Three Layers of AI Transformation Acceleration
Most AI programs I see in the middle market start with a tool. Someone buys seats, runs a pilot, waits for a number to move. The other trend is hiring a consultant to come in and build a tool. This may give a story to the board or lenders, but it dodges the goal, AI acceleration.
Companies that build durable AI acceleration, the type of acceleration that compounds, run it in a different order: people, data, process.
People
Starting with people means enabling and upskilling your team. Sometimes that is new talent. More often it is latent talent you already have. What it always means is giving your people the tools and the opportunity to understand what AI is. It is not magic. It is a tool for accelerating their capabilities.
Alexander Sukharevsky runs QuantumBlack, McKinsey's AI practice, and he put the move plainly on CXOTalk. "It's literally about taking your most talented colleagues, putting them in the room, teaching them technology, and then trying to reinvent domains." Not hiring someone to reinvent the domain for you. Teaching the people who already know it.
His team looked across thousands of transformations. The ones that worked focused on two domains or fewer, and fewer than a hundred companies have captured more than two thirds of the value on the table. Asked what he would prioritize if he were building from scratch today, model quality or proprietary data or workflow integration or user adoption, he named adoption as the hard one, and warned against "a quest toward creating the perfect technological platform, creating the perfect data lake."
AI may come for some jobs. It will certainly come for the people who do not use it.
Data
This is about mapping your data and making it accessible, inside an appropriate governance framework, to the people in the organization who need it. Giving people training and tools without that access just creates risk and slows the transformation down. Data is the fuel. It has to be securely available before any of this works.
Bharathi Rajan is VP of data, digital and AI at Swire Coca-Cola, one of the top five Coke bottlers in North America, six plants and sixty distribution centers. She refuses the usual framing. "Governance is not red tape. If you want to make decisions with the right data, you need to have governance, you have to have the quality data flowing in."
Her account of where the work actually started is this leg in one sentence: "making sure that we have access to the data and we are giving the data in the hands of the people that actually are going to use the data." Not a lake. Not a platform. The right data reaching the person who has to decide something, on time and in a format they can use.
Sukharevsky adds the part that stops this becoming a two year infrastructure project. "There is no situation when there is a perfect data." You work out what you have, what you could have, and what the domain actually needs. He also points out that opening up the flow is rarely a technical decision. It is a CEO decision, because it is about the distribution of power and influence inside the organization.
Process
Once you have people who are excited about AI and the data to let them get things done, you turn to process.
Why not start with process? Because if you get the first two right, they will do it for you. Smart people with good data redefine their own workflows, and they do it better than any process map you could have bought.
Joshua McKenzie, chief technology officer of ELMO Software Group, ran this over about twelve months. It began, in his words, with "a combination of training, brown bags, showcases, demos, pair programming, all of that normal stuff." Ordinary enablement. Where it ended was not ordinary. It "eventually built all the way up to redefining that entire process that we have and specific roles within that process. And that was all done collaboratively."
Nobody handed his team a target operating model. They arrived at one. That is where the acceleration hits the ground and the productivity becomes visible.
Sukharevsky's word for the sequence is the one that should bother anyone running it backwards. He calls the journey "invariant." There is one way through it. "If you miss one or few of them, the value is not getting unlocked."
This is not easy. It takes conviction from leaders and the board to do and support the hard work. Most companies are at the very beginning of this journey. If you look across PE podcasts, AI came up in about half of episodes in late 2025. Over the last three months it is five in six. Almost everyone is talking about it. Far fewer are doing the hard work.
To drive the transformation, start with the people, find the great ones, free your data, and let those two change how your business runs. That is how you start AI acceleration and get the compounding effect of change.
Sources: Alexander Sukharevsky (QuantumBlack, McKinsey), CXOTalk, "McKinsey: Why Agentic AI Pilots Stall," Jun 24, 2026. Bharathi Rajan (Swire Coca-Cola), The Data Chief, "How Swire Coca-Cola Turned Governance Into an AI Growth Engine," Jul 8, 2026. Joshua McKenzie (ELMO Software), The Data Chief, "How to Transform a SaaS Company with AI from ELMO CTO," Jun 24, 2026.Jun 24, 2026.
(AI reach and sentiment: MeetBri.ai podcast analysis, PE vertical, Aug-Oct 2025 vs. May-Jul 2026, n=112 episodes.)