Should I replace my staff with AI?

Probably not, and most small businesses using AI haven’t. The headlines are about big companies cutting staff, but in a Goldman Sachs and Babson survey of small business owners this year, 84% of the owners using AI reported no change to their workforce because of it. In the US Chamber’s 2026 report, 47% of small businesses said AI is creating jobs and 6% said it’s letting them cut. Replace the busywork inside each role instead, so your people spend their hours on the client work only a person can do, and only hire when growth actually needs someone.

Why does it feel like everyone is cutting staff for AI?

If you’ve read any business news in the last couple of weeks, it probably feels like every company out there is swapping people for AI. On October 26th, Reuters reported that Microsoft is offering voluntary buyouts to around 60,000 of its employees. That’s about a quarter of a company of roughly 223,000 people. At the same time, some of the biggest AI videos on YouTube this week are about free packs of 230 “AI employees”.

To put Microsoft in small business terms, a quarter of an eight-person agency is two people. So if you read that headline as an owner, it’s pretty easy to look at your own team and wonder if you’re supposed to be letting two of them go.

But the Microsoft story has very little to do with AI doing anybody’s work. Microsoft says AI isn’t what’s driving the buyouts, and it expects under 5% of the people who are eligible to actually take one. Reuters tied the offer to containing costs while the company pours money into AI, and Fortune reported that Microsoft’s yearly spending on property and equipment went from $64.6 billion to $115.9 billion. A small business isn’t spending on infrastructure to create and train AI. Most of these small businesses aren’t even spending enough money on AI itself to justify buying out their employees either.

And it isn’t only owners reading these headlines. The people on your payroll are reading the same ones. I’m watching this play out with one of my own clients right now, whose main employee is worried about their job being taken away in place of AI.

So, before you make a decision about anybody on your team, it’s worth looking at what small businesses are actually doing with AI. It’s close to the opposite of the headlines.

Sources: Microsoft offers buyouts to around 60,000 employees to contain costs (Reuters, via US News) · Microsoft’s 60,000-person buyout offer (Fortune)

What are small businesses actually doing with AI?

Last week, Goldman Sachs shared a survey that Babson College ran with graduates of its 10,000 Small Businesses program, and 2,473 owners responded. 86% of them use AI in their business, up from 56% in 2024. Of the owners using it, 84% reported no change to their workforce because of AI. And 73% of the owners using AI spend less than $100 a month on it.

The number to pay attention to is a smaller one. Only 15% of them say AI is fully built into their core operations. So most small business owners are using AI, very few have built it into how the work actually gets done, and almost none of them have cut anybody because of it.

The US Chamber of Commerce found something similar in its 2026 small business technology report. Two thirds of small businesses now use AI, and 47% say AI is creating jobs for them today, compared to 6% who say it’s letting them reduce headcount. That’s almost eight businesses adding jobs for every one cutting.

One thing to be upfront about is that neither survey breaks down how big the businesses are. But both of them are surveys of small businesses, and both of them land in the same place.

And even the big companies that are cutting aren’t clearly getting paid for it. Apollo’s chief economist, Torsten Slok, pointed out in late October that revenue per worker is rising for the seven biggest tech companies, but it’s flat for the other 493 companies in the S&P 500. Profit margins look the same way. For a small firm, flat revenue per worker looks like paying for AI subscriptions every month while the work still gets done the same way, by the same people, at the same speed. That’s where most of the owners in the Goldman survey are sitting right now. They’re using AI, but it hasn’t changed how the work gets done.

So the small businesses using AI mostly kept their people. The more useful question is what happens to the ones who went the other way.

Sources: Goldman Sachs Small Business AI Institute and survey (TNW) · 2026 Empowering Small Business Report (US Chamber of Commerce) · No signs of AI payoff outside the tech sector (Apollo)

Measure yours in 3 minutes: the free Task Audit scores how much of your recurring work could run without you.

What happens when you cut someone and the work is still there?

A lot of the companies that cut people for AI are already hiring them back. Robert Half asked US hiring managers about this, and 32% said their company eliminated a role primarily because of AI, then later rehired for that exact role. Orgvue found that 39% of business leaders had let people go because of AI, and 55% of those leaders now say it was the wrong call.

