Let me tell you three stories.

The Next Revolution: Three Stories About Building in the Storm of Progress

First Story: Standing on Shifting Ground

When you stand at the beginning—bright-eyed, code humming, vision clear—you see gaps in the world. You build bridges to cross them.

But here’s the thing no one tells you: the ground itself is shifting.

I’ll never forget when I held the iPhone 3g in my hand. It wasn’t just a phone—it was a promise. Emails, music, web browsing, effortlessly nestled in my pocket. It felt magical, almost impossible.

Yet today, life without it is unimaginable.

We are standing at the brink of another such moment.

AI doesn’t wait. What took years in the old world now happens in months. Foundation models, like Claude and ChatGPT, aren’t merely improving—they’re traversing capability thresholds at exponential rates. That startup coding furiously to solve yesterday’s problem? They’re like sculptors carving ice under a rising sun, watching their masterpiece dissolve into puddles of irrelevance.

This isn’t mere change; it’s a technological revolution in real-time.

By month 22, what was once a chasm becomes a hairline fracture in the substrate—and GPT-5’s attention mechanisms stride over it without even a downsampling hiccup.

Where are the visionaries?

Second Story: Love, Loss, and Letting Go

Listening to Andrej Karpathy, you can see it at Tesla and OpenAI—the gravitational pull of trusted architectures, beloved activation functions, and established tools.

Teams cling to architectures like life rafts. They optimise for performance on yesterday’s benchmarks, not seeing the tsunami of scale approaching on the horizon.

Imagine this: In 2023, a company passionately builds an AI solution to painstakingly extract structured data from PDFs, cleverly fine-tuning GPT-4 parameters for eighteen months. Suddenly Claude 4 emerges in 2025—deciphering complex tables and diagrams intuitively, rendering their hard-earned innovation obsolete overnight.

This isn’t failure; it’s the inevitable physics of exponential progress.

When the foundations of technology shift this dramatically, solutions either ride the wave or vanish beneath it.

Where are the visionaries?

Third Story: Death and Rebirth

Every startup faces two potential deaths: running out of cash or becoming irrelevant. Yet visionary leaders embrace a third possibility—rebirth.

Karpathy’s playbook whispers loudly, a fundamental truth:

Build not for today’s model, but tomorrow’s computational substrate.

That is the vibe. Those that survive aren’t optimising against current capability horizons. They’re constructing entirely new possibility spaces. Pause and think about this. Survive? What does it take to thrive!?

Lambda doesn’t merely sell GPUs; they supply oxygen for AI’s evolutionary fire.

Perplexity doesn’t simply index data; they distil latent knowledge from vast information spaces.

These aren’t features—they’re foundational primitives. Bedrock beneath ever-shifting sands, the atomic units that remain valuable regardless of which models dominate next quarter.

Where are the visionaries?

Stay Hungry. Stay Flexible.

The lesson isn’t grim—it’s liberation.

This is not just for product builders—THIS is for enterprise leaders implementing products.

• Are you designing for obsolescence? Just as Karpathy designed Tesla’s neural networks to silently evolve overnight, bake continuous reinvention deep within your organisation’s DNA. Create systems with modular interfaces designed to absorb new capabilities rather than be rendered obsolete by them.
• Have you identified your atomic unit? Do you even own the atomic piece? Pinpoint irreducible values—unique datasets, data moats, irreplaceable human interaction graphs, tangible physical-world hooks—that no AI advancement can render obsolete. Not even an API update can erase. What persists when the current generation of models becomes commoditised?
• Are you becoming the platform? Become the platform. Don’t merely enhance existing functions like “document parsing”; become the layer upon which future AI-driven innovations unfold. Create the document understanding layer for the LLM operating system. Position at the nexus of data flows, not at the endpoints.

When the ground shifts daily, survival belongs to those who see beyond the immediate horizon. They anticipate tectonic changes. They build the future. The great builders—like Steve Jobs’ teams that turned computers into pockets—don’t chase gaps.

They create tectonic plates.

Where are the visionaries? Steve Jobs didn’t chase momentary opportunities; he looked beyond the visible horizon, capturing entirely new possibilities.

One More Thing…

That PDF startup I mentioned? Their path forward isn’t incremental refinement. Instead, they become the Rosetta Stone for multimodal reasoning. Teaching Claude 4 how to decode the visual semantics of ancient tables and diagrams. Moving from filling gaps to expanding frontiers.

Consider Stability AI, pioneering open-source models that democratise AI’s potential, empowering thousands of enterprises to customise, adapt, and scale innovation on their own terms.

This is how legends emerge: not by solving today’s puzzles, but by redrawing the board. That is where the visionaries are found.

The question isn’t whether your AI strategy will be obsolete in 18 months.

It will be.

The question is whether you’re building a company that thrives precisely because of that obsolescence.

Enterprise—Your Moment Is Now

Enterprise leaders stand at an inflection point comparable to the dawn of the internet—perhaps even more transformative. The weight on their shoulders is immense. Autonomous AI agents promise more than mere automation—they promise to unlock unprecedented human creativity and redefine purpose.

Yet here’s a critical insight many organisations overlook:

• Is your data infrastructure truly ready? Data readiness isn’t optional; it’s foundational.
• How are you addressing human adoption? AI success isn’t automatic; it’s cultural.
• Is your “build versus buy” strategy flexible enough? Today’s strategic choices are tomorrow’s competitive edges.
• Are your interfaces truly human-centred? Technology is nothing without profoundly human experiences.

Companies dominating post-iPhone weren’t those digitising existing services—they were the pioneers who imagined entirely new worlds.

Now, this is your chance—your moment—to define the AI future:

To orchestrate, not just automate.

To lead, not just follow.

To create, rather than react.

Don’t wait—the future won’t.

Will you seize this moment?