At CincyDeliver in July, I spent twenty minutes with Matt Anderson before either of us went on stage. We were in different rooms at different times, and neither of us had seen the other’s deck. I could tell I was going to love Matt’s talk just from the title: ‘You Can’t AI Your Way Out of an Organization That Couldn’t Agile.’ Later that day, he closed his session by telling his room to go see mine, which is a generous thing to do for someone whose talk you haven’t watched yet.
Turns out we spent the afternoon making the same argument from opposite ends. His was a diagnostic. Mine was a design.
His argument was that the organizational failures that limit agility are about to limit AI, with considerably higher stakes. The part I’ve been chewing on since is his diagnosis of what an agile transformation actually changes.
Renaming the meeting a stand-up while the same ten people report status to the same manager. Putting Jira on the budget and calling it planning, even though funding still flows through annual cycles and executive horse-trading. Adding a Slack channel to the war room and calling it DevOps. The surface changed, but the structure didn’t. I’d guess most of us have lived through at least one of those.
He also did something worth noting on its own. He published a full account of how AI helped him build the talk: two research passes, one deliberately designed to prove him wrong, a second model brought in to audit the first, and a change ledger with every correction traceable to a source. The tool argued with him. It never decided anything. I made the same commitment in the preface of my book, though in a plainer form than his ledger. I think any author or speaker working in this way with AI owes their audience that account.
But notice what his account actually demonstrates. AI amplified what he already had, namely a conviction and a method. Point the same tools at an organization with neither, and they will amplify that instead — the silos, the hoarded information, the decision bottleneck two levels above the people with the facts, all encoded and accelerated. It’s the same argument his talk makes about organizations, running at the scale of one person.
So here’s the thing I want leaders to sit with.
The destination was never agile. It isn’t AI either. Both are capabilities, and a capability only becomes an advantage in proportion to the organization’s ability to absorb it. In my experience, that absorption capacity has very little to do with talent, tooling, or budget. It’s set by the operating model — how strategy gets made, how product decisions are owned, how work flows, how the organization is designed, and how leaders behave when someone turns out to be wrong. Every dysfunction your agile transformation didn’t reach is sitting exactly where you left it, and AI is going to run into it at speed.
Which means you already know where it will hit. Your agile transformation spent years finding out which parts of the organization would bend and which ones wouldn’t, and it produced a list: the exceptions that got granted, the teams that were carved out, the conversations that ended with ‘yes, but not that part.’ That list tells you exactly how ready you are for AI. Most companies never looked at it honestly, so they don’t know the uglies on it. The rest know and are happily ignoring them.
And your list is not unusual. Compare notes with peers at other companies and they look remarkably alike. Funding still moves on an annual cycle, so the work can only move as fast as the money does. The people closest to the information still aren’t the ones deciding. Value flows sideways across the organization while teams and reporting lines run up and down. Governance wants a plan and a projected return before it will fund anything truly uncertain, which rules out most of what’s worth learning. And in many rooms, being wrong is still treated as a performance problem rather than as information.
None of that is a mystery to anyone who works there. The question is which of them has ever been touched by someone with the authority to change it.
That’s your ceiling. It was your agile ceiling, and it’s about to be your AI ceiling. The second one will cost more because technology moves monthly while organizational change moves in years. The gap compounds, and so does your transformation debt.
I’ve heard companies write “use AI” into their OKRs, which is as crazy as it sounds. Before any AI initiative, the question remains the same as it has always been: what problem are we actually trying to solve? Without an answer, you get exactly what you asked for and nothing in return. A good operating model doesn’t make AI irrelevant. It’s what makes AI investable.
I describe this very point in my book, Assembled. Aligned. Adaptive. Building an adaptive organization means you no longer need a new transformation program every time the industry hands you a new capability. You build the environment once, deliberately, and disruption becomes something you absorb rather than something you survive.
If you want to take a closer look at where your organization stands, Matt built the tool for it. ‘The AI Readiness Field Guide’ opens with four ceiling questions and includes a retro you can run with your own team: edgeofcomplexity.com/2026cincydeliverhandout