The first question most organisations ask about AI is "what can it do?" The better question is "what do we need it to do?"
This distinction sounds minor. It is not. Starting with capability leads to pilot projects chosen because the technology is impressive, not because the outcome matters. Starting with strategy leads to initiatives chosen because they address a board-level priority that AI can uniquely solve.
We see this pattern repeatedly. A leadership team surveys the landscape of AI tools, picks something promising, runs a proof of concept, and declares progress. Six months later, the tool sits unused. The business case was sound. It was just never connected to anything that mattered enough to change how people work.
The organisations that get this right evaluate AI initiatives on two dimensions: strategic value and cultural value. Strategic value explores if the approach towards AI has strategic relevance and financial merit. Cultural value considers the implications on job role design, talent development, operating models, and how decisions are made about AI. Consider the talents of your people, and their own sense of agency in applying these talents, and then how AI can enhance and augment talent (vs task).
The most impactful AI initiatives score high on both dimensions. They matter to the organisation's strategy and people actually adopt them. This intersection is where returns compound.
Initiatives that score high on only one dimension fail in predictable ways. High strategic value with low adoption produces expensive shelfware: a capability the board approved that nobody uses. High individual value with low strategic alignment produces productivity noise. People enjoy the tool, but it does not move any metric the board tracks.
Neither failure looks like a failure at first. The shelfware has an impressive business case. The productivity tool has enthusiastic users. Both consume budget, attention, and credibility. Neither delivers the return that justifies what comes next.
This is why the sequencing matters. If you start with technology, you spend your early credibility on initiatives that may land in either failure mode. If you start with strategy, you concentrate your effort on the intersection where both value and adoption are high, and you build the organisational evidence that funds everything after.
Your board does not need an AI strategy. It needs a strategy that happens to use AI where AI is the best answer. The difference is not semantic. It determines whether your AI investment compounds or stalls.
Start with what your board has already said matters. Then ask where AI is the only way to get there. That is the map.