AI-generated article. This article was researched and drafted using AI tools and published automatically, and its featured image was generated by AI. Facts are drawn from the sources cited in the text.
Headlines about machines replacing workers usually get written from inside large companies, where a hiring freeze is announced on a quarterly call and felt by thousands of people at once. Main Street rarely shows up in that story. New payroll research from Gusto, published on September 10, 2026, puts small firms back into it, and the finding runs the opposite direction from the one most people expect. On the question of AI and small business hiring, the businesses that adopted the technology grew their teams faster than the ones that did not.
What the payroll data says about AI and small business hiring
Gusto compared businesses on its payroll platform that had adopted AI against a comparable group that was aware of the technology but had not adopted it. Adopters grew headcount about 7% more than non-adopters in the year following adoption. The gap was not a one-time bump. It widened steadily, from roughly 4.5% to 6.8% across four quarters, which is the shape of a trend rather than a blip.
The smallest firms gained the most. Companies with fewer than ten employees grew their teams about 10% more than comparable non-adopters. For a nine-person business, that is one extra person on the payroll within a year. Health care and social assistance led every sector, with dental offices and mental health practices under ten employees growing headcount nearly 20% more than their non-adopting peers.
Who got hired matters more than how many
This is the part worth sitting with. The new hires were not prompt engineers. Gusto found that adopters added teachers, technicians, cooks, front-desk staff and care providers. Dental hygienists. Therapists. People who do the work the customer actually pays for.
Nich Tremper, senior economist at Gusto, framed it as a shift back toward expertise. Small teams have always had people doing tasks well outside their job description, and AI may be handing some of that overflow work to software so the humans can return to what they were hired for. If that reading holds, the mechanism is not replacement. It is capacity. Owners got time back, spent it on serving more customers, and then needed more hands to serve them.
Where the evidence gets thinner
Now the other side, because the study does not carry as much weight as the headline suggests.
The sample is not random. It covers Gusto customers who responded to a survey, which means businesses already organized enough to run modern payroll software and inclined enough to answer questions about technology. Axios flagged this directly in its coverage. Firms that adopt AI early may simply be the firms that were already growing, and growing businesses tend to buy new tools. Correlation and causation are genuinely hard to separate here.
The wider evidence is also much less flattering. MIT’s NANDA research on enterprise AI found that only about 5% of generative AI pilots reached measurable profit and loss impact. A 2026 NBER survey of nearly 6,000 executives across four countries found that 89% reported no measurable effect on their firm’s labor productivity over three years. Most projects still stall somewhere between the demo and daily use.
Adoption itself is also far from universal. The Census Bureau’s Business Trends and Outlook Survey put US business AI use at about 19.8% in May 2026, with firms under five employees sitting below 20%. Roughly 82% of the very smallest firms say AI simply does not apply to them. That is not resistance. That is a reasonable reading of a tool nobody has shown them how to use in their own context.
Two true things at once
Both of these hold. Among small businesses that adopted AI and kept using it, the payroll record points toward growth rather than contraction. And most small businesses have adopted nothing, most projects that start do not finish, and the productivity gains promised at the macro level have not shown up in the aggregate numbers yet.
The more useful question is not whether AI creates or destroys jobs. It is what a specific business does with the hours it gets back. The Gusto data is consistent with owners reinvesting that time into customers. It is equally consistent with owners who never got any time back at all, because the tool was bought, never really learned, and quietly abandoned a few weeks later.
One thing to try this week
Pick the single task that eats the most of your week and is not the thing customers pay you for. Chasing invoices. Appointment reminders. Writing the same three emails again. Just one. Spend two hours seeing whether a tool can take half of it off your hands, before committing to anything with a recurring monthly cost.
That is a small test, and small is the point. The businesses in the Gusto data did not run transformation programs. They adopted something, kept using it, and the effect turned up a year later in payroll records.
Whether the wider economy ends up looking like the Gusto sample or like the NBER one is still open. What the data does suggest is that the freed-up time is the real asset, and what happens to it is a decision rather than an outcome.
