“The greatest obstacle to discovering the shape of the earth, the continents, and the oceans was not ignorance, but the illusion of knowledge.” — Daniel J. Boorstin, The Discoverers (1983)
A founder I know became noticeably more productive over the past year. Faster with proposals, faster with board papers, able to sit down at ten at night and think something through with a tool instead of waiting for the Monday meeting where he’d normally have done that thinking out loud with someone else in the room. By any reasonable measure, he’s better at his job than he was eighteen months ago.
Despite this, the business he’s leading looks much as it did before any of that started. The same bottlenecks show up in the same places. The same person is still making the same decisions, and the same things still only happen when he’s present to make them happen. Nobody in that business would tell you nothing has changed, because plenty has. It’s just that AI adoption in business, when you look closely at what actually changed and what merely felt different, doesn’t always touch the things that needed touching.
How It Actually Spreads
Nobody decided this, in the sense that boards usually mean when they say something was decided. There was no rollout, project plan, or line item that anyone signed off. A finance manager started using a tool for board packs because it saved her most of an evening every month. A salesperson started drafting proposals with AI tools for business because the first draft, the blank-page part, used to be the worst part of the job and now isn’t.
Within a year, half the business is using AI for something, and not one person asked permission first, simply because there was no one obvious to ask. In many businesses, by the time the leadership team realises AI is being used widely, it is already woven into daily work in ways that were never formally discussed.
This is worth taking seriously rather than waving away as a fad working itself out, because the gains in individual productivity are completely real. The finance manager genuinely is better at her job now, the sales person produces better proposals. The mistake, if there is one, isn’t in what they did. It’s in assuming that getting better at their jobs is the same thing as the business getting better at its job. Those two facts can both be true and have almost nothing to do with each other.
The Information No One Can Hold Together Anymore
There’s a point most growing businesses reach where no single person can comfortably hold every customer issue, every sales trend, every operational constraint and every financial signal in their head at the same time. Founders tend to keep trying anyway, often because they’ve always managed it before and can’t quite locate the moment it stopped being true. The issue here isn’t usually effort, it’s the limits of what one person’s attention can comfortably cover.
This is one of the places AI starts to look like a genuinely different kind of tool, not because it can write a paragraph or summarise a meeting, but because it can identify patterns across systems, reports and conversations that would otherwise stay scattered in different inboxes and different people’s heads. A founder who’s drowning in disconnected signals isn’t short of business information. They’re short of a way to bring it together. That’s not quite the same thing as helping one person work faster. Bringing it together with AI can change what the business can actually see.
Two Conversations, Same Sentence, Different Meanings
I had two conversations with two different founders, several months apart, and both used almost exactly the same sentence. “We’re using AI now.” Said with the same tone, the same mild pride, the same sense that something had shifted.
The first founder meant that three people in his business had got noticeably quicker at their own work. Reports that took a day now took an hour. The second meant something else entirely: a change in decision-making itself, a recurring decision that used to require him personally in the room, a pricing exception, say, or a judgement call on which customer complaint needed escalating, now got made the same way whether he was there or not. Same sentence. Almost opposite meanings underneath it.
The first changed how fast some individuals got their work done. The second changed who the business actually depends on to get it done at all, which is a rarer shift, and a considerably harder one to engineer on purpose.
What’s interesting is that neither founder seemed especially aware which kind of change he was describing, and both felt they were benefitting. The technology, in both cases, looked roughly the same from the outside. The results, though, were markedly different. It reminded me that the technology can work while the underlying business remains largely unchanged.
Why the Easy Win Wins
There’s a reason AI adoption in business spreads faster at the individual level than the organisational one, and it isn’t enthusiasm in one camp and resistance in the other. A person can open a tool tomorrow morning and start getting value from it without asking anyone’s permission, without agreeing anything with anyone else, without the business as a whole having decided a single thing. That’s most of why it spreads the way it does.
