The Same Old Same Old at New Speeds
Why these 3 Ds of innovation matter more now, not less
I built four profitable businesses in the first 25 years of my career. I’m on track to build four THIS YEAR with the help of two things I never had before.
The first is a team of partners who think in systems, build at speed, and challenge every assumption I make.
The second isn’t AI. It’s what we use AI for. Not just expediency. Expansion.
Most of the AI conversation right now is about going faster. Same direction, cheaper, quicker. That matters. But it’s not the superpower.
The superpower is going wider and deeper. Diverging farther. Sensing more. Learning more. Testing more possibilities before committing to a single one. Then converging — hard — on what actually deserves to exist.
That distinction is the difference between building the right thing and building the wrong thing faster than ever before.
The Unchanged Rhythm of Innovation
One of my favorite lines of The Insider’s Guide to Innovation at Microsoft is in the conclusion:
“The best innovators have learned to diverge farther than everyone else. They dive deeper and stay submerged in the possibilities longer, surfacing with insights that propel them forward in the continuous cycle of innovation.”
That was true when we wrote it. It’s exponentially more true now that AI extends how far you can diverge, how deep you can dive, and how long you can stay submerged without drowning.
Discover. Design. Develop. Within each phase: diverge, converge, synthesize. This is the rhythm of innovation found in all the case studies we explored and those we lived.
The other side of this immutable pattern is that undisciplined divergence doesn’t work either. A controlled study on AI-assisted brainstorming found that when AI generated broadly divergent ideas without structure, participants shut down — they selected fewer ideas and trusted the output less.¹
That’s not evidence against expansion. That’s evidence for the discipline.
Discover-design-develop, with diverge-converge-synthesize operating within each phase, is what keeps expansion from becoming noise. The discipline isn’t opposed to expansion. It’s what makes expansion usable.
When building gets cheap and fast, the cost of skipping divergence feels like zero. AI whispers, you could just build it now and find out. But the cost isn’t zero. It’s deferred. And it compounds.
Three hours on an experiment? Great. Three months building something you never validated? That’s not an AI success story. That’s the same old same old at new speeds.
Expansion + Expediency in Practice
On a recent Friday, I sat down with a question about where our next product should go. By that afternoon, my team and I had pressure-tested the idea through multiple expert perspectives, researched the competitive landscape in real time, identified a market gap nobody else had named, and built a proof of concept we could throw darts at.
A year ago that would have taken weeks. Six months ago, days. Now, hours. Not because we skipped steps. Because every step got wider and faster at the same time. More divergence within discovery. More options evaluated within design. More rapid validation within development.
The discipline held. The aperture expanded.
Two AI Futures
Two futures are forming right now (at least), and most organizations don’t realize they’re choosing.
Future 1: AI automates existing processes. Things get faster. Costs drop. When the market shifts, these organizations discover they’ve been sprinting in the wrong direction — efficiently.
Future 2: AI widens discovery, deepens sensing, and accelerates the ability to test before building. When the market shifts, these organizations have already seen it coming — because they invested in perception, not just production.
My friend and former teammate at Microsoft, Yue Tu, asked the right question this week. Our Regenerative Datacenter of the Future team had done deep discovery work and predicted today’s community sentiment would be a major constraint on data centers. We even quantified the financial impact years before it showed up in the headlines. His question now: “What’s next, and in what time horizon?”
That’s the discover question.
No living system optimizes for speed alone. The organisms that survive sense broadly before they commit resources — reading light, moisture, and climate across hours, seasons, and decades simultaneously. The best discovery works the same way: reading signals across multiple time horizons at once. What’s happening now. What’s shifting this year. What’s emerging over the next three to five years. AI makes that multi-temporal read possible at a pace that wasn’t available before.
The organizations that use AI only for speed will be outrun by the ones that use it for sight. The ones who go wider before they go faster. The ones who earned the right to be efficient because they already know they’re building the right thing.
How far ahead do you need to look to stay ahead when everything’s changing this fast?
I know where the answer starts: with discovery. It always does.
THAT SAID — discipline is not the only problem. As we’ve discussed before, we can’t change people’s behavior, only their context. What happens when the incentive architecture of an organization actively punishes divergence?
Product managers are measured on releases. Engineering teams are measured on velocity. Sales teams on bookings. Boards measure quarters. In most organizations, skipping discovery, design, and divergence throughout to do more, faster is the most rational career move available. AI just made that path even more rewarding — ship faster, show output sooner, get promoted.
Discipline matters. But if the structure rewards expediency, asking someone to practice expansion is asking them to bear a cost the organization isn’t paying for.
More on that next time.
Thank you for reading and listening. Don’t forget to register for our live monthly recording of Innovating out Loud. Our next guest is….TBD! But we’re aligning calendars with a few interesting folks. Stay tuned!
Connections to The Insider’s Guide to Innovation at Microsoft:
Discover-Design-Develop with Diverge-Converge-Synthesize within each phase
Adaptive Cycle of Innovation — The cycle that once played out over years now plays out in weeks and months. Expansion doesn’t slow the cycle — it ensures each loop is aimed at the right thing.
BXT (Business, Experience, Technology) — Cross-disciplinary discovery that expands the aperture before design begins
Sources:
Komura and Yamada, “Deepening ideas vs. exploring new ones: AI strategy effects in human-AI creative collaboration,” PLoS ONE, January 2026. Link
This piece was created with the help of AI — specifically Claude, Perplexity, and a team of expert personas built by Regenerous Labs. Direction, judgment and final decisions by me. Say it ugly, then build it better. Onward!
