A few weeks ago, I found myself sketching ideas in my notebook, trying to make sense of where businesses fit into the whole Generative AI (GenAI) wave.
Everywhere I looked, companies were either talking about AI, launching something with AI, or worrying about being left behind. But it didn’t feel random — there was a pattern to it.

As I kept thinking, it became clear that businesses were lining up along a kind of spectrum. On one end, there are the pioneers, like OpenAI and DeepMind, building the foundational models that power everything else. They’re not really making products for everyday people — they’re building the engines that others use.

Then there’s another group taking these models and fine-tuning them for specific industries. Most of the time, they’re not working with the biggest AI models. Instead, they’re using smaller, more practical ones — often called SLMs (SLMs — Small Language Models — are growing in popularity because they offer faster performance, lower costs, and are easier to fine-tune for specific tasks compared to massive LLMs) — because they’re faster, cheaper, and easier to adapt. It’s not about outbuilding OpenAI; it’s about making AI useful for doctors, lawyers, bankers, and others who need real-world solutions that fit their work.

A little further along the spectrum are companies that don’t tweak the models much at all — they just integrate them into apps, websites, and software. They take what’s already available and make it part of the user experience. It’s a different kind of innovation, one that’s closer to the end user.

And then, on a different track, there are the businesses building the hardware — the headsets, cameras, and devices that make all these AI experiences smoother and more natural. They’re not coding AI models, but without them, a lot of AI-driven communication wouldn’t feel nearly as human.

When I looked at it this way, I realized something important: you don’t have to be a foundational AI company to have a huge impact.
In fact, some of the most exciting opportunities are for businesses that find clever ways to customize and apply AI in specific areas.

Building a new AI model from scratch is expensive, slow, and risky. But fine-tuning existing models — making them smarter for a particular industry, or even a particular type of customer — is within reach for a lot more companies. It’s faster, cheaper, and honestly, probably a lot more useful for real people.

In my opinion, the real winners over the next few years will be the ones who don’t just chase AI for the sake of it. They’ll be the ones who blend great user experiences, smart technology, and thoughtful specialization.
They’ll find ways to make AI feel natural, personal, and perfectly suited to the people using it.

That’s the kind of future I’m excited about — and the kind of work I hope to be part of.