In a rapidly changing world, adaptability wins
There is a lot of energy going into predicting what AI will mean for organisations over the next few years.
Which capabilities will matter, which roles will change, what agents will be able to do, what the organisation of the future will look like.
Some of those predictions will be right. Plenty probably will not be.
The more important point, I think, is that we should be paying far more attention to the pace of change, and what that might mean for what we choose to focus on as organisations. Roles, products, markets, technologies and operating models are moving quickly, and some of the things organisations are designing around today will inevitably change again.
That should probably make us a little more humble about predicting exactly where this goes. But it does not leave organisations without something useful to optimise for. Quite the opposite.
In a world that is changing as quickly as this one, and continuing to accelerate as it does so, perhaps the strongest organisations will be those most capable of adapting well.
Adaptability is more than agility
The word adaptability gets used fairly loosely, often as a synonym for agility. There is obviously a relationship between the two. Organisations should be able to respond quickly when circumstances change, stop doing things that are no longer working, move resources, alter a product or revisit a strategic direction when the evidence tells them to.
And there is nothing inherently flippant about agility. A change can be thoughtful, evidence-based and extremely well reasoned. But speed of response alone is not adaptability.
An organisation can make a perfectly sensible change today and another perfectly sensible change six months from now, while learning remarkably little between the two.
The people in each room can analyse the information properly, debate the options and arrive at a sound answer. But if the reasoning disappears once the meeting ends, the next group is still forced to reconstruct much of the thinking for itself.
That is where I think the deeper necessary capability sits.
In a world of accelerating change, the advantage may not come from how fast people or AI agents can work, or how rapidly the organisation can pivot, but from how fast the organisation can learn as a network, turning the decisions, reasoning and learning created through that work into shared organisational knowledge.
Because adaptability is not simply the ability to change. In a rapidly changing environment, it depends on whether the organisation can learn from experience, carry that learning forward, and make it available to what happens next.
Adaptation depends on inheritance
The principle is much older than organisations. Humanity has become extraordinarily adaptable. We have made homes across vastly different climates and environments, crossed oceans, taken to the skies and developed technologies that allow us to operate in places our biology alone never could. We have continually found ways to change what we do in response to the conditions around us. But much of that adaptability comes from something more powerful than the ability of an individual human to adjust.
Human capability has compounded across generations because learning did not have to be rediscovered every time.
Much of this adaptability rested on our ability to share ideas and learning beyond the individual who first developed them. Language and storytelling allowed knowledge to move between people and across generations. Writing then changed the durability of that inheritance. An idea no longer had to survive only through memory and retelling.
But the value was not simply that something had been written down. It was what could be carried forward in the writing: the reasoning, assumptions, evidence, derivations and context behind a conclusion. Later forms of distribution made that inheritance easier to reproduce, challenge and extend. The next person could see not only what had been concluded, but how and why, and therefore test it, disagree with it, refine it or build beyond it.
The text did not replace human interaction. It raised the altitude from which the next interaction could begin. In other words, the conversation did not have to restart from zero.
Einstein did not leave later generations only the shorthand E=mc². His 1905 work, including On the Electrodynamics of Moving Bodies and Does the Inertia of a Body Depend Upon Its Energy Content?, preserved the assumptions and reasoning behind claims that would fundamentally alter our understanding of space, time, mass and energy. Newton, centuries earlier, did not leave only a law of gravity or three laws of motion. Philosophiæ Naturalis Principia Mathematica set out the mathematical framework and chain of reasoning that later thinkers, including Einstein, could inspect, use and ultimately extend beyond.
The same is true outside science. Adam Smith did not leave later generations only the phrase “the invisible hand”. The Wealth of Nations developed a much broader body of reasoning around the division of labour, exchange, incentives, markets, capital and institutions, leaving later economists something they could agree with, reject, refine and build upon.
What survived was not simply the answer. It was enough of the thinking behind the answer for the next person to start higher.
That is an important part of how human capability has been able to compound, rather than each generation continuously beginning again. And the same principle applies inside an organisation.
What should an organisation inherit?
If what one team learns does not carry forward, the next team begins lower than it should.
If each leadership group has to reconstruct why an earlier strategic choice was made, the organisation retains the outcome but loses the reasoning that would allow that outcome to be understood. And the reality is that they often do not reconstruct it. They fill the gaps with assumptions, interpretations and recollections that may be only loosely substantiated by the thinking that shaped the decision in the first place.
Seen in that light, it is surprisingly difficult to justify how much valuable thinking organisations routinely allow to disappear.
The broader issue is therefore not only memory. It is inheritance.
Organisations produce this potential inheritance constantly through the normal course of work. Somebody reasons through a problem and reaches a conclusion. A leadership team weighs alternatives and chooses one for reasons that matter. A frontline employee discovers an exception the formal process did not anticipate. Research changes what a team believes. A failed experiment removes an apparently sensible option from future consideration.
The organisation is a knowledge-generating system. Every day it produces decisions, explanations, exceptions, interpretations and learning. The question is what survives.
Consider an important decision. Four different things may have been produced through the act of making it:
what was decided,
why it was decided,
what else was considered (and explicitly decided against),
and what had to be assumed for the decision to remain sound.
Those layers decay differently. The decision itself usually survives. The rationale may be partly remembered. Rejected alternatives disappear quickly, and assumptions often vanish first because, in the moment, they feel like context rather than knowledge.
Their future value runs in almost the opposite direction to the rate at which they decay.
There is an irony in that distinction between context and knowledge. So much of the AI discourse at the moment is about giving systems enough context, but we have not necessarily been precise enough about what good organisational context actually is. It is not simply more information around a question. Often, it is exactly this material: the assumptions that were being made, the alternatives considered, the constraints that mattered, and the reasoning that made a particular decision sensible at the time. Yet these are often the first things organisations allow to disappear.
