My board asked for an AI strategy and I don't know where to start

Start with the outcome, not the technology. When a board asks for an AI strategy, what they are really asking is "where will this make us money or make us better, and who is steering this." So begin there: pick the two or three business results you want in the next year, attach a pilot to each, name an owner, and decide how you will measure success. That one page is your strategy. Everything else is detail you can buy from a specialist.

Here is the part worth holding onto. You feel behind. You are not. The pressure in your chest is the gap between how fast the field is moving and how clearly anyone has explained where a leader should stand. That gap is closeable, and it closes with a frame, not a vendor.

Where do I actually start with an AI strategy?

You start at the boardroom table, with a question most strategies skip. What does good look like in twelve months? Lower cost to serve? Faster product cycles? Write those down first. They are the destination. The models, the platforms, the agents: those are how you travel, and they change every quarter, so you choose them last.

You need not be the technologist in the room. You need to be the one who knows what the business exists to do. I put it this way in my book: you don't need a qualification as a car mechanic to run a corporate fleet. You need the principles of fleet management, and the judgement to hire the mechanic. AI is the same. The leader's job is to hold the outcome steady while the engineers choose the engine.

Are we behind if we don't have one yet?

You are ahead of where the headlines suggest. Gartner found in 2025 that just 23% of organisations have a formal AI strategy in place. So a board that writes a clear one this quarter joins a leading minority rather than a catching-up crowd. The catch-up framing is the trap; it sends you chasing tools to look busy. Lead from where you intend to be instead, and the tool choices get simpler.

There is a quieter signal here too. Deloitte's 2025 Governance of AI survey, covering around 700 directors and executives across 56 countries, found that 31% of boards still keep AI off the agenda entirely, down from 45% a year earlier, and two-thirds of directors describe their board as having limited to no knowledge of the technology. The boards that capture value are simply the ones that keep AI in front of them. Your board asking the question is the first move of the group that wins.

What goes in a board-ready AI strategy?

Less than you think, on purpose. A board wants to see the destination, the route, the driver, and the dashboard. Build it in this order, because the order is where most strategies go quietly wrong.

1. Name the outcomes. Pick two or three measurable business results for the next twelve months. Margin, speed, capacity, customer experience. These anchor every later decision.
2. Choose three pilots tied to P&L. Each pilot maps to one outcome and carries a number. A pilot with no number is a hobby.
3. Redesign the work around the pilot. Decide how the workflow itself changes, not just which tool sits on top. This is the step that separates value from noise.
4. Name an owner and a budget. One accountable executive per pilot. AI without an owner drifts back into the old way of working.
5. Set the dashboard. Agree the handful of metrics the board reviews every meeting. What gets seen gets steered.

Notice step three. The Wharton 2025 AI Adoption Report (a multi-year study of 800 senior decision-makers at large US enterprises, run by the Wharton School and Gartner-backed researchers) found that 74% of firms already report a positive return on generative AI, and 72% now measure that return formally against profit and productivity. Yet the factor most tightly linked to real impact is workflow redesign, and only about one in five firms has reshaped how the work actually flows. Most leaders are trying to install new software on broken hardware: the same processes, the same habits, a clever tool bolted on the side. The tool is fine. The hardware underneath was never upgraded.

Why do most AI efforts stall, and how do we land instead?

They stall in the gap between the demo and the organisation. MIT's 2025 study of enterprise generative AI (the NANDA initiative, MIT's research programme on agentic AI in the workplace) found that around 95% of pilots delivered no measurable impact on profit and loss, while about 5% accelerated real revenue. The cause was almost never the model. It was the learning gap: tools dropped into workflows that kept their old shape.

So here is what makes a pilot land. Buy capability from specialists where you can; MIT found vendor partnerships succeeded around twice as often as pure internal builds. Then give the work permission to change. The bottleneck is no longer the technology. It is the clarity of the people deciding how the work should flow.

This is the arc behind the whole shift. Phase One, the Age of Effort: work hard, get a little more, linear growth. Phase Two, the Age of Scale: build once, sell to millions, exponential growth. Phase Three, the Age of Acceleration: output decoupled from human effort almost entirely, the phase AI unlocks. Phase One was muscle. Phase Two was machine. Phase Three is mind, and the upgrade that matters most is the one at the head of the table.

That is the frame to carry into the boardroom. The stance is "we are early, we know what we want, and we will redesign the work to get it." Same data. A completely different place to stand.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.