Making AI change how decision-making gets done, the kind of organisational leverage that actually compounds, requires the business to have already settled questions that many growing businesses leave surprisingly vague. The way important decisions get made matters long before technology enters the picture. So does clarity around decision rights and accountability when circumstances become less straightforward than expected. These are fundamental issues that any growing business needs to consider if it is to maintain the rate of growth.
McKinsey’s most recent global survey on AI adoption in business found much the same thing from a different angle: the organisations actually capturing value from AI are the ones that redesigned how the work gets done, rather than the ones that simply layered a tool onto a process nobody had thought to redesign. That’s the harder path, and most businesses, for entirely sensible reasons, take the easier one first.
The Founder Who Got Faster
The uncomfortable part of this discussion is that the founder may be the person who benefits most quickly. That is not surprising. Founders are often generalists by necessity, carrying a mixture of commercial judgement, customer knowledge, operational memory, people issues and financial concern that rarely fits neatly into one role. A tool that helps them draft, compare, analyse, summarise or test thinking can feel immediately valuable because it touches so many parts of their working life.
It’s not really about AI at all, but about what relief does to urgency. A founder who’s personally getting better at using these tools can quietly take the pressure off the very problem that most needed solving. If they’re drafting faster, thinking things through alone at night instead of needing other people in the room, the whole business starts to feel less stretched to them, even though nothing about who that business actually depends on has moved an inch.
They feel relieved. The dependency is exactly where it was. It’s entirely possible, and I suspect more common than anyone admits, for a founder to feel their business coped better this year, in a year when, by the one measure that actually counts, it became no less dependent on them than it was the year before. The technology that looks like leverage from where they sit, can actually be reinforcing the very thing it appeared to be solving, if nothing at the organisational level develops alongside it. I’ve watched this pattern before in businesses going through growth without structural change, where the founder’s own greater capability becomes the thing masking the absence of anyone else’s.
This is not a reason to avoid AI. It is simply a reason to be honest about what kind of progress is being made. There is a difference between a business becoming more capable and a founder becoming better equipped to compensate for the business not yet being capable enough, where visible performance improvements can disguise deeper founder dependency rather than reduce it.
What Nobody Decided
There’s a second cost to AI arriving person by person rather than being given a proper place in the business, and it has nothing to do with leverage at all. Nobody has decided what client information is fine to put into these tools and what isn’t, and in most SMEs, nobody has actually checked. Not because anyone involved is being careless. Because the same thing that made adoption fast in the first place, the fact that no one needed to ask, also meant no one was ever in a position to say no.
This pattern will be familiar to many growing businesses, whether or not technology is involved. Practices often emerge before structure catches up. The sales team develops its own way of pricing exceptions. Operations finds workarounds that were never formally approved. Managers make assumptions about authority because nobody has stated it clearly. AI may feel new, but the organisational behaviour around it is not. It is another example of authority existing through habit rather than clarity, where the organisation absorbs a practice long before it consciously designs for it. The risk sits quietly in the same gap the rest of this article has been describing.
Closing
There seems little doubt that AI is already helping many SME leaders and managers. The usefulness is too visible to deny. People are saving time, improving drafts, preparing more effectively, analysing more information and, in many cases, reducing the drag in work that previously consumed too much attention for too little return.
As AI adoption in business continues, the more difficult question is where those gains are actually accumulating. As with many forms of technology investment, technology often reveals more than it changes. In some businesses, they appear to sit mainly with individuals who have become better at doing what they already did. In others, the gains seem to be building organisational capability itself, changing how information is recognised, how decision-making is done, and how much depends on one or two people holding the whole picture together. Or perhaps, more often than either extreme, the answer sits somewhere uncomfortably between the two.
Most businesses are probably somewhere between those positions, which is why the question is not as simple as whether AI is useful. It clearly is. The more revealing question is whether its usefulness is strengthening the business, or simply making the existing pattern easier to carry for a while longer.
When you look at the AI tools now embedded in your business, what is actually becoming stronger: the organisation itself, or the people who were already carrying most of the weight?
If you enjoyed this article you can subscribe here to receive future articles.
—
0 Comments