Last week I drove up to Boston in my usual rented pickup truck to give one of my favorite talks of all time.
The team at Bain Capital invited me to present to a room full of data leaders. These are people who have heard every AI pitch there is, so I didn't pitch. Instead, I shared what I've learned from more than 100 AI transformations, including the parts that don't make it into vendor decks.
A rare vantage point
While preparing (credit to Nicole Falasco for the idea), I realized that despite being one of the youngest people in the room, I had an experience most people there didn't: I'd been part of more than one AI transformation. I've seen several dozen in the past year alone. Most leaders haven't seen even one end to end.
So I shared the three things I've seen matter most in the implementations that succeed: data, people, and process.
Data: stop warehousing, start explaining
Data readiness is overplayed. You don't need one warehouse, perfectly clean data, or a complete ontology before you start.
What you do need is an understanding of what your data actually means, plus the tribal knowledge that only lives in people's heads. A semantic layer, if you will. My mantra here was simple: stop warehousing, start explaining.
People: one owner, tied to dollars
Transformations don't fail on technology. They fail on ownership.
The single best predictor of success I've seen is one owner whose compensation is tied to P&L impact. Not adoption rates. Not pilots launched. Dollars.
Process: start where the work is messy
"We're investing in data and AI" won't get funded. "We're fixing real-time inventory visibility on the floor to reduce stockouts and overstock" will.
Start where the work is essential, complex, and messy: supply chain, AP/AR, quoting. Plot every use case on cost versus value, and start in the corner where value is high and cost is low. Customer service looks like the obvious first move. It's often the wrong one.
Most of the company never noticed
The line I opened and closed with is the one I believe most: the best AI transformations I've seen, most of the company never noticed.
Agents run the control plans. Agents run AP and AR. Nobody on the floor feels a big, loud change. The only place you see it is in the numbers.
Great transformations are quiet, subtle, and, most importantly, they work. You don't need to move the mountain to get there. Just focus on the first stone.
Thank you to Emily Ashworth, Nicole Falasco, Florence Evina-Ze Baez, David Humphrey, and the broader Bain Capital and Bain Capital Ventures team for having me, and to everyone who stuck around afterward to keep the conversation going.
