Nobody Actually Wants AI. They Want the Work Already Done.

Nobody wakes up at six in the morning wanting artificial intelligence.

You wake up wanting the invoices already sent. The inbox already answered. The follow-up already written, the no-show already rescheduled, the weekly report already sitting where it belongs. Strip away the noise and the actual want is almost embarrassingly simple: work, finished, without you.

The industry heard that and answered with the strangest possible offer: here is a technology. Learn it. Prompt it. Subscribe to it. Build a strategy around it. And with that offer came a new kind of pressure that has nothing to do with your operation: everyone around you seems to be “doing AI,” and you’re quietly wondering what you’re supposed to want from it.

The honest answer is nothing. You don’t want AI — you want what comes out the other end of it. That sounds like wordplay until you watch what it does to buying decisions.

The AI-strategy trap#

Ask an owner why they’re evaluating AI right now, and listen closely to the answer. It’s rarely “my invoices go out late” or “our response time is costing us clients.” It’s usually some version of “we need to figure out our AI.” A peer brought it up at dinner. A competitor posted about their chatbot. An article announced that this is the year small businesses fall behind.

So businesses end up owning AI the way people own gym memberships in February. Acquired to relieve a pressure, used for two weeks, kept out of guilt, measured by nothing.

There’s a one-line test for whether any of it is real: count the hours of recurring work that no longer happen in your business because of it. Not hours “saved” in a demo. Recurring work that used to land on someone’s desk and now arrives finished. If that number is zero, call it what it is: AI expenses, not an AI strategy.

We’ve circled this failure from two sides already. Most small businesses see nothing from AI because they point it at work that was never the constraint. And even when they find the right work, they buy it like software instead of handing it off like delegation. But there’s a failure upstream of both, and it’s the frame the whole industry runs on: wanting AI at all. As a category. As a thing a serious business is supposed to have.

Start from the work, not the technology#

If nobody actually wants AI, then no sane buying process should start with AI.

“What can AI do?” is a useless question. The answer is “almost anything, somewhere between badly and brilliantly, depending on everything else,” and you can’t build a decision on that. The useful question is smaller and more uncomfortable: which recurring work in my business do I want to keep getting done without me, or my team, doing it?

That question has real answers. The invoices. The follow-ups that die in drafts. The intake emails answered on day three instead of day one. Write them down and you have something no vendor deck will ever give you: a shopping list written in your operation’s own language, where every line already has a definition of done and someone who currently pays for it in hours.

This is the entire logic behind how we work. Diagnose finds the recurring work sitting on your real constraint. Automate means that work starts arriving finished. Unlock is what the freed hours buy you. You’ll notice the word AI appears nowhere in those three steps. That’s not modesty. The technology is how the work gets delivered; it was never what anyone was buying.

What “already done” actually looks like#

At Académie Lavalloise, the 250-student school that runs on Avelle, nobody experiences artificial intelligence. Parents ask a question and an answer comes back the same day. The administration doesn’t attend AI trainings and doesn’t maintain prompts; sensitive communications show up as drafts, a human reads them, a human sends them. Twelve recurring processes run this way, an estimated 35 to 50 hours of work a week arriving finished instead of waiting for someone’s evening.

Nobody at that school wanted AI either. They wanted enrollment season to stop swallowing the office whole, and parent questions to stop sitting unanswered until Thursday. That’s what got delivered. The fact that an AI operations team is doing the delivering is an implementation detail, and it should be. When your bookkeeper closes your month, you don’t ask which software did the reconciling.

The question that replaces the strategy#

The next time the pressure shows up (a vendor, a peer, a headline about falling behind), swap the question:

Not what’s our AI strategy. Which recurring work do we want delivered finished, and who’s accountable for delivering it?

Everything sorts itself under that question. A pitch that opens with the technology is asking you to want AI, and you don’t. That’s not a gap in your sophistication; it’s the sanest instinct you have. A pitch that opens with your work, names what done looks like, and takes accountability for the result is describing something you’ve bought happily your whole career.

You were never in the market for intelligence. You’re in the market for finished work.

Name the work you want back#

Take ten minutes and write the list: every piece of recurring work you or your team touched this week that didn’t need your judgment, just your time. That list is your real AI strategy, and most owners have never written it down. If you want a second pair of eyes on it, a thirty-minute conversation is enough for us to tell you which lines can realistically arrive finished, and which ones can’t yet.

Nobody actually wants AI. Wanting the work already done is enough.

---

Matthew Lacopo leads client engagement and content at The Agensee, a managed AI operations firm based in Montréal serving SMBs across North America, which he cofounded with Anthony Lacopo.