In July, small business owners told Congress that AI is helping them grow headcount, not cut it. The data backs them up — and it exposes the difference between treating AI as a replacement and treating it as a coworker.
On July 16, small business owners sat in front of the House Small Business Committee and told a story that cut against the headline everyone expects. AI was not thinning their teams. It was helping them grow.
That is not the narrative most people carry into a conversation about AI and jobs. The reflex is subtraction: adopt AI, cut roles, book the savings. But the operators actually running AI inside small companies keep describing something else — they are taking on more work, serving more customers, and hiring more people to do it.
The tension is real, and a recent Thryv survey captured it bluntly: a striking share of small business owners said that if forced to choose, they would rather add AI capability than make their next human hire. Read fast, that sounds like replacement. Read carefully, it says something different. It says AI has become the first capacity decision owners reach for — and when that capacity pays off, the human hire tends to follow, not disappear.
This is the gap between two ways of thinking about the same technology. One treats AI as a cheaper substitute for a person. The other treats it as a coworker that makes each person more valuable. The businesses growing headcount are almost always in the second camp.
The word you use for AI quietly decides how you deploy it.
If AI is a replacement, the goal is to remove a cost. You find a role, automate the tasks inside it, and count the salary you did not pay. The ceiling on that strategy is low by design — you can only save what you were already spending. Once the headcount is gone, the upside is gone with it.
If AI is a coworker, the goal is to add capacity. You find work your team cannot get to — the follow-ups that slip, the reports nobody has time to write, the leads that go cold over the weekend — and you put an AI coworker on it. Now the ceiling is your market, not your payroll. The human you would have replaced instead gets handed the higher-value work the AI surfaced.
The distinction is not semantic. It changes what you measure, what you build, and whether your team grows or shrinks.
A home services company decides AI will replace its front-desk coordinator. It buys a chatbot, points it at the phone line, and lets the coordinator go. Bookings hold steady for a month, then dip. The bot mishandles the messy calls — the reschedules, the angry customers, the permit questions — because nobody encoded how the business actually wants those handled. There is no one left who owns the outcome. The savings on paper turn into lost jobs in practice, and the owner quietly rehires three months later, now behind on both revenue and trust.
A permitting business puts an AI coworker on intake instead. It reads every incoming request, drafts the initial application, flags the ones that need a licensed human to review, and escalates anything ambiguous to a named person. The coordinator does not get cut — she stops drowning in data entry and starts handling the judgment calls and the client relationships. Within a quarter the business is processing more applications than it could before, so the owner hires a second coordinator to keep up with the growth the AI coworker made possible. Headcount went up. So did margin.
Same technology. Opposite outcome. The difference was whether the AI was pointed at removing a person or at extending one.
Three forces push AI-forward small businesses toward more hiring, not less.
Capacity opens up demand. Most small businesses are not sitting on excess staff. They are turning work away because they cannot keep up. When an AI coworker absorbs the repetitive load, the business can finally say yes to work it used to decline — and that new volume needs humans to deliver, sell, and support it. A plumbing company that could only quote ten jobs a week does not stay at ten jobs a week once an AI coworker handles the quoting; it grows into the demand it was already leaving on the table.
AI raises the value of judgment. The more routine work an AI coworker handles, the more a business's success rides on the decisions that are still irreducibly human: which customer to prioritize, when to bend a policy, how to handle the exception. Those decisions do not get cheaper as AI spreads. They get more valuable, which makes the people who make them more worth hiring.
Someone has to run the team. An AI coworker is not a set-and-forget appliance. It needs someone to define what it should never do without a human, to review its work, to adjust it as the business changes. The companies getting real value from AI are creating roles around it, not eliminating roles because of it.
None of this happens automatically. It happens when a business treats an AI coworker as a real member of the team — with a defined role, clear boundaries, and a human it answers to. That is the design decision underneath every hiring-up story.
