There’s a new narrative shaping the conversation around AI, and it’s one that feels very different from just a year ago. Back then, the priority was getting people comfortable with the idea of using AI. Companies rolled out tools at scale, invested in upskilling, and encouraged experimentation at every level.
We did the same. In 2023, shortly after ChatGPT arrived, we set up a dedicated AI Transformation Team, one of the earlier moves of its kind in our industry. We built two programmes to carry the work forward: AI Infusion, focused on building AI capability, and DigiCrew, covering the RPA, analytics and automation work.
The question in front of us is harder: are our leaders ready?
From experimentation to accountability
The Financial Times reported in June 2026 that companies including Amazon, Uber, Walmart, Cisco and Meta, all early and enthusiastic AI adopters, have started introducing caps, steering employees toward more efficient models, and discouraging “AI for the sake of AI.” What started as a push for adoption is becoming, in many businesses, a need for control, cost management and most importantly of all, value.
The conversation is shifting, and quickly, from experimentation to scrutiny. The ask is not whether companies are using AI, but whether they are getting enough out of it.
Goldman Sachs analysts, cited in the same report, expect token consumption to increase by 24-fold by 2030, driven by AI agents. And that’s a lot of spend to leave unmanaged.
We recognise the pattern, but it isn’t our story. From day one, our AI Transformation Team has worked to a simple discipline: prove the value before you scale a use case and retire the ones that don’t earn their place. We built a strategy that tells us where to focus, and it has kept generating value we can point to rather than being an activity we have to explain.
You can call that discipline if you want. I would call it a leadership choice, and it’s where I think the industry’s attention needs to go next.
No such thing as an AI strategy
There is no such thing as an “AI strategy.” There is only company strategy, driven and enabled by AI. AI transformation was never about the technology, it was never the goal, and using AI for its own sake was never the point. You start with a business problem and design the best solution for it. Sometimes that solution involves AI, often it doesn’t. The moment AI becomes something you deploy because it’s there, rather than in service of an outcome you actually care about, you’ve already lost the thread.
That means understanding enough to question existing assumptions and identifying opportunities that simply weren’t possible before.
Which means the deciding factor in this next phase isn’t a tool, a model, or a budget line. It’s whether the people at the top understand enough of what’s now possible, data science, machine learning, generative AI, agentic systems, automation, to challenge their own strategy with it. Leaders who can do that start treating AI as a default assumption in how they design the business.
Once you reach this stage, there’s a temptation to skip straight to the exciting part: new business models, new revenue lines, a company reinvented around AI from the ground up. But you get there by doing the less glamorous work first, embedding AI properly into how the business runs, measuring what it changes, learning where it helps and where it doesn’t, and only then reimagining what’s possible. If you skip that step, you’re only decorating the old model with a new label.
The rise of the AI manager
There is another shift worth mentioning as it will define the next few years of work: as agentic AI matures, more of our people will spend part of their day managing not just colleagues, but teams of agents, training them, checking their output, deciding what they should and shouldn’t be trusted with. Nobody was taught that management skill in business school. We’ll now need to teach it.
I’m not saying the technology doesn’t matter. I’m saying leadership matters more than it has in years. Being first or fastest to adopt won’t win this. The companies that come out ahead in this next phase will be the ones whose leaders rebuilt how they operate around AI and did it faster while the rest are still figuring out what to do.
