AI Replaces Busywork, Not People: Here's the Line

This piece is about that line: where it falls, why it matters most for small teams, and how to use it instead of fearing it.

Where the Fear Comes From, and Why It’s Aimed at the Wrong Target

The fear makes sense, because of how we describe jobs. We say “she does invoicing” or “he runs the front desk,” as if the job is the task. So when a tool can handle the task, it sounds like the job is gone.

But that’s not what a job is. Pull any role apart and you find two very different kinds of work tangled together. There’s the repetitive, rules-based part: copying data, scheduling, chasing, formatting, the same email for the hundredth time. And there’s the part that needs a human in the room: judgment, persuasion, care, knowing when the rule shouldn’t apply. We bundle them under one title and call it a job. AI doesn’t see the title. It only ever reaches for one of those two piles.

A 2023 McKinsey analysis put a number on it: current AI could automate activities that absorb 60 to 70% of the time employees spend at work today, not 60 to 70% of jobs, but of the time inside them. That gap between “tasks” and “jobs” is the whole story. The tasks are going. The jobs are being rebuilt around what’s left.

The Line: Busywork vs. Judgment

Here’s the line you can draw through almost any role:

Busywork is anything where the right answer is determined before you start. Judgment is everything where the right answer depends on you.

Busywork is repetitive, rules-based, and predictable. Move this number into that cell. Send the reminder when the date is near. Tag the email by topic. There’s a correct output and a clear path to it, which is exactly what software is good at. Nobody grows up dreaming of doing it, and nobody on your team will miss it.

Judgment is the opposite. It’s the call you can’t write a rule for. Should we extend this client credit? Is this patient more anxious than they’re letting on? Does this campaign actually sound like us? It draws on context, relationships, taste, and accountability, the things that don’t reduce to a formula because the right answer changes with the situation.

AI lives almost entirely on the busywork side of that line. It’s a spectacular engine for “the answer is knowable, just tedious to produce,” and a poor substitute for “the answer depends on a human who’s paying attention.” Keep that line in your head and the question stops being will AI take this job? It becomes the far more useful how much of this job is busywork, and what could this person do with that time back?

What This Looks Like in Three Real Jobs

The line is abstract until you lay it over an actual role. So let’s do that with three you’ll recognize.

The One-Person Finance Team

In a lot of small businesses, “finance” is one person and a heroic amount of patience. Picture their week. Invoices arrive in five formats and get matched by hand against orders. Receipts are typed into a spreadsheet. Numbers are reconciled. The same “just following up on invoice #1234” email goes out again and again. Most of that week is busywork: clearly knowable answers, just slow and mind-numbing to produce.

Hand that pile to AI and the matching, the data entry, the reminder emails, the first-pass categorization all get absorbed. What’s left isn’t nothing, it’s the part that was always the actual job. Deciding whether to chase a good client harder or give them room. Spotting that cash is about to get tight three weeks out. Knowing which vendor relationship is worth protecting. That person didn’t get replaced. They got promoted out of data entry and into the judgment they were hired for in the first place.

The Clinic Front Desk

A clinic receptionist is doing two jobs at once, and only one of them is replaceable. Job one: scheduling, reminders, confirming insurance details, answering “what are your hours?” for the thousandth time, playing phone tag to move an appointment. Pure busywork: knowable answers on repeat.

Job two is the patient standing at the desk who’s scared about a diagnosis and needs a human being to slow down and be kind. No model does that. When AI takes the scheduling, the reminders, and the FAQ off their plate, it isn’t emptying the role, it’s clearing away everything that kept them from the part of the job that only a person can do. The desk gets more human, not less.

The Two-Person Marketing Team

A small marketing team lives in a permanent backlog. Resize the asset for six channels. Reformat the blog into a newsletter, then into five posts. Schedule everything. Pull last week’s numbers into a deck. It’s a treadmill of production tasks, and AI is built for exactly this. Drafting, repurposing, resizing, scheduling: handed over, the backlog shrinks fast.

