For more than twenty-five years, I have watched technology cycles oscillate between the “magic” phase and the “utility” phase. From my early days at Nokia during the first mobile explosion to leading digital at easyJet, I’ve seen this pattern repeat: a burst of erratic innovation is followed by a plateau of “Enshittification” as companies prioritise extraction over value, before eventually settling into a quiet transformation where the tech becomes invisible infrastructure. Today, we are witnessing a similar pivot. The “shiny object” phase of Generative AI is hitting the hard reality of the “Rot Economy.”
The Ghost in the Corporate Machine
There is a growing tension in boardrooms across Europe and beyond. On one hand, there is the undeniable urge to automate everything to drive efficiency; on the other, a creeping realization that excessive automation is leading to a structural brittleness. When we automate a flawed process, we don’t fix the process—we simply accelerate the rate at which it fails. We are seeing the rise of what some call “successful mediocrity”: organisations doing AI well enough to satisfy governance requirements, but not well enough to actually transform their competitive position.
The real risk today isn’t that AI fails to work; it’s that it works too well in ways we cannot audit or explain. This “Agentic Era” requires us to move beyond the chatbox. If your product strategy still relies on a “top-of-funnel” traffic model from search engines, you are standing on a floor that is being pulled out from under you. As recent industry reports suggest, publishers and digital businesses are bracing for a massive decline in search referrals because Google and Gemini are becoming “Answer Engines” rather than traffic aggregators.
Systems Velocity Over Model Brilliance
Many CIOs and CDOs are currently trapped in a data governance loop. They assume AI success depends primarily on model quality. While data is critical, the true differentiator is systems velocity. In my experience building connected home divisions at EDF Energy, the challenge was never just the hardware; it was the feedback loops and the human-in-the-loop workflows that allowed us to iterate at speed.
- Platform Infrastructure: Moving away from isolated pilots to integrated agent orchestration.
- Agentic Governance: Shifting from “what the model says” to “how the system acts.”
- Predictability over Intelligence: A CIO doesn’t need the smartest model; they need the most predictable one.
Consider Salesforce, which has faced scrutiny over its “data tolls”—increasing fees for connectors that allow other apps to tap into its ecosystem. This is a classic symptom of the Rot Economy. When platforms start taxing the data flow rather than enabling value creation, they invite disruption. Enlightened leaders aren’t treating AI as a feature; they are treating it as invisible infrastructure, with the same rigour applied to a payment gateway or a cloud server.
The Human Side of the Algorithm
Technical brilliance is easy to hire; product craftsmanship is hard to build. As highlighted in recent reflections on Google’s long-term product lessons, the lever that turns technical potential into real impact is user obsession. It is easy to quote “users first” in a slide deck, but far harder to translate that into productive constraints. We must align our teams around problems, not features.
In the insurance sector, specifically during my time with MORE TH>N and across Eastern Europe, we found that the most sophisticated digital solutions failed if they didn’t respect the human habits of the customer. AI adoption follows the same rule. The real payoff comes when employees apply these tools to solve specific workflow pains, not when they are forced to use a generic corporate “Copilot” because it was part of a bundle.
Building for Resilience, Not Just Efficiency
To survive this cycle, we must reject the “Enshittification” of our products. Efficiency is a trap if it kills the distinctiveness that made your brand valuable in the first place. We are entering a period where proprietary systems of intelligence will be the only sustainable moat. This isn’t just about prompt engineering; it’s about context engineering—defining valuable problems within your unique business workflows that no off-the-shelf model can solve effectively.
Rather than seeking the next “magic” AI moment, focus on the quiet transformation. Audit your current automated processes: are they creating value for the user, or are they just reducing costs for the business? If it’s the latter, you are effectively building technical debt for a future failure. The winners of this era won’t be the ones with the flashiest agents, but those who integrated technology so seamlessly that the user forgets it’s even there. It’s time to stop experimenting in safe playgrounds and start building reliable, responsible systems that actually run the business.
Leave a Reply