Approach
Capability isn’t conversion.
AI capability is developing extraordinarily quickly. But an organisation does not realise value simply because the technology becomes more capable.
There is quite a lot that has to happen in between. Someone has to choose a problem worth solving. The technology has to fit the work. People need to understand how to use it and when to apply judgement. The organisation needs enough confidence to rely on it. Accountability has to remain clear. The cost of building and operating it needs to make sense. And ultimately something worthwhile has to change.
I think of the organisation as the thing that performs that conversion.
That idea shapes how I approach the work.
01 · Design
Technology should be intuitive.
We have traditionally treated technology adoption as something that happens after implementation. We deploy a tool, train people to use it, communicate the benefits and try to build new habits.
There will always be an enablement component to change. But AI raises another question: how much of the adoption problem should have been solved in the design?
If people need extensive training to navigate unnecessary complexity, remember unintuitive processes or constantly compensate for the technology, we have created an ongoing adoption burden.
This becomes even more important as people move beyond using AI and start building with it. They are now designing experiences and workflows that somebody else may need to adopt. At the point where AI materially changes a process, the question gets bigger again. We need to think about the work itself.
02 · Trust
Trust should be designed.
A policy can describe what should happen. An assessment can record a decision. A governance forum can provide oversight. But ultimately trust has to work operationally.
People need to know what they are accountable for. Builders need practical boundaries. Leaders need useful evidence. Someone needs to know when intervention is required and what happens if a capability no longer behaves as expected.
This is why I like to ask what happens on the worst day. If something goes wrong, can we see it? Can we intervene? Do we know who owns the decision? Can we recover? Could we explain and stand behind what happened?
If the answer is unclear, the trust problem probably has not been solved by the documentation.
03 · Value
Value should be realised.
AI gives us plenty of things that are easy to count. Licences. Active users. Prompts. Agents. Use cases. They are useful measures, but they are not the outcome.
The adoption work I have developed looks at the progression from access and activity through to capability, behaviour, outcomes and ultimately value. The point is to understand whether AI has actually changed the work, and what happened as a result.
That distinction matters because more AI is not necessarily the objective. Better work might be. More capacity might be. Lower cost, better decisions, reduced risk or a better customer experience might be.
The technology is valuable when it helps create an outcome worth the total effort and cost required to achieve it.
In a sentence
Designed, not documented.
I do not think organisations need an entirely new bureaucracy because AI exists. Most already have strategy, delegations, risk management, technology controls, data governance, people practices, procurement and accountability structures.
AI changes some of those things. It exposes gaps in others. Occasionally it creates a genuinely new requirement. The work is to understand the difference and design what is actually needed.