Speed is no longer the constraint. Your organization is.

Anything you can describe clearly can now be built, by almost anyone, in days – and every company I talk to is already changing because of it, one department at a time, without anyone looking at the whole. I help chief executives and boards accelerate the company by leveraging AI rather than replacing people with it, starting small enough to prove in a quarter or two.

The room that froze

Seven engineering leaders arrived with plans after two weeks of discovery. I arrived with the thing built and in their inbox, and the silence that followed is the reason this practice exists.

The projects that don't die

The initiative on its ninth change request, the system that has been ninety percent done for two years. They are not failures of the people running them; they are what happens when a project needs a piece of the company nobody owns.

How many agents are running in your company today?

And who could tell you what they are doing, what they are spending, and what data they are reaching? Most companies cannot answer, and governing that is what the model is built for.

At three sizes

Small — Around $500MM in revenue, or a division of that size inside something larger

The proof fits in a quarter with the right setup going in, and a second domain in the first year is realistic. The shape it takes here is the same shape it takes at every size above it – that is the whole point of starting here.

Medium — Around $5B

Two quarters or more to the working example. Security and architecture join data and red tape as the things that stretch it, and change management and strategy become part of the proof team.

Large — Around $25B and up

Three quarters or more, because competing priorities and loss of focus are the biggest risk at scale – which is exactly why the first step stays small even when the company is not, and why change management sits inside the proof team from the first day.

Transforming companies is what I have done for most of my career. Five times I have taken a business and rebuilt how it creates products and technology, and those transformations worked, which is why I can say with some confidence that the one in front of us now is different in kind. Every previous transformation was designed, whether anyone said so or not, to ration a scarce thing – the capacity to build. The machinery we built to ration it (requirements gathering, roadmaps, quarterly planning, steering committees, the annual technology budget) is now the thing standing between the people you already employ and the work they could be doing by Tuesday.

The word "accelerated" is chosen deliberately, over every alternative. This is not a plan to replace people with machines; it is a plan to remove the machinery that stops the people you already have from building, and to give every function the capacity that only engineering used to have. The scarce thing in the accelerated organization is still people, the ones who know the domain and will own it, and what changes is what they spend their days doing.

What I do

I work with the chief executive and the board rather than with the engineering department. The change has to start with the organization and not with the technology, and in my experience nobody below the top has the authority to decide it (the room that froze taught me that). The engagement starts with a short diagnostic that produces your own number – what your stalled and duplicated portfolio is actually worth – and if that number deserves it, we move to a proof in one domain that the rest of the company can see and build on. I have run the operating model underneath this at three company sizes, roughly five hundred million in revenue, five billion, and twenty-five billion. I am careful about how I claim that, because the costs and the benefits scale very differently at each, and the first step has to stay small even when the company is not.

The honest line

The product operating model beneath this has been run five times, before AI, and it worked every time. The full accelerated model has been run in prototypes, in pieces inside a real company, and in one engagement that saw it clearly and froze. It has not yet been run end to end at scale. That is exactly what the proof is for, and it is why the first yes I ask for is a small one.