Every owner of a private company has now been told that AI will change the business. Very few have been told where, in what order, or how they would know whether it worked. The gap between those two conversations is where most of the money spent on AI disappears.
The test we apply is simple. If a change does not show up in the financial statements within a reasonable period, as lower cost, faster cash, higher revenue or better margin, it is not an operating improvement. It may be interesting. It is not what the owner is paying for.
What changed, and why it matters to a company of fifty people
Two things changed at once. The models became able to reason through a task rather than simply complete a sentence, and they became able to act: read a document, look something up, draft a reply, update a record, and hand the result to a person for approval. Together, those turn AI from a feature inside software into something closer to a workforce that can be configured.
That capability is available to a company with fifty employees on the same terms as one with fifty thousand. What the smaller company lacks is not access. It is the judgement about which work to hand over, and the discipline to keep a person accountable for the result.
The six places it reaches the financial statements
In roughly the order the return usually arrives:
1. Quoting, estimating and proposals
In most owner-operated companies a quote waits on one or two people who hold the pricing logic. Built from the company’s own history and pricing rules, a quote can be drafted in minutes and reviewed by a person before it goes out. The effect shows up as a shorter sales cycle, a higher win rate on work that was previously lost to slow turnaround, and pricing that no longer depends on who is in the office that day.
2. The back office
Receivables follow-up, coding payables, scheduling, routine reporting and the month-end pack are repetitive, rules-based and well documented in the company’s own records. They are the most reliable source of overhead reduction, and the easiest to measure, because the hours they consume are already known.
3. Sales and customer follow-up
Enquiries that are answered late, quotes that are never chased and dormant accounts nobody revisits are revenue the company has already paid to generate and then let go. AI that answers, follows up and reactivates in the company’s own voice, with a person stepping in at the right moment, recovers part of it.
4. The owner’s knowledge
The pricing logic, vendor terms, exceptions and judgement calls that live in the owner’s head are the single largest operational risk in most private companies, and the single largest discount a buyer applies at sale. Capturing them into systems the company can run on reduces the risk and, for an owner who may one day sell, raises what the company is worth.
5. Margin and decision support
Most private companies do not know, with any confidence, which customers, jobs or product lines lose them money. Margin reported continuously rather than reconstructed once a year changes pricing and customer decisions within a quarter.
6. Growth without proportional headcount
Service, onboarding, documentation and internal knowledge handled by AI allow revenue to grow faster than payroll. That is operating leverage, and it changes the growth story from “add people” to something a buyer will pay more for.
The order matters. The first items should be chosen because they pay for the rest, not because they are the most impressive in a demonstration.
Three mistakes that cost owners the most
Starting with the tool rather than the work
Buying a platform and then looking for something to do with it inverts the sequence. The right starting point is an honest account of where the hours and the errors are, and only then the question of which tool fits.
Adopting without a named owner
Every tool needs a person inside the company who is accountable for it before it goes live. Without one, the system drifts, nobody notices when it is wrong, and the business ends up with a new dependency instead of a removed one.
Measuring nothing
Without a baseline taken before the work starts, nobody can say afterwards what changed. A monthly measurement against that baseline, in hours, cycle time, margin or revenue, is the only way to know whether the investment paid, and the only defensible way to decide what to do next.
What it means for value
Because Turris also sells companies, there is a second discipline underneath the first. A company that runs on documented systems rather than on its owner, with margins a buyer can verify, is worth more, whether or not it is ever sold. Every tool adopted should be documented, assigned to a named owner and understood by the company, to the standard a buyer’s diligence team will one day expect.
The first conversation should not be about which model to buy. It should be about where the hours go, and which tools already in the building are going unused.