The businesses that shrink after adopting AI usually made the opposite choice without realizing it. They bought a tool, pointed it at a cost center, and never designed the coworker relationship at all. The technology was identical to what the growing companies used. The intent behind it was not.
You do not get the "hire more people" outcome by accident. Here is how to set it up deliberately.
1. Start with the work you are turning away, not the roles you want to cut. List the tasks your team never gets to — the backlog, the follow-ups, the after-hours requests. That is where an AI coworker adds capacity without threatening anyone's job. Automating what you already do poorly just makes the failures faster; adding capacity where you have none grows the business.
2. Give the AI coworker a real role, not a task. A task is "answer this email." A role is "own first-response for inbound leads, qualify them against these criteria, and hand the qualified ones to a human within the hour." Roles compound. Tasks stay flat.
3. Define the boundaries before you deploy. Write down what the AI coworker can decide on its own, what it must confirm with a human, and what it must escalate. This is the single most important step, and the one replacement-thinking skips. Clear boundaries are what let you trust an AI coworker with real work — and what keep the human in the loop where judgment matters.
4. Point your people at the work the AI surfaces. When the AI coworker handles intake, your coordinator moves to relationships. When it drafts the reports, your analyst moves to interpreting them. Redeploy, do not remove. The whole return on an AI coworker comes from what your humans do with the time it gives back.
5. Create the role that runs it. Someone needs to own the AI coworker's output, review its work, and tune it over time. On a small team this might be part of an existing job. As you grow, it becomes its own. Either way, plan for it — an unmanaged AI coworker drifts, and a drifting coworker destroys trust fast.
Q: Does using AI mean I will need fewer employees? A: Not for most small businesses. The companies adopting AI fastest are generally hiring more people, because AI absorbs the work they were turning away and creates new capacity that humans deliver on. Replacement is possible for narrow, fully repetitive roles, but it caps your upside at whatever you were already spending.
Q: What is the difference between an AI coworker and an AI tool? A: A tool waits for you to operate it. An AI coworker holds a role — it owns a slice of work, operates within defined boundaries, escalates when it should, and runs continuously. The coworker frame is what produces the "hire more people" outcome, because it adds capacity rather than just speeding up a task.
Q: Isn't it cheaper to just replace a role with AI? A: On a spreadsheet, sometimes. In practice, replacement often removes the person who owned the outcome, and the messy edge cases the AI cannot handle become nobody's job. Most businesses that cut first end up rehiring. Adding capacity where you had none is the more durable play.
Q: What is the first job I should give an AI coworker? A: Pick the highest-volume work your team cannot keep up with and that follows clear rules most of the time — lead intake, first-response, follow-up sequences, routine reporting. Give it a defined role and clear escalation boundaries, and keep a human on the judgment calls.
Q: How do I keep an AI coworker from making bad decisions? A: Define its boundaries explicitly before deployment: what it can decide alone, what it must confirm, what it must escalate to a named human. Review its work regularly. Structural boundaries — where the AI structurally cannot act outside its lane — beat hoping it behaves.
If you're ready to stop using AI tools and start running a real team of AI coworkers, Associates AI Teammates gives you a 14-day free trial with no credit card required. Start your free trial at associatesai.team.
Written by
Founder, Associates AI
Mike is a self-taught technologist who has spent his career proving that unconventional thinking produces the most powerful solutions. He built Associates AI on the belief that every business — regardless of size — deserves AI that actually works for them: custom-built, fully managed, and getting smarter over time. When he's not building agent systems, he's finding the outside-of-the-box answer to problems that have existed for generations.
More from the blog
OpenAI's ChatGPT Work runs a task on its own for hours, then hands you a finished file. That sounds...
A July 2026 Pax8 survey found nearly one in three AI-using small businesses is frozen between testin...
A new open-source benchmark found that a dynamic AI router completed a coding task run for $13.21, c...
Want to go deeper?
Get started today. Hire your first Teammate in minutes and put it to work on what you're reading about.
Get Started