Built, not Used
More than 260 software operators on the AI they shipped, the customers who never used it, and what the ones ahead do differently.
Why this, and why now
If you run a vertical market software company, the last year has been loud. Every conference, every board call, every competitor's release: AI, AI, AI. And underneath it, a quieter worry: "Are we moving fast enough?"
I wanted to know what was actually true for companies like yours. Not the enterprise, not the venture darlings. The operator who built something real over years, often without outside money, who now has to decide how hard to push. So we asked, and more than 260 of you answered. Some of it will steady your nerves. Some will sting. We operate more than 120 software companies ourselves, which makes this gap as much our opportunity as yours. Read it as an operator's-eye look at where you actually stand.
If you want to talk any of it through, I or anyone on our team is always happy to. This market is changing fast, and we spend our days in it.
Max RisenPresident, M&A, Banyan Software
Six things your peers told us
You shipped the AI. Most of your customers are not using it. More than half of operators see fewer than one in four customers touch what they built. Shipping was the easy part. This is the deployment gap, and it is the whole report.
Good news first: it is not about money or age. Bootstrapped operators pull ahead just as often as the PE-backed ones. What separates them is what they do, not what they have.
Ask operators what they would change, and most say the same thing: they would have moved sooner. Pace is the regret. But speed alone is not what wins.
When AI works, it shows up where you live: retention and revenue. The operators ahead see it lift both. Most of the field sees neither, yet.
The ones winning used AI to do more with the team they have, not to cut it. That single choice tracks with real revenue.
And they are not drowning in tools. Operators pick one or two and go deep, while the enterprise juggles five or more.
You are right to feel behind
If that worry sits in the back of your mind, you are not imagining it. And if you have been deliberately waiting, this has something for you, too. Either way, here is the part that should help.
What separates the operators ahead is not their size, their funding, or how long they have been around. It is none of the things you cannot change. Here is the share of each group who rank as AI leaders, grouped by the things you have no control over:
They did not just move fast.
They moved sooner, and measured what happened.
So the lesson is not simply go faster. Plenty of operators moved fast and blind, and that is exactly how you end up with AI no one uses. The ones who pulled ahead moved sooner than felt comfortable, and watched what happened closely enough to know what was working. Motion plus measurement. Everything that follows is what that looks like.
A spectrum, not a divide
Most operators are mid-pack. Almost everyone is bunched in the middle, with a smaller group pulling ahead. There is no club you are locked out of, just a spectrum, and a question of how far back you are. The posture here is opportunity, not defense: across the field, far more operators are leaning into AI than bracing against it. The ones pulling ahead see no bigger a threat than anyone else. What sets them apart is appetite. They are deliberate, and they move.
Where the gap shows up
Here is the same split, measured on the numbers you actually run on: revenue, growth, retention. None of these feed the maturity score, so they show the gap honestly. Each row compares the operators ahead (green) with the rest of the field (blue).
The clearest gap: 45% of the operators ahead report significant revenue from AI, against 9% of the field. Five times the rate.
Six behaviors, part 1 (1-3)
Move sooner, not recklessly 44% name pace
The clearest regret is pace. The leaders did not wait for certainty. The discipline that keeps sooner from becoming wasteful is the rest of this list.
Build AI into the product 45% vs 9% revenue
Put AI where customers touch it, not just the back office. The bar is higher: it has to be good, not just present. Watch error rates, keep a human in the loop where mistakes cost.
Measure it: activation and KPIs 51% gap
Where the deployment gap is won or lost. Not whether your team is busy with AI, but whether customers use it and whether it is any good.
Six behaviors, continued (4-6)
Put one person in charge of it #2 regret
The second most common regret was scattered tinkering, where AI is everyone's job and so nobody's. Pick your top use cases, give it a regular check-in, and put one person on the hook. It did not matter how operators set up ownership on paper. It mattered that someone owned it.
Optimize, do not cut 32% vs 7%
Use AI to get more from the team you have. Operators who kept headcount and raised output report significant revenue impact far more often than those who stood still.
Prove it, and make it provable 40% vs 6%
The payoff. Wire AI to results you can show, so your board, your team, and a future buyer can see it without you in the room explaining. Substance first.
Three stories that did not hold
Picking, not piling up
The clearest pattern is consolidation. Operators concentrate on one or two AI tools. Enterprises, by contrast, run five or more.
Asked, unprompted, for the one tool they would most hate to lose, operators named one by a wide margin: Claude. It led the next most-named tool by more than six to one.
This market moves fast, and the survey is a snapshot. Since we fielded it, OpenAI has pushed hard into enterprise coding, so we would expect Codex to climb in a rerun.
Share naming each tool as the one they would most hate to lose. Open-text and unprompted, so a multi-tool answer counts toward each tool named.
Where do you land against the field?
You have seen where the field sits. Now find your own spot on the same curve.
Pick all five to see where you land.
Built, not Used.
The operators a step ahead are not smarter or braver. They stopped waiting for adoption to happen and started managing it, and not one of them needed money you do not have or a company younger than yours.
Closing the gap is not about building more. It is about building what customers truly value, and then making sure they use it. Usage is the verdict on value: if they will not use it, the market is telling you it is not yet worth using. Where you start is yours to choose.
How we did this
We fielded this in 2026 among operators of business software companies, reaching them through operator and founder networks. More than 260 gave substantive answers, most of them founders or CEOs of vertical market software companies under $50M in revenue. Because respondents opted in, the sample leans toward operators already working with AI, so read the findings as a portrait of that group rather than a census of every software company.
We scored AI maturity on three things: how far a company has deployed AI, whether it tracks customer use, and how rigorously it measures impact. Leaders are the operators above a fixed maturity bar, which we hold constant so future editions can show whether the share clearing it grows. The outcomes we report sit outside that score, so they are associations rather than proof of cause, and the differences reliable enough to trust carry a quiet star.
Questions or comments? We would love to hear from you at info@banyansoftware.com.
