Adrian McPhee has built from zero and scaled inside giants, yet his north star has barely moved. In any setting there are a hundred things you could do. The craft, he said, is choosing what you should do – in what order and why. He starts with the customer and works back, setting an ambitious target as if it were already launched, then forcing a debate on risks and assumptions.
“Good product managers are really risk mitigation engineers,” he said. “What’s the riskiest thing here, and how do we test it quickly?” That rhythm clears the noise and lets teams pivot fast when the evidence says they are wrong.
The instinct goes back to a Commodore 64 and the early internet – a mix of problem-solving, curiosity and play that never really left. His first paid build automated a recruitment firm, then a leasing calculator that became business-critical. The lesson landed early: fix the boring pain and value compounds.
Customer-first clarity
Across startups and large enterprises the trap is the same – endless options masquerading as progress. McPhee’s antidote is a working-backwards discipline: write the six-month press release, state why customers should care, list risks, test the biggest one first. “I’ve learned to be ruthless about priorities,” he said. The method is dull on purpose. It imposes order when enthusiasm tries to do everything at once.
He carries that lens into heavily regulated work. Banks want the power of AI yet cannot duck control. “I see LLMs and agents as brilliant but unreliable consultants,” he said.
“You cannot simply unleash them.” He argues for audit trails, immutable logs and human oversight, with proportionality by use case – a sales campaign is not the same as credit risk. The point is to embed AI inside existing risk frameworks so it becomes part of the bank’s DNA. Ignore it and risk irrelevance. Act responsibly and you shape the future.
Startup speed, enterprise discipline
McPhee has moved between worlds and borrows from both. In a startup he will test a generative model that writes code from specifications. In a corporate he will insist on infrastructure-as-code and rigorous security practice. Speed and creativity on one side, structure and resilience on the other – together they give you an edge. The two mindsets also correct each other. Startup zeal can skip controls. Enterprise muscle can smother ideas. Held in tension, they raise the standard of what gets shipped.
Not every call lands. In the mid-2000s he built a neighbourhood platform, declined an acquisition and discovered he was a decade too early. “Being too early can be as fatal as being too late,” he said. He would now partner sooner, add complementary skills and think harder about customer acquisition than code. Timing, model and team – miss one and the rest strain.
Hype never bothered him as much as wishful thinking. The late ’90s web was hype and revolution at once. So is AI. Dismiss it and you become a cautionary tale. Chase it without substance and you burn out. Winners keep to fundamentals – clean architecture, sound commercial models and talent close to the technology. Discipline is not glamour, yet it is what survives the cycle.
From clicks to intent
McPhee believes the interface itself is changing. “The old model of screens and buttons is on its way out,” he said. He expects intent-driven experiences – hybrid chat, voice, dynamic systems that infer what the user wants, not what they press. Banking or shopping through conversation will feel less like navigation and more like stating an aim. Companies that embrace intent will win. Those that cling to old paradigms will watch traffic and relevance slip.
That shift only works if the foundations are ready. “AI will amplify whatever you already have,” he said. “If your systems are solid, you’ll get extraordinary leverage. If your house is shaky, AI will expose those cracks brutally.” His advice to tech leaders is blunt. Do not ignore hype and do not be blinded by it. Focus on architecture, specifications, governance and people. Prepare for transformation rather than react to it.
The personal cadence has not changed much since the Commodore. “Even on holidays, I’m usually tinkering,” he said. It reads as a habit not a hustle – curiosity applied with patience. Start with the customer. De-risk the next step. Insist on controls where they matter. Then move with purpose. In McPhee’s world, progress is not the pile of things you could do – it is the small set you choose, test and deliver, in the order that compounds.
