In the fall of 2022, I sat down with my co-founders to discuss what we wanted to achieve by building Cogno. Our goal was simple: gain the expertise to build and scale so we’d be ready when we faced an opportunity truly worth pursuing. We wanted to learn by doing, to understand the mechanics of creation before the next big wave hit.
Two months later, ChatGPT launched.
After getting my hands on it during Thanksgiving break, I knew it was going to mark a new era of what’s possible. The interface was crude, the responses inconsistent, but the glimpse of capability was unmistakable. Building AI agents suddenly shifted from distant science fiction to immediate possibility—a combination of imagination and rapid prototyping that investors found genuinely interesting.
We were at the right timing, or so it seemed.
The Momentum Lesson
A year later, we had built a functioning system with real progress. Yet marketing to SMBs for tech adoption proved to be more of an ordeal than building an intricate system. The technical challenges—the ones we loved solving—turned out to be the easy part. The hard part was convincing small businesses to change how they worked, to trust a new system, to invest time in learning something different.
It was during this exact period that Turbo.ai started to take off. I remember thinking our product had more technical depth, that we’d built something more defensible with better architecture. I was focused on the elegance of the solution rather than the urgency of the problem it solved.
Turns out, I was more than wrong.
In this new age of AI, momentum is all that matters. Technical depth, while valuable, takes a backseat to speed, market positioning, and the ability to capitalize on windows of opportunity before they close. Turbo understood something we missed: in rapidly evolving markets, being first often trumps being best.
The Commercialization Reality
My subsequent experience at ByteDance and TikTok reinforced another uncomfortable truth: commercialization of AI remains a far stretch for many consumer products. Working on Symphony products at TikTok, I witnessed the gap between AI capability and real commercial value. The technology was impressive, but translating that into sustainable business models—especially in consumer contexts—proved consistently challenging.
The enterprise world moves differently. B2B customers will pay for clear ROI and efficiency gains. But consumer applications require something deeper: they need to become habits, to solve problems people didn’t know they had, to create new behaviors rather than just optimize existing ones.
The Funding Signal
Now, a year later, as funding rounds get bigger and bigger, I’m starting to wonder if we’re witnessing preparation for a funding winter. When capital becomes abundant for AI startups, it often signals that investors are placing their bets before a potential downturn. Companies are sealing their positions at the table, raising large rounds to survive the deeper waters ahead.
There’s a palpable sense of urgency in the ecosystem. Not the good kind of urgency that comes from clear market demand, but the anxious kind that comes from feeling like tickets for opportunities are selling out. Every week brings news of another massive AI round, another player securing their spot in what feels like an increasingly exclusive game.
The Window Frame
This brings me to the concept I’ve been thinking about: the closing window frame of AI applications.
In the early days of any transformative technology, there’s a period of experimentation where many approaches seem viable. Multiple technical paths, business models, and market strategies coexist. This is the wide-open window phase—lots of room for different players to find their niche.
But as the technology matures and market dynamics clarify, the window frame starts to close. Successful patterns emerge. Distribution advantages compound. Network effects kick in. Capital requirements increase. What once felt like an open field becomes a defined arena with clear winners and everybody else.
I believe we’re watching this transition happen in real-time with AI applications. The experimental phase is ending. The consolidation phase is beginning.
The Agentic Opportunity
A generation of agentic startups is taking shape. Werdelin believes we’re still in the earliest days of figuring out what “agentic businesses”—companies built around AI agents—will actually look like. He compares it to the early 2000s when he was building an internet video startup before YouTube came around. Back then no one knew how to answer simple questions like: Should videos autoplay when you open a link, or require a click? Should the next one start automatically? As he builds Audos, Werdelin is realizing that many of these basic features simply don’t exist yet for generative AI.
This observation reinforces the urgency I feel about timing. We’re not just in the early days of AI adoption—we’re in the early days of understanding what AI-native experiences should even look like. The fundamental interaction patterns, the basic UX conventions, the core product assumptions are all still being established. This represents both massive opportunity and massive risk, depending on how quickly you can move and how well you can read the emerging patterns.
The Urgency I Feel
There’s a specific urgency I feel watching this unfold. It’s not FOMO exactly—it’s more like watching the last few seats at a table fill up, knowing that the conversation happening at that table will determine the next decade of technological progress.
The question isn’t whether AI will transform everything—that’s already decided. The question is who will build the infrastructure, platforms, and applications that define how that transformation happens. And increasingly, it feels like those positions are being claimed right now.
For those of us who’ve been building in this space, who’ve learned the hard lessons about technical depth versus market momentum, who understand both the possibilities and the pitfalls—this feels like the moment where preparation meets opportunity. The window frame is closing, but it’s not closed yet.
The tickets aren’t sold out. But they’re selling fast.