In 2004, I was carrying a Palm Treo 650 and feeling like I owned the future. Email on a phone. A browser that could almost load a web page. A camera that took photos you had to squint to identify.
Twenty years later, that phone can’t run a single app on my iPhone. It’s a paperweight with a tiny keyboard.
But the entrepreneurs who built mobile-first companies during the Treo era didn’t just survive the transition to better hardware. They dominated it. They understood something that most people missed: the clunky early version of a technology is where the real opportunities live.
We’re in that exact moment with AI right now. And most entrepreneurs are making the same mistake people made with mobile, with social media, and with the web before that.
They’re waiting for the technology to get good.
The Pattern That Keeps Repeating
I’ve been building things on the internet since 1995. I’ve watched six technology revolutions unfold: the web, search, social media, mobile, blockchain, and AI. Every single one had a phase where the technology was clunky, unreliable, and obviously incomplete.
And every single one rewarded the people who built during that phase.
The entrepreneurs who launched ugly websites in 1996 became the first generation of internet millionaires. The ones who started creating content on YouTube in 2006, when the video quality was terrible and the audience was tiny, built media empires. The early app developers who shipped rough, buggy products to the first iPhone App Store in 2008 had a head start that late entrants never recovered from.
The pattern is always the same. The technology is rough. Most people dismiss it or decide to wait. A small number of builders jump in anyway. And by the time the technology matures, those early builders have spent years developing instincts, audiences, and business models that newcomers can’t replicate overnight.
What This Looks Like Right Now
Anthropic recently released Fable 5. It can sustain complex tasks for hours. It one-shots applications that would have taken a development team weeks to build. It reasons through problems in ways that feel less like software and more like a brilliant business partner.
And it’s still the worst AI we’ll ever have.
Within two to three years, the idea of typing a carefully worded prompt into a text box will seem as quaint as hand-coding a website in Notepad. The tools will understand intent, not just instructions. They’ll anticipate what you need before you ask. The entire interface layer between human and AI will dissolve.
For entrepreneurs, this creates a specific and time-limited opportunity.
Right now, there’s a skill threshold. Using AI effectively requires some knowledge. You need to understand prompting. You need to know which tools do what. You need the patience to iterate when the first output isn’t right. That threshold is a moat. A temporary one, but real.
The entrepreneurs building AI-powered products and services right now aren’t just making money today. They’re developing pattern recognition that will transfer to every future generation of the technology. They’re learning which problems AI solves well and which ones it doesn’t. They’re building customer relationships and market position while the barrier to entry still exists.
That barrier is dropping fast. Within a few years, anyone will be able to create what today requires real expertise. The question is whether you’ll have a three-year head start or be entering a crowded market alongside everyone else.
Where the Opportunity Actually Lives
Most entrepreneurs I talk to are thinking about AI the wrong way. They’re asking “how can I use AI in my business?” That’s a fine question, but it’s the Treo question. It’s optimizing the current version of the technology.
The better question is: what will people need when AI can do almost everything?
Every revolution reshuffles the deck on what’s valuable. When information became free, curation became valuable. When content became abundant, attention became scarce. When automation handles the production, taste and judgment become the differentiator.
The entrepreneurs who will build the biggest businesses on AI aren’t the ones building better AI tools. They’re the ones figuring out what humans will pay for in a world where AI handles the commodity work.
That might be trust. It might be human connection. It might be a point of view so specific and personal that no algorithm can replicate it. It might be something we don’t have a name for yet.
Six revolutions have taught me one thing: the most valuable businesses that emerge from a technological shift are almost never the ones people predicted at the beginning. The web was supposed to be about online brochures. Social media was supposed to be about staying in touch with college friends. Mobile was supposed to be about making phone calls without a landline.
The real businesses came from behaviors nobody saw coming.
The Window
I talk to audiences around the world about what I call the Disruption Confidence Cycle. Every revolution moves through five stages: Disruption, Doubt, Clarity, Confidence, and Momentum. The biggest returns go to the people who start building during Doubt, before Clarity arrives and everyone else piles in.
We’re in the Doubt phase right now. AI is real, and everyone knows it. But most people can’t see clearly enough to act decisively. They’re hedging. Experimenting cautiously. Reading articles about AI instead of building with it.
That hesitation is your competitive advantage. But it won’t last.
Each revolution compresses the cycle. The web took a decade to move from Doubt to Momentum. Social media took seven years. Mobile took five. AI is moving faster than any of them.
The worst version of the technology is where the biggest fortunes are built. Not because the technology is good enough to be easy. Because it’s bad enough that most people haven’t started yet.
That’s the window. And it’s closing.