An assumption may seem relatively unimportant on the day the decision is made. Six months later, when the market changes or a new technology appears, it can become the most important piece of the entire decision because it tells the organisation whether the original choice should still stand.
Without it, the organisation can see what it decided but struggle to understand whether the reasoning is still valid.
This is where the difference between movement and adaptation becomes clearer. If circumstances change and the organisation simply changes direction, it may well be demonstrating agility. If it can see which assumptions have changed, understand which parts of the previous reasoning still hold, remember which alternatives were already considered and why, integrate what has been learned since, and then change direction from that inherited position, something more valuable is happening.
The organisation is adapting from what it has learned rather than merely reacting to what it can currently see.
Choosing what matters is a skill
That does not mean every decision needs an encyclopaedic record. Nor should organisations attempt to retain every debate, meeting or passing thought. The standard is much more practical: what would another person need in order to act, decide or reason well without depending on the original people being present?
There is a capability question hidden inside that test. If the people making an important decision struggle to identify what another person would need to understand about it later, that is worth paying attention to. Were the assumptions clear? Were the trade-offs understood? Was it obvious which parts of the discussion actually shaped the choice, and which were simply part of getting there?
That does not necessarily mean the decision was poor. It may simply expose a capability most organisations have never deliberately developed.
We have historically asked people to make good decisions, not to make the thinking behind those decisions reusable. But as organisations become more distributed, and as AI carries more of the mechanical work, that ability becomes increasingly valuable.
Being able to recognise what matters, articulate why it mattered, and judge what another person will need later is a skill. Like any other professional capability, it can be developed. And doing so should make the person better not only at contributing knowledge, but potentially at the decision itself, because making the reasoning explicit forces assumptions, trade-offs and decision criteria into clearer view.
The answer, therefore, is not to retain everything because choosing is difficult. The choosing is the required skill.
That becomes especially useful across recurring cycles. When each strategy review, operating forum, product cycle or major decision leaves enough of its reasoning behind, the current cycle can be compared with what came before. Priorities that quietly disappear become visible. Assumptions that change while the decision remains unchanged can be spotted. Issues that keep returning stop looking new. Differences between stated intent and subsequent action become harder to ignore.
Cycle-over-cycle comparison turns organisational memory into something more useful than an archive. It allows the organisation to ask whether its reasoning is compounding or drifting.
Across enough decisions, patterns can also become visible in how the organisation makes trade-offs, which risks it protects against, which assumptions recur and what it consistently deprioritises. No single meeting contains “how the organisation thinks”. That pattern appears only when enough reasoning survives to be compared.
That is how an organisation can begin to scale more than instructions. It can start to scale judgement.
Augmenting the company, not just the person
AI makes all of this considerably more important, partly because it increases the rate at which work, decisions and outputs can now be produced. AI can reduce the time required to research, draft, compare, summarise and analyse. Agents will increasingly carry work across larger parts of organisational workflows. Those capabilities can create substantial value.
But faster work does not automatically create faster organisational learning.
One person can solve a problem brilliantly with AI and leave the reasoning inside a private working context. Another person can solve a closely related problem three months later without ever seeing the first. A marketing team can run an excellent series of conversion-rate optimisation tests and retain the winning variation, while losing the hypotheses they tested, what they learned about customer behaviour and why one approach outperformed another. Six months later, a new campaign team sees the result but has little of the thinking that would allow it to build intelligently on what came before.
The same can happen operationally. A team can work through an unusual customer or process failure, make a good judgement call and solve the immediate problem, even make the right change to the process, but leave the exception logic and reasoning in a meeting or message thread. When the same underlying pattern appears elsewhere, another team has to reason through it again from scratch.
In every case the work itself may have been excellent. What is missing is the mechanism by which the value of that work becomes available beyond the people who did it.
This is the difference between augmenting a worker and augmenting the company.
An organisation is not simply a collection of individuals. It is a network of people whose collective capability depends on how effectively they can align around shared context, combine different expertise, coordinate their contributions, make decisions that take account of one another, and leave the organisation stronger through the work they do together.
The important question is therefore not only how much faster AI allows work to happen. It is whether the company becomes more capable because that work happened.
Did an important decision become understandable beyond the people who made it? Did the assumptions survive? Did the failed approach become useful knowledge rather than something another team will rediscover later? Did new evidence change what the organisation knows? Can the next person begin from what has already been learned?
If the answer is no, activity may be accelerating considerably faster than organisational learning.
This is why I increasingly see Knowledge Architecture as part of the foundation for adaptability rather than simply as an answer to fragmented documentation or AI readiness. Its purpose is to preserve learning beyond the people who created it, make important reasoning and assumptions available to the next person, establish authority deliberately rather than by repetition, and help the organisation adapt without continuously fragmenting its own understanding.
That last part matters. Adaptability does not require an organisation to become loose, constantly rewrite itself or treat every new signal as a reason to pivot. A strong knowledge foundation should make change more coherent, not less.
It allows the organisation to understand what it currently believes and why, recognise when the conditions beneath that belief have materially changed, and integrate new learning without losing everything that still holds.
We do not need to know exactly what AI will look like five years from now to recognise the value of that capability. The environment will change. Some assumptions will turn out to be wrong. New technologies, competitors and ways of working will emerge. Strategies will have to evolve.
The more uncertain that becomes, the more important it is that each response leaves the organisation better positioned for the one after it.
Because in a rapidly changing environment, the advantage may ultimately belong less to the organisation that predicts the future correctly, and more to the one that can adapt by drawing on what it has already learned and applying that to what it does next.
Comments