A Simple Framework for CEOs Who Feel Lost on AI
Most leadership teams I talk to feel overwhelmed by AI. Not because they're ignoring it — the opposite. They're reading about it constantly, hearing about it in every conference, watching competitors make announcements. But when it comes to deciding what their company should actually do, there's a lot of uncertainty.
The confusion is dominated by two extremes: apocalyptic warnings about job losses and vendor hype about transformation. Neither helps you decide what to do on Monday morning.
This is a simple way I've found useful when working with leadership teams on AI.
It's not an AI strategy. It's a way to get moving.
Two buckets, not one
AI initiatives tend to fall into two categories, and it's worth separating them clearly.
The first is productivity — using AI to do what you already do, but faster or cheaper or with fewer errors. This is where most companies should start. It's lower risk, easier to measure, and builds organizational capability for bigger moves later.
The second is business model — exploring whether AI enables you to do something fundamentally different. New offerings, new markets, new ways of delivering value. This is higher stakes, longer horizon, and requires dedicated strategic attention.
Most leadership teams mix the two. They talk about "AI strategy" as if it's one thing. It's not. Productivity improvements and business model exploration require different approaches, different timelines, and often different people.
The productivity path
Within productivity, I see three types of work.
AI literacy training comes first. Everyone needs a baseline understanding of what these tools can and can't do. This isn't about making everyone a prompt engineer — it's about removing fear and building enough fluency that people can recognise opportunities in their own work.
Workflow development and coaching is the next step. This is where individuals and teams actually change how they work. It's not enough to train people and hope they figure it out. You need dedicated support — what I'd call AI workflow coaches — helping people build new habits and integrate tools into their daily routines.
Process improvement at the function level is where it gets more structured. Think of this as applying Lean methodology to AI opportunities. Which processes have the most waste? Where are the bottlenecks? AI becomes a tool for continuous improvement, not a one-off technology project.
The business model path
Business model exploration is different. It sits at leadership team level — this isn't something you delegate to IT or a task force.
Competitor and market intelligence is the starting point. Systematic scanning of how AI is reshaping your industry and adjacent spaces. What are competitors doing? What are startups doing? What's happening in industries that faced similar disruptions earlier?
External expert advisory fills in the blind spots. Bringing in specialists who've seen more implementations, more failures, more edge cases than you'll encounter internally. Not to outsource thinking, but to pressure-test your assumptions.
AI-enabled exploration and prototyping is where you actually test ideas. Using AI tools in-house to research possibilities, build rough prototypes, experiment with new concepts before committing resources.
Who owns this?
This cannot be delegated.
Yes, HR should own competence development. Yes, IT should be involved in infrastructure and governance. But AI strategy — especially the business model exploration — must be owned and led by the CEO and leadership team.
And not just as oversight. Leaders need to visibly participate. Try the tools themselves. Share what they're learning. Be open about uncertainty.
There's a temptation to look confident even when you are not. People see through that. What employees actually want to see is initiative and effort — the sense that leadership is actively figuring this out, not waiting for someone else to provide the answers.
Direction is unclear. But there has never been a better way of finding direction than starting to move.
Where to start
If this feels like a lot, here are three actions with almost no downside:
Train everyone on AI basics. Not a two-hour webinar — a proper literacy programme that builds real understanding.
Get dedicated workflow coaching in place. Someone whose job is helping teams actually adopt these tools, not just learn about them. If needed, bring in external help to get started quickly, and build a lean internal capability as you go.
Set clear upskilling targets. This also becomes a lens on workforce adaptability — and that's information you'll need regardless of which direction AI takes your industry.
Illustration: Productivity vs Business Model work
