I have described working in a flow-state with AI is like putting on an Iron Man suit. At its best, it extends your thinking, helps you work through complexity, and gives you a level of capability that would otherwise take much longer to access.
That is powerful. But it is only the beginning.
Most of the AI conversation in business is still about individual productivity: writing faster, summarising documents, preparing a report, analysing a spreadsheet, or reducing administration. These are worthwhile gains. I see them regularly in New Zealand businesses.
But a task saving is not automatically a business improvement.
AI is changing more than the tools people use. It is changing some of the constraints around which organisations were designed. If information can be found, compared, drafted, analysed and routed more quickly, then the process around that work needs another look. So do the roles, hand-offs, reporting lines and decisions built around it.
This matters in New Zealand.
Many of our businesses operate with lean teams, constrained capital and limited management bandwidth. Good people often carry years of practical knowledge in their heads. A scheduler, administrator, estimator, operations manager or technical specialist can be holding together processes that have evolved over a long time. There is rarely the luxury of a large programme or a team available to stop work and redesign everything.
At the same time, the pressure to improve productivity is real. Markets move, customers expect more, experienced people leave or retire, and the work does not get any simpler. We need to get more value from the capability already inside our businesses, while protecting the judgement, relationships and local knowledge that make them work.
An Adaptive Organisation is one that can repeatedly adjust how work, decisions and learning happen as conditions change.
It does not mean chasing every new AI tool. There is already plenty of hype, and AI is not a silver bullet. It is an amplifier of ability. It can magnify good thinking and strong processes, but it can also magnify poor information, weak judgement and existing inefficiencies.
There is also a lure-of-the-ring aspect to AI. New power is seductive. It is fast, fluent and always available, which makes it easy to use for every answer, every document and every decision.
Used without judgement, that can produce AI slop, weaker critical thinking and uncontrolled token costs and risks without a matching business benefit. People can begin to accept plausible outputs too readily, outsource too much of their thinking, or generate far more content than anyone needs.
The place to start is with how work is actually done.
Where does time go? What data gets rekeyed, chased, checked or reported repeatedly? Where are people waiting for information, approvals or someone else’s knowledge? Which decisions need experienced judgement, and which parts could be better prepared or supported by AI?
Sometimes the first opportunity is straightforward. A process that currently takes six hours may be redesigned to take four. That is useful, but the real value is not just the two hours saved. It is what happens next.
In Good to Great, Jim Collins describes the flywheel: sustained progress comes from many deliberate, reinforcing actions rather than one dramatic breakthrough. The idea applies well to AI. Time saved in one workflow can be reinvested in better information, better decisions, stronger customer contact or the next improvement. Those gains start to compound.
It does not happen by accident. Without an intention for the time AI creates, it is easily consumed by more emails, more reporting and more of the same work. AI then helps the existing system run faster, rather than helping the business improve.
That is also why I think the work is about teaching people how to fish.
The aim is not to show people a few clever AI tricks and leave them dependent on an expert. It is to build AI agency: the confidence and judgement to spot an opportunity, give AI the right context, test the output, check it properly and improve the approach over time.
Once people understand what AI can and cannot do, and see it applied to their own work, they develop a new set of eyes. They start to notice repetitive tasks, unnecessary hand-offs, buried knowledge and decisions that could be better supported. Opportunities they had not previously seen become more obvious.
That is when capability starts to spread through a business.
There is a further challenge. As humans, we have a self-serving bias in the stories we tell about AI’s impact. It is easier to imagine AI removing the drudgery while leaving our existing roles, professional boundaries and management structures fundamentally intact.
Sometimes that will be right. Human judgement, relationships, context, accountability and trust will remain essential in many areas. AI is capable in some tasks and unreliable in others, so people must continue to test, challenge and verify its outputs.
But we should not decide in advance that every existing task, role or layer of management must remain. Some work should remain human-led. Some can be improved by AI assistance. Some can be handled largely by AI, with sensible guardrails. Some work may no longer need to exist, and that's ok. New roles will also emerge.
The human consequences matter. People need role clarity, support and the space to build new capability. Yet being human-centred should not mean protecting old ways of working regardless of whether they continue to serve the customer or the business. It should mean human agency and purposeful design.
For me, the opportunity is not simply AI adoption. It is helping New Zealand individuals and businesses become better able to adapt themselves.
That means building trust and capability, testing AI in practical workflows, measuring what works, keeping human judgement where it adds value, and deliberately reinvesting the gains into the next improvement.
AI is the catalyst. The lasting capability is adaptability.