Before you spend a dollar on AI, ask one thing: what's the one task your team complains about every single week? Not a chatbot. Not a receptionist.