Jeff Bezos’s two-way doors policy says it’s cheaper to try a reversible idea than to argue about it. It’s the framework behind AWS, behind the two-pizza team, and behind Amazon’s invention rate. In 2026, with AI collapsing the cost of experimentation, it’s also the single most important decision framework for any enterprise serious about AI transformation.
In 1994, anthropologist Eugenie Scott coined the term “Gish gallop” to describe a debate technique where a speaker overwhelms an opponent with so many false claims that no rebuttal is possible in the time available. The technique has now migrated from creationist debates into the default information environment of senior decision-making — and AI has industrialized it. Here’s how to recognize it and how to defeat it.
Sir Ken Robinson once joked from a stage about a future where a $1,000 laptop would be as smart as an adult human brain. He died in 2020. The Kurzweil curve he was citing turned out to be directionally right. And the question he was actually asking — “how’s that going to feel?” — is the one almost every senior leader is quietly avoiding in 2026.
Steve Wozniak called the Macintosh a failure. He was right. The Apple III melted. The Lisa was priced out of reach. The original Macintosh nearly bankrupted the company and got Steve Jobs fired. The real history of Apple is a string of near-fatal failures — and it’s the most important lesson available to anyone leading an AI transformation today.
A waterfall would never sound melodious if there weren’t rocks in its way. The very thing we love about waterfalls is created entirely by the obstacles we’d remove if we could. Most executives spend their careers trying to smooth the path. That’s exactly why their careers are silent — and why AI is the biggest rock the executive class has encountered in a generation.
In 1950, an F1 pit stop took 67 seconds. Today the world record is 1.80. The 96% reduction has almost nothing to do with making mechanics faster — and everything to do with redesigning the work itself. The same lesson saved thousands of children’s lives at Great Ormond Street Hospital. The same lesson explains why most AI transformations are failing right now.
EY Canada just published research showing the most successful AI training programs are also the ones causing the most employee turnover. Employees with fewer than 4 hours of AI training express 21% intent to quit. Employees with 81+ hours express 45%. The retention problem isn’t an AI problem. It’s a career architecture problem most enterprises haven’t updated since the 1990s. Here’s what’s actually going on — and what to do about it.
In 2021, a Tufts neuroscientist named Erik Hoel proposed that dreams exist to keep the brain from overfitting to everyday life — exactly the way dropout prevents overfitting in AI. The hypothesis explains why senior leaders, calibrated against decades of same-industry inputs, gradually lose the ability to lead through change. The fix — what I call “executive dropout” — is the most important leadership discipline of the next decade.
Edward de Bono coined the term “lateral thinking” in 1967 and compressed the entire framework into one sentence: “You cannot dig a hole in a different place by digging the same hole deeper.” Most senior executives in 2026 are digging a deep hole in the wrong location — with all the wrong instincts, all the right credentials, and a culture that rewards intensification over reorientation. This is what to do instead.
A clickbait Reel is going viral claiming Microsoft banned engineers from using AI because AI costs more than humans. Almost every part of that framing is wrong. The actual story — about token-based pricing creating a new enterprise cost class that finance teams aren’t prepared for — is harder, more important, and the most useful lesson available to any leader running an AI program in 2026.
Tell us about your goals — we'll be in touch.