In a big company, a rehire is a line on an HR report. In a small business it’s a lot more personal, and a lot more expensive.

Say you run the eight-person agency and you cut your one coordinator because AI can draft the client updates now. You could give AI the rest of what they knew, too. Which client wants everything in bullet points, which one always pays late, where the files for every project live, all of that can go into a few files and an agent setup. But somebody has to write those files, set the agents up, and check what comes back every day, and the person who knows all of it best is the coordinator you just let go. So the day they leave, that job lands on you, on top of everything you were already doing.

And there’s a part AI can’t take at all. Your clients are used to talking to that person, and managing clients is still a human job. You might be the owner, and maybe you’ve met or sold the clients, but it’s the coordinator who is building that relationship and doing their work.

So, you decide you need someone after all. In my issue on handing a task to AI like a new hire, I went through what that costs. A hire costs 1.25 to 1.4 times their salary in the first year and takes three to eight months to get fully productive. And for those three to eight months, you’re the one training them, which is exactly the job you were trying to get out of.

Size makes this worse, too. One person leaving Microsoft barely shows up in the numbers. One person in a team of eight is 12% of your company, and probably a much bigger share of your client relationships.

So AI can take on a lot of what that person was doing. Somebody still has to run it and look after the clients, and when you cut the person to get rid of the busywork, you’re the one left doing both.

Sources: Employers who laid off workers citing AI are already starting to regret it (CNBC, on Robert Half and Orgvue)

What’s the difference between replacing a person and replacing their busywork?

This is happening in one of my clients’ businesses right now. They have one main employee for client and project management, and the owner has been building a lot of AI workflows to do the work around that employee. Day by day, the employee is worrying more and more about their job, because they think that eventually they won’t have anything left to do.

And as far as the busywork goes, they’re right. The employee thinks the admin and busywork is why they were hired. In the first place, sure, it probably was. But for the business to grow, that employee needs to be able to manage more clients, and for the most part, managing clients is a human-only job.

If the employee is spending their day on busywork, they won’t be able to manage more clients.

So adding the AI workflows is a way of giving the employee more responsibility, rather than taking their job away.

This is where small businesses are different from the companies in the headlines. Large businesses are the ones cutting their headcount, and small businesses don’t need to cut theirs. Smaller businesses should be using AI to stay leaner for longer. That basically means using AI agents to get more work done without hiring more people. That way you only hire when it’s absolutely necessary to grow the business, rather than hiring when you need somebody to do busywork.

What makes this possible now is how much of a role AI can actually take. Before, the only way to get busywork off somebody’s plate was to hand it to another person, which meant another full hire or a VA. Now a lot of it, like the status updates, the meeting notes, the onboarding emails and the invoice reminders, can be drafted by AI and checked by the person who used to write it. That person goes from doing the busywork to approving it, and the hours they get back go to the clients.

Replacing the busywork keeps the person, and the person is the one who can take on more clients.

What does a team that carries more look like?

My publishing client, Laura Di Franco, is the best example I have of this. Her company, Brave Healer Productions, runs on one manager and 15 to 20 contractors, with six to eight book projects going at once and over 200 tasks per project. I’ve written about what her AI Operating System does in an earlier issue, so I’ll keep this part short. Claude helps her run the project tracking, the launch days and the workshop replays, so she only spends her time checking what comes back, and she counts 25 to 30-plus hours a month back.

Her core team didn’t shrink when we built it. What changed was how much each of them could see and do without coming to her first. In her words:

“My entire team can be in the tracker and we can each see what’s been done.”

To be fair about it, she also estimates she saves around $2,000+ a month on VAs and subscriptions, so some outside help did go away. She’s honest about that side of it too. When I interviewed her for her case study, this is how she put it:

“Some human jobs are going away because these systems can do the thing. But what I really want people to know about that is, if you have a person that can run your system, they will be a thousand times more productive for you now.”

Right now she runs the whole AI Operating System herself. Her goal, in her words, is that eventually “someone else is sitting here running the whole thing.” So the end state she’s building toward is one person on her team running the AIOS and getting a lot more done, with the core team still in place.

For a five to ten person firm, this looks the same, just role by role. In my issue on onboarding a new client, I talked about the client experience person, the one in charge of onboarding everybody. Once AI is drafting the invoice, the welcome email and the nudges for whatever the client still owes, that person stops being the one sending everything and setting everything up. They’re the one approving it, and they’re able to onboard a lot more people.

You can do the same thing for every seat. Your project manager gets the status updates drafted from the call notes. Your account manager gets the follow-up drafted right after the meeting. Whoever handles admin gets the invoice reminders drafted for them. In each case, AI helps that person run the repeat part of their job, so they only spend their time reading and approving the drafts, and the rest of their week goes to the clients.

You keep the same team, and each person on it can handle more clients.

When should you actually hire?

Staying leaner for longer still includes hiring. You just hire when growth actually needs a person. If the client work, the sales calls and the relationships are more than your current people can carry, even after the busywork is gone, that’s when you hire. If the reason you want to hire is that everybody’s buried in admin, that’s a sign to take the admin out of their roles first.

The other thing I’d pay attention to is the employee in my client’s story. They’re worrying more every day, and nobody has explained to them what the workflows are actually for. If you start building AI into somebody’s role, tell them what it’s for before they work it out from the headlines. When the person knows the busywork is going so they can take on more clients, they’ll usually help you find more of it.

If you want to start this week, here’s how.

Firstly, write down what each person on your team does in a normal week, including you. Try to get every recurring task, not the ideal version of the week.

Next, go through each list and mark which tasks are busywork and which ones actually need a person, like client conversations and judgment calls.

Then pick the role with the most busywork and take the biggest piece of it out first, using AI to draft and that person to approve.

Finally, sit down with that person and decide where their hours are going to go before they get them back. That might be more clients, or the part of the job they never had time for.

Or, if you’d rather I just look at your business directly, I do a free discovery call.

Frequently asked questions

Why does it feel like everyone is cutting staff for AI? Because the headlines are about big companies. On October 26th, Reuters reported that Microsoft is offering voluntary buyouts to around 60,000 of its roughly 223,000 employees, which in small business terms is two people in an eight-person agency. But Microsoft says AI isn’t what’s driving the buyouts, and it expects under 5% of the people who are eligible to take one.

What are small businesses actually doing with AI? Most of them are using it and keeping their people. In a Goldman Sachs and Babson survey of 2,473 small business owners, 86% use AI, 84% of the owners using it reported no change to their workforce because of it, and only 15% say AI is fully built into their core operations. The US Chamber found 47% of small businesses say AI is creating jobs for them, compared to 6% who say it’s letting them reduce headcount.

What happens when you cut someone and the work is still there? A lot of companies end up hiring them back: 32% of US hiring managers in a Robert Half survey said their company eliminated a role because of AI and later rehired for it. AI can take on a lot of what that person was doing, but somebody still has to run it and look after the clients, and when you cut the person, you’re the one left doing both. A rehire then costs 1.25 to 1.4 times their salary in the first year and takes three to eight months to get fully productive.

What’s the difference between replacing a person and replacing their busywork? When you replace the busywork, AI drafts things like the status updates, the meeting notes, the onboarding emails and the invoice reminders, and the person who used to write them checks them instead. The hours they get back go to the clients. Replacing the busywork keeps the person, and the person is the one who can take on more clients.

What does a team that carries more look like? It’s the same team, with each person able to handle more clients. One publishing client’s core team didn’t shrink, and Claude helps her run the project tracking, the launch days and the workshop replays, so she only spends her time checking what comes back, and she counts 25 to 30-plus hours a month back. In a five to ten person firm, it works the same way role by role, with AI drafting the repeat part of each job and that person approving it.

When should you actually hire? Hire when growth actually needs a person, meaning the client work, the sales calls and the relationships are more than your current people can carry, even after the busywork is gone. If the reason you want to hire is that everybody’s buried in admin, take the admin out of their roles first. And if you start building AI into somebody’s role, tell them what it’s for before they work it out from the headlines.

Photo by Jud Mackrill on Unsplash