the leading AI models. But the next stage of the boom will increasingly be about companies taking that intelligence and applying it to industries that already exist.

We’re beginning to see examples everywhere.

Duolingo (DUOL) is using generative AI to create educational content and give language learners the ability to practice conversations with an AI rather than another human. One of the biggest challenges in education has always been personalization. A teacher can’t give 30 students 30 completely different lessons at the same time. AI increasingly can.

Then there’s a company most investors probably don’t associate with AI at all.

CCC Intelligent Solutions (CCC) operates behind the scenes of the auto-insurance and collision-repair industry. Its technology connects insurers, repair shops, automakers, and other participants in the claims process. Now it’s using AI to analyze photos of damaged vehicles, help generate repair estimates, and automate portions of the claims process. One large collision-repair customer is already initiating 98% of its estimates through CCC’s AI-powered technology.

And consider Procore Technologies (PCOR), which provides software to the construction industry. Procore is introducing what it calls “Digital Coworkers” – AI agents designed to help construction companies review contracts, search project documents, manage information requests, and handle some of the administrative work surrounding massive construction projects.

Language learning. Car accidents. Construction sites.

Those may sound like three completely unrelated businesses. But that’s exactly the point.

A Coworker, Not a Pink Slip

There’s another misconception about AI that I think is going to disappear over the next several years: The technology only creates value if it replaces human workers.

I don’t believe that’s how most companies will initially use it.

A construction company doesn’t necessarily want to eliminate its project manager. It wants that project manager spending less time searching through hundreds of documents. An insurance company doesn’t necessarily want to eliminate everyone involved in a claim. It wants to process that claim faster and more accurately.

And a student learning Spanish doesn’t necessarily need AI to replace a teacher. He might simply want someone (or something) available at 10 p.m. to practice a conversation.

Multiply those productivity improvements across millions of workers and thousands of businesses, and the economic impact becomes enormous.

That’s why the next phase of AI adoption could look very different from the first.

Phase Two Pays the Bills

The first phase of the AI boom was about creating the technology. We needed GPUs, enormous data centers, massive language models, and billions of dollars of investment to make modern AI possible.

Now we’re moving into what I call the Integration Phase.

This is when AI begins working its way into healthcare, finance, manufacturing, insurance, construction, education, transportation, and virtually every other major industry. Some applications will be revolutionary. Others will be almost boring – automating paperwork, answering routine questions, or saving employees 20 minutes on a task.

But boring can make a lot of money.

That’s what I’ll increasingly be watching as this AI boom matures. I want to find companies that already have customers, valuable data, and expertise in a particular industry, and can use AI to make their existing businesses dramatically better.

Eventually, I think we’ll stop talking about “AI companies” altogether. We don’t call Amazon (AMZN) an internet company anymore. The internet simply became part of how modern business operates.

AI is headed there fast.

And if I’m right, the biggest opportunity won’t simply be figuring out who builds the best AI.

It’ll be figuring out who puts it to work.

P.S. Here’s the twist to everything I just told you…

Every company putting AI to work is paying someone for the brains behind it, and a huge share of that money flows into just two companies, OpenAI and Anthropic. Neither is public yet, but Anthropic is reportedly eyeing a listing before Thanksgiving at up to $2 trillion.

In my new presentation, I show you how to get in ahead of the IPO through already-public companies holding big stakes in these two AI giants. Watch it here.