Long before I was keynoting about AI, I was wearing a headset in pit lane at the 1990 Australian Grand Prix, making sure the critical race communications kept working.
Without that, there was no race.
That experience still shapes how I think about AI today.
It’s not about more horsepower. It’s about a sharper strategy.
This week, I presented at AI Roadmap London at the F1® Arcade alongside Tom Godden and Gregory Eckert from Amazon Web Services (AWS), Nikola Pitman and Maike Strudthoff from Pegasystems, and Rina Madlani from Accenture.
Our venue, complete with 60 F1 simulators, was the perfect setting, because the point I wanted to make is this: AI strategy is race strategy.
My F1 experience at the Adelaide Grand Prix never left me.
And it’s exactly the lens I use when I look at how organisations are approaching AI today.
Here are the 5 decisions every leader needs to make right now:
1️⃣ Pick ONE workflow. Too many pilots is the biggest mistake I see. If you can’t name the workflow, the owner, and the success metric in one sentence, you haven’t chosen yet. Pick one. Own it. Prove it. Scale it.
2️⃣ Decide what stays human. In F1, the driver stays in the car. The pit crew changes the tyres. AI reassigns people, it doesn’t replace them. Keep humans in the loop for any decision where being wrong carries a serious cost.
3️⃣ Define your production metric. The chief strategist doesn’t care how fast the engine runs in isolation. They care about lap time. Write down three business outcomes before you scale.
4️⃣ Install minimum viable governance. The team principal, not the engine manufacturer, is responsible for race day performance. You need a named business owner, a named risk owner, an audit trail, a human override, and a review point. You can close this gap in 30 days. It doesn’t require a committee. It requires decisions.
5️⃣ Build for agents now. The teams dominating F1 this season made their engineering decisions back in 2023. They anticipated where the rules were heading and built for it. The winners won’t be the firms with the most AI.
They’ll be the firms that redesign their work around it.
I also shared a 90-day plan:
→ Days 1–30: One workflow. One owner. One metric.
→ Days 31–60: Design the controls, data boundaries, and human override.
→ Days 61–90: Deploy one contained use case.
Review it formally. Then scale.
You don’t win the championship with your first race. You win Silverstone. Then Monza. Then you build a championship.
The race has already started. None of this requires budget approval. None of it requires a board meeting.
It requires decisions.