But notice what AI can’t hand back: whether the idea is any good. What the brand actually sounds like when it’s being honest. Which risk is worth taking this quarter. Strategy and taste are judgment calls, and they’re where a two-person team either punches above its weight or doesn’t. Take the production grind away and you don’t have a smaller team. You have the same two people finally doing the work that moves the business.

Why Small Teams Have the Most to Gain

You’d think the big enterprise with a thousand employees benefits most from AI. It’s usually the opposite, and the reason is simple math.

When busywork eats ten hours of a 200-person department’s week, it’s a rounding error, there’s slack everywhere. When it eats ten hours of a five-person company’s week, that’s a person and a half, and it’s strangling the work that actually grows the business. Small teams feel busywork more sharply because there’s nowhere for it to hide. So when it lifts, the relief is proportionally enormous. The freed hours don’t disappear into a bureaucracy. They go straight back into selling, building, and serving customers, because in a small team, everyone is close to the work that matters.

There’s a second advantage too: small companies have less to unwind. No fifteen-year-old approval process, no committee that owns the spreadsheet, no legacy system everything is duct-taped to. You can redraw a role around its judgment in an afternoon, not a fiscal year. That nimbleness is exactly the leapfrog opportunity in front of leaner businesses right now, including across Latin America, where many companies are building their processes for the first time and can build them automated from day one.

The Honest Part: Some Roles Really Do Change

This is where most “AI won’t take your job” pieces get dishonest, so let’s not.

If a role is only busywork, if every hour of it sits on the knowable-answer side of the line, then yes, that role changes, and pretending otherwise helps no one. The honest response isn’t denial. It’s redesign. The same automation that hollows out a purely repetitive role is what funds a better one: the person who spent the day on data entry is exactly who you want learning to run the automations, handle the exceptions software can’t, and own the customer relationships that just got more time. The work doesn’t vanish. It moves up the line, toward judgment, toward people, toward the things that were always worth paying a human to do.

The companies that handle this well treat AI as a reason to invest in their people, not an excuse to stop. The ones that handle it badly will cut a head, pocket the saving, and quietly lose the judgment that head was carrying. The line between busywork and judgment isn’t just a way to think about AI. It’s a way to decide what kind of employer you want to be.

What to Do Monday Morning

You don’t need a strategy deck to start. You need to find your busywork. Try this:

1. Pick one role. Your own counts.
2. List a week of it. Every recurring task, honestly.
3. Draw the line through the list. Mark each task: is the right answer knowable before you start (busywork), or does it depend on a human paying attention (judgment)?
4. Add up the busywork hours. That number is your opportunity, the time AI could hand back.
5. Ask the better question: what would this person do with those hours if they never touched the busywork again?

That last answer is almost always something the business has been starving for. That’s the whole point. AI isn’t here to remove the people. It’s here to remove the reasons they can’t do their best work.

Frequently Asked Questions

Will AI eliminate jobs at my company?

It will eliminate tasks, the repetitive, rules-based busywork inside jobs. Whether that becomes lost jobs or upgraded ones is a decision you make, not something AI decides for you. The businesses that redesign roles around judgment keep their people and get more from them.

How do I know which work is safe to automate?

Draw the line: if the correct answer is knowable before the work starts, it’s a candidate for automation. If the right answer depends on context, relationships, or accountability, keep it human. Most roles are a mix: automate the first kind, protect the second.

Isn’t this just a nicer way of saying “do more with fewer people”?

No, it’s “do more with the same people.” The goal isn’t to shrink the team. It’s to stop spending your team’s hours on work that never needed a human, so those hours go to work that does.

We’re a small team with no AI experience. Where do we even start?

Start with one painful, repetitive task, not a grand plan. Map it, automate it, and feel the time come back. That first win shows you the line clearly, and the next steps get obvious from there.

Keep the People. Lose the Busywork.

The fear isn’t wrong to exist, it’s just aimed one level too high. AI doesn’t come for the job. It comes for the busywork inside it, and it leaves behind the part that needed a human all along. For a small team, that’s the best trade there is: less of the work nobody wanted, more of the work only your people can do.

Want help finding the busywork in your business, and freeing your team to do their best work? Talk to our team → and we’ll help you draw the line.