AI made real, proven by real-life use cases, is what our Innovation Day events are all about. It’s a day where we share expert insights and hands-on demos of what is already being built, deployed and learned in practice.

However, at his opening keynote, David Billings (CSO at Empathy Lab) asked our London audience to lift its head above familiar conversations about AI implementation. Not because those use cases don’t matter, but because they sit on top of a much deeper shift that is already underway.

Most businesses still focus on the application layer: tools, copilots, assistants, and automation opportunities. That framing is understandable... yet incomplete. Beneath it, something more fundamental is changing: the digital infrastructure we all rely on is being rebuilt for a world where machines are not just tools, but primary actors.

When machines take the lead

David illustrated this with a story that stayed with the room. 

When the technical foundation of the web was written in 1991, the HTTP protocol included a status code called 402 Payment Required. It was never used. The authors reserved it for a future they could barely imagine: one in which machines might need to pay each other directly, without a human in the loop.

For over thirty years, that line sat dormant in the specification. Last year, it was activated. A coalition including Cloudflare and Coinbase launched the 402 initiative, backed by companies such as Google, Microsoft, AWS, Visa and Stripe, to build a payment layer for autonomous agents transacting on our behalf.

A placeholder written decades ago has quietly become live infrastructure. Not as an experiment, but as something intended to scale. That detail matters because it reframes the moment we are in.

“This is not just another cycle of digital tools being layered onto existing systems. The systems themselves are changing.”
David Billings, CSO Empathy Lab

When the shift is structural, intent matters

Every major technology wave claims to be unprecedented. Most are not. This one, David argued, genuinely is. Previous industrial revolutions optimized physical labor. The digital revolution organized information and built the infrastructure modern businesses run on. AI is different. It is the first technology designed, explicitly and from the ground up, to automate human intellectual labor.

That speed is not accidental. It is driven by unprecedented, debt‑funded investment in data centers, compute and model development by the world’s largest technology companies. Those investments now need returns. The incentives to move fast are structural.

At the same time, the context into which AI is arriving is more nuanced than the usual dystopia‑versus‑utopia framing. Ageing populations, rising labor costs, stagnant productivity and spiraling public debt are problems for which governments have few credible answers. These are exactly the kinds of structural challenges this technology could help address.

Whether it does so constructively is not predetermined. Big Tech will shape outcomes. Governments will too. But neither group is the main interface between this technology and the real economy. That role sits much closer to home: with operators, technologists and business leaders making everyday decisions inside organizations.

“The choices we make about how this technology is deployed will leave a legacy that goes well beyond the next earnings cycle.”
David Billings, CSO Empathy Lab

Slowing down is not a neutral response. In a global economy, it has consequences. Capability gaps become competitive gaps, and competitive gaps turn into acquisition targets. Europe’s history of building world-class technology only to see it scaled and monetized elsewhere is already well documented.

The question is not whether organizations will adopt AI. It is how well they will do it.

Beyond use cases: from pilots to orchestration

This is where David challenged one of the most common starting points for AI initiatives: the use case mindset. Focusing narrowly on individual use cases feels pragmatic. It creates momentum. It appears low risk. But on its own, it caps value. Left unchecked, it fragments effort and turns AI into a collection of disconnected experiments that never compound.

David was equally clear about another trap: reducing the conversation to headcount replacement. Automation is real, and it would be naïve to pretend otherwise. But when AI ambition collapses into a workforce debate alone, organizations solve the wrong problem and leave the majority of value untouched.

“If the sum total of your AI ambition is a list of use cases or a headcount discussion, you’re solving the wrong problem – and leaving most of the value on the table.”
David Billings, CSO Empathy Lab

The real advantage comes from solving the coordination problems at the heart of complex businesses. You get the competitive edge from building systems where intelligence flows across functions, where capabilities reinforce each other, and where decision‑making improves because insights are connected rather than siloed.

This is the shift from pilots to production. From proof of concept to enterprise reality. And it is, as David noted, far harder than building any single pilot in isolation.

It is also the logic behind what we at Empathy Lab describe as a Growth Operating System. It’s not a single platform or tool, but an orchestrated way of connecting data, intelligence, teams and execution. The result is compounding value over time, not a reset with every new initiative.

If AI is becoming infrastructural, then growth can no longer depend on isolated moments of brilliance. It depends on systems that allow organizations to learn, adapt and scale coherently.

Execution is the differentiator

David made it clear that this moment doesn't call for panic or caution. Instead, he set the audience a challenge. AI is already reshaping markets. The differentiator will not be vision decks, experimentation theatre or the number of pilots an organization can point to. It will be execution.

It comes down to whether AI is embedded in how value is actually created: connected across teams, capabilities and systems, and able to compound over time. In the coming years, the gap will widen between organizations that can turn AI into real, sustained impact and those that can't.

Do you need help in building this kind of executional backbone? Then download our Growth OS white paper, where we unpack how orchestration replaces siloed optimization.