If intelligence is everywhere, why isn't advantage?

Over the past two years, I have found myself moving between two very different worlds. In one, I spend time with people building frontier artificial intelligence, where the conversation is about what might become possible next. In the other, I sit with boards, CEOs and executive teams trying to work out what those advances actually mean for their organisations.

It is an unusual vantage point, and lately I have been thinking about the distance between the two.

The people building frontier AI are asking:

It is a question they are answering at extraordinary speed. Each generation of models can reason further, work across more complex tasks and do things that, not long ago, we assumed would remain exclusively human.

But when I leave those conversations and walk into an organisation, the question changes.

I have become increasingly interested in the difference between those two questions.

For almost two decades, my work has involved trying to turn new technology into enterprise value, across banking, energy, telecommunications and, more recently, alongside organisations putting AI to work. What that experience has taught me is that access to technology has rarely been enough to explain the difference in outcomes. Organisations in the same industries buy from the same vendors, build on the same platforms and compete for the same talent. Yet some turn new capability into something meaningful while others struggle to move beyond pockets of progress.

AI makes that gap harder to ignore.

The capability of the technology is now advancing outside the organisation, at a pace no individual enterprise controls. A new model can arrive overnight. A capability that was scarce can become widely available within months. What one organisation can access, its competitors can increasingly access too.

But access to more capable intelligence does not make the organisations using it equally more capable.

That is the part I keep coming back to.

I suspect the answer has less to do with the intelligence we are building than with the organisations we are asking to absorb it.

The assumption underneath most AI programmes is that better technology should make an organisation better. Give people stronger tools, automate more of the work, add intelligence to more processes, and performance should follow.

In practice, the relationship is far less direct.

The first response in most organisations is additive. Intelligence is placed into a workflow that already exists. A copilot is added to a role that already exists. An agent takes over part of a process nobody has examined in years. The gains are real and they report well: faster execution, less effort, higher throughput.

The organisation itself remains as it was.

Some of that is reasonable, and worth saying plainly. There are roles where a copilot on an existing workflow is the right answer and the return is genuine. Not everything needs to be reinvented.

But a model that completes a task in seconds cannot tell an organisation whether the task should exist. An agent can execute part of a workflow and has no view on who should be accountable when it does. Nor can intelligence tell you what happens to the economics of a business when something that used to be expensive costs almost nothing.

Those questions sit somewhere else. They sit in strategy, operating models, funding, governance, incentives, organisation design and the way people are equipped to work. Most of those systems were built around assumptions that made sense when expertise was scarce, when work was largely human, and when technology moved slowly enough to plan around.

The frontier will keep increasing the supply of intelligence. What decides the outcome is how much of it an organisation can absorb. Take it in, put it to work, and reorganise around it.

The answer is not another technology programme. It is a different set of conditions.

That took me a long time to see, because the conditions themselves do not look new.

AI has changed my view of them. When the technology improves continuously, everything around it has to be capable of moving, and most of it was designed on the assumption that it would not have to.

A strategy written around today's economics is overtaken when intelligence changes the cost of something the organisation had assumed was fixed. A funding process built for multi-year programmes struggles with work where the value has to be discovered before it can be forecast. An operating model built on hand-offs between functions becomes hard to defend once intelligent systems work straight across them. Risk frameworks that control by slowing things down and adding checkpoints stop being a control and become the constraint.

I have seen the conditions move in both directions, in the same decade, in two organisations I was proud to work in.

The first was before generative AI. We had built the capability everyone says an organisation needs. Models in production rather than in pilot, running against real decisions in the business rather than beside it. It was difficult work and it was producing.

Then the company went through a structural change that had nothing to do with the programme. Leadership changed and the sponsorship changed with it. What surprised me was that nobody cancelled anything. The work simply mattered less to the people who came next. Funding conversations became harder with each cycle. The models kept running while the ambition around them narrowed.

Nothing about our capability had changed. Everything around it had.

I had thought of sponsorship as something that supported the conditions. I have come to think it is one of the most important conditions.

The second was a different organisation, some years later, and the conditions moved the other way. That was not luck. Four things were different, and they were different on purpose.

Comparable capability, two very different outcomes. In one, the conditions changed around the work and the work went with them. In the other, the conditions were built on purpose, and the capability compounded on top of them.

None of those decisions sat in the technology function. That is not a criticism of technology leaders, most of whom see the problem more clearly than anyone. It is an observation about where these decisions actually live.

The conditions also compound, which is the part that rewards patience. When leaders agree where intelligence should create value, funding becomes clearer. When funding pays for learning rather than delivery alone, teams redesign work instead of automating it. When governance gives people confidence, adoption stops needing to be pushed. When capability sits across the business rather than inside one team, responsibility moves closer to the people who understand the work.

None of them is sufficient on its own. Together they are the reason two organisations with the same vendors end up somewhere entirely different. One introduces intelligence into a company whose assumptions remain intact.

The other begins to change the assumptions.

This is what I mean by cultivating the conditions. It is the deliberate work of building an organisation able to absorb new intelligence and reorganise around it. It is slower than buying things, and if a board is counting licences deployed and pilots running, it will look as though very little is happening. That is worth settling early, because it decides how much room there is to do the work properly.

The frontier companies will keep making artificial intelligence more capable, whether or not the rest of us are ready. The work of leadership is different. It is to build organisations that become more capable because of it. Whether strategy is rethought as the economics of intelligence change. Whether work is redesigned rather than automated. Whether the conditions exist for intelligence to be absorbed rather than only deployed.

The technology arrives from outside. The conditions have to be grown inside, by the people already there.

Which also means they can be lost from inside, by people who never intended to lose them.

Artificial intelligence may prove to be one of the most significant technological advances of our lifetime. Whether it becomes one of the most significant advances in organisational capability is a leadership decision.