Eighty-six percent. That is the share of US small businesses now using AI, up from 56 percent two years ago, according to Goldman Sachs' October 2026 small business report. Read quickly, it sounds like the story is over. Nearly everyone has arrived.
Then comes the second number. Among the businesses already using AI, only 18 percent say they have fully integrated it into the way the business actually runs. Roughly four in five are somewhere in the middle: a few tools open, a few habits formed, nothing quite settled.
That gap is a simplicity problem before it is a skills problem. Wide use means many tools, many prompts and many small decisions made fresh every day. Deep use means one piece of work that now runs the same way each time, with fewer steps, fewer choices and nothing to rediscover on Monday morning.
------------- Context -------------
Most of us have been told, in one form or another, to try more. Try the new model, try the new assistant, try the feature that arrived this week. It is sensible advice for the first month. After that, it quietly turns into a habit of sampling.
Sampling feels like progress because every trial produces something. A draft appears. A summary arrives. A list gets sorted. But none of it connects to anything else, so each result stands alone and each new task starts with the same questions: which tool, which wording, what do I give it, who looks at it afterwards?
The hidden cost is not any single choice. It is the number of choices. A solo owner who uses AI in six different places, in six slightly different ways, is making dozens of small decisions a week that never settle into a routine. That is the opposite of simple, however modern each individual tool feels.
The better lens is depth. Depth means taking one repeatable piece of work all the way through: where the input comes from, what AI does with it, who checks it, and where the result goes. Once that path is written down, it stops being a decision and becomes a route.
This matters because the scorecard for simplicity is not how many tools we use or how impressive the demo looked. It is how many decisions we no longer have to make. A deep workflow removes them. A wide spread of experiments multiplies them.
------------- Wide Use Multiplies Decisions, Deep Use Removes Them -------------
One of the quietest costs of trying everything is that every use is a fresh negotiation. Which assistant do we open? What do we paste in? How do we word the request? Do we trust the answer or check it? None of these questions is hard. Together they are exhausting.
When we take a single workflow deep, most of those questions are answered once and then retired. The tool is chosen. The starting material is fixed. The wording is saved. The check is named. What remains is the work itself, which is the part we wanted help with in the first place.
Consider a consultant who sends a weekly client update. Sampling looks like this: some weeks she pastes her notes into one assistant, some weeks into another, rewrites the opening each time, and spends ten minutes deciding whether the summary is good enough to send. Going deep looks like one saved structure, one place the notes live, one two-minute read-through against a short checklist. Eight decisions become two.
That is the simplicity gain in plain terms. The update still takes some effort, but the effort is spent on what to say, not on how to get started. And because the route is the same every week, it gets faster without anyone trying to speed it up.
------------- Integration Is Mostly Writing Things Down -------------
It is tempting to imagine that the 18 percent who have fully integrated AI have better tools, bigger budgets or more technical staff. More often, integration looks far less glamorous. It looks like a short written description of how a piece of work now gets done.
That description does not need to be long. A half page is enough: what triggers the work, what goes in, what AI is asked to do, what a person checks, and where the finished result is filed. Written once, it can be followed by the owner on a tired Friday, by a new team member, or by a contractor covering a holiday.
Without it, the workflow lives in one person's head and one person's chat history. It works while that person remembers, and it quietly stops working when they are busy. That is why so many businesses sit in the middle: the habit exists, but nothing holds it in place.
A small team that writes down its client onboarding route, for example, removes the repeated question of how this one should be handled. New clients start from the same page. Nobody asks around. Nobody reinvents the opening email. The team spends its attention on the client, not on remembering the process.
The simplicity win here is clarity. There is one path, it is visible, and anyone can follow it without asking.
------------- Training Works Better When It Has Somewhere to Land -------------
The same Goldman Sachs survey found that 73 percent of small businesses say they would benefit from more training and resources to put AI to work. That is a sensible thing to want. It is also easy to spend badly.
Training that covers everything produces knowledge with nowhere to go. We come away having seen twenty things, and we apply none of them, because none of them is attached to a piece of real work. A week later the notes are unopened.
Training attached to one workflow behaves differently. We learn exactly what that route needs, apply it the same afternoon, and see whether it removed anything. If it did, it is kept. If it did not, it is dropped without ceremony, and no one feels they have failed a course.
Picture a small business owner who spends one hour learning how to turn voice notes into a tidy quote, then uses it for every quote that week. The hour has a destination. Compare that with an hour of general tips, which leaves the same owner with more options and no more clarity.
So training is not the opposite of simplicity. Unfocused training is. When learning is pointed at a single route, it removes steps. When it is pointed at everything, it adds them.
------------- A Deep Workflow Frees the Rest of the Week -------------
There is a final benefit that is easy to miss. Once one route is settled, it stops asking for attention. It sits in the background and runs, and the mind that used to hold it is free for something else.
That is what simplicity feels like in practice: not fewer features, but fewer open loops. A business with three deep workflows and nothing else can feel calmer than one with fifteen shallow experiments, even though it technically does less with AI.
It also makes the next step easier. When the first route is solid, adding a second is a matter of copying a pattern we already trust. We stop asking what else AI can do and start asking what else we would like to stop thinking about.
The 18 percent are not ahead because they use more. They are ahead because less of their week depends on remembering how.
------------- Practical Moves -------------
First, choose one piece of work you repeat at least weekly, such as a client update, a quote or a meeting follow-up, and decide that this is the one you will take all the way through before trying anything else.
Second, write the route on half a page: what starts it, what goes in, what AI is asked to do, who checks it, and where the result is kept. Five lines is plenty, and it removes most of the daily deciding.
Third, retire the spare options. If you currently use two or three tools for this job, keep the one that already fits best and stop opening the others for it, so the choice is made once instead of every time.
Fourth, run the route for two weeks without changing it, then note which steps still feel heavy. Fix only those. A route that stays the same for a fortnight becomes a habit rather than an experiment.
Fifth, point any training, tutorial or learning time at that route alone. If something you learn does not remove a step from it, set it aside for later and keep the path short.
------------- Reflection -------------
The headline numbers about AI adoption are true and slightly misleading. Almost everyone is using it. Very few have settled into it. The difference between those two states is mostly the number of decisions still being made by hand, every day, from scratch.
That is why depth is a simplicity idea. A workflow taken all the way through replaces a cloud of small choices with a single route. The tool matters less than the fact that the path is fixed, written down and easy to follow.
In the end, the shift worth making is small. Not more AI in more places, but one place where it finally stops being a question. When that happens once, the habit of doing it again becomes easy, and the week gets lighter without anything dramatic being added.
Which piece of your work do you repeat often enough that it deserves a written route, and what is stopping you from writing it?
Where are you currently making the same small AI decisions again and again, and what would it feel like to make them once?
If you could only take one workflow all the way through this month, which one would give you back the most clarity?