Take Control of AI Adoption
AI capabilities are advancing faster than most organisations can absorb. Models are more powerful, easier to use, and dramatically cheaper. In just a few years, the cost of accessing GPT-4-class AI capabilities has fallen by by roughly three orders of magnitude, making advanced AI widely accessible across organisations.(1) From the outside, it looks like an explosion of opportunity.
Inside organisations, however, many leaders feel a growing sense that AI adoption is happening to them, rather than being deliberately shaped by them.
Pilots appear across teams, tools spread informally, and expectations rise. But control, over direction, impact, and risk, feels increasingly fragile.
And this is not a technology problem, it is a leadership one.
Losing control doesn’t look like failure
Most organisations don’t lose control of AI adoption overnight, it happens gradually.
At first, leaders encourage experimentation, they want teams to explore possibilities and learn. And this is healthy. But without clear guidance, experimentation starts to fragment and, as a result, two familiar patterns emerge as we’ve seen in multiple situations.
Direction
In some organisations, leadership invests heavily in direction. AI strategies are written, and ambition is articulated. However, the organisation lacks the skills, routines, and confidence to execute, and AI remains something discussed at the top, not practiced throughout the organisation.
This pattern has been visible in financial institutions such as Goldman Sachs(2), where leadership articulated clear AI ambitions early on, but embedding AI use safely and confidently across the organisation took longer.
Direction was clear, but building organisation-wide capability within regulatory and cultural constraints proved more demanding.
Capability
In others, teams build real capability. While people learn fast and use cases multiply, leadership hasn’t yet clarified where AI should truly matter. Activity increases, but focus weakens. As a result, AI adoption accelerates without intent.
Variations of this pattern have been visible in organisations like Shopify(3), where teams rapidly embraced AI tools in day-to-day work, building strong local capability. Only later did leadership step in to clarify where AI should be standardised, where autonomy was encouraged, and how individual initiatives connected to broader priorities.
Capability moved fast; direction had to catch up.
Both situations feel busy and both feel progressive. However, neither gives leaders real control.
Control comes from alignment, not restriction
Taking control of AI adoption does not mean slowing everything down or centralising every decision.
It means aligning two forces that often drift apart:
AI capabilities - what people and teams can reliably do with AI, supported by skills and organisational change
AI direction - how leadership guides that effort through strategy and governance toward meaningful outcomes
When these evolve together, leaders gain control without killing momentum.
When they drift apart, AI adoption becomes reactive.
To make this visible, I use a simple diagnostic: the Zero to 100 Matrix.
From reaction to intention
The Zero to 100 Matrix does not rank organisations by maturity, it helps leaders understand why control is slipping.
Some organisations remain Observers, watching AI closely but hesitating to commit. In these cases, control feels safe, but progress stalls.
Others become Thinkers, strong on vision, but weak on execution. Control here exists on paper, not in practice.
Then there are the Doers, the teams who move fast, but where leadership struggles to steer. In these, control is lost not because people act irresponsibly, but because direction is unclear.
Lastly, only a small group reaches the Leaders position. These are the organisations that combine capability and direction, allowing AI adoption to compound rather than sprawl.
The insight is simple but demanding:
You don’t regain control by doing more with AI.
You regain control by building what is missing, in the right order.
What taking control looks like in practice
For some organisations, control means building capability first: giving people the skills, confidence, and shared language to work with AI responsibly.
For others, it means clarifying direction: deciding where AI should create advantage, and where it shouldn’t.
The mistake I see is organizations trying to fix everything at once. Progress from zero to 100 should come from identifying the current situation and addressing constraints deliberately, not from reacting to every new tool or opportunity.
Before you move on, take a moment to reflect.
Where has AI adoption started to run ahead of leadership intent in your organisation, and what would it take to bring them back into alignment?
This framework is part of a broader playbook I’m developing for leaders navigating AI, the AI Strategy Playbook: From Zero to 100. If you’d like early access to the book and related materials as they become available, you can join the waitlist here.
_
(1) https://a16z.com/llmflation-llm-inference-cost/
(2) https://www.ft.com/content/d570c5a6-02ad-4a28-8129-2f1a02e63603
(3) https://www.forbes.com/sites/douglaslaney/2025/04/09/selling-ai-strategy-to-employees-shopify-ceos-manifesto/