You post the job. You wait. The three applicants who reply are not a fit, or already took something else by the time you called. Six weeks later the work is still sitting on the people who were already full.

Most owners across the lakeshore, orchard country, and the river valleys will recognize that loop. It is usually described as a hiring problem. Increasingly it is an arithmetic problem, and it is worth looking at the arithmetic before deciding what to do about it.

The pool got smaller. The work didn’t.

In June, the Michigan labor force stood at 4,907,000, which is 124,000 fewer people (down 2.5%) than a year earlier. The number of employed residents fell by nearly the same amount. Over that same year, payroll jobs in the state rose by 1,000. So there are roughly the same number of positions and noticeably fewer people available to fill them.

Part of the national drop is a statistical correction. The St. Louis Fed’s breakdown attributes about 43% of the decline since December to a January revision to the population estimates, about 16% to ongoing aging, and about 41% to real changes in behavior. So the exact size of the drop is worth holding loosely. The direction is not in dispute. State officials describe older workers retiring and not being replaced by younger ones, and a University of Michigan economist put it as demographics setting a speed limit on growth.

Capacity is not the same as headcount

When you cannot hire, the tempting answers are to work longer, turn work away, or let quality slip. All three cost money, and none of them show up as a line item. The overtime does. The quote that never got sent does not.

The other answer is to take low-value work off the people you already have. That is a different claim from “AI replaces staff,” and the evidence supports the narrower one. In new Census Bureau research, 66% of firms using AI rely on it solely to augment tasks, and AI-related employment decreases occurred in only 2% of firms. Earlier Census data from the Annual Business Survey found that the most common answer among AI-using firms was that the number of workers did not change overall.

Small firms are not where most of that adoption is happening yet. Census reports that between December 2025 and May 2026, AI use rose among firms with at least 20 employees but did not change significantly among firms with fewer than 20. The teams with the fewest spare hands are the ones least likely to have tried this.

Where the hours actually come back

The work that can move is the work nobody applied for in the first place: retyping, chasing, sorting, reformatting. It sits in the same four places in most small operations.

When it isn’t the answer

Some gaps are real people gaps. If the bottleneck is a licensed technician, a driver, or someone who has to be in the room, software does not fill the chair. It can clear the paperwork around that person so their day is spent on the work only they can do, but it will not do the work.

And if the hours you are losing are a handful a month, the fix should cost less than the hours. Do it by hand and spend the money elsewhere. We would rather say that on a call than after a project.

Start with one thing

The useful version of this is small. Ask the people who are stretched thinnest what they retype, chase, or re-check every week. Pick one. Get a working fix in hand early, in the tools your team already uses, and write it down so your people can run it without us. Then look at what the hours went toward.

You may still need to hire. The point is to make the hire you can find a better job, and to stop losing the one you can’t.

A note on the numbers: labor statistics and AI adoption surveys disagree with each other because they ask different questions and revise their own baselines. Every figure above links to its primary source so you can check the definition yourself. We keep a running Sources page grading each citation by its distance from original research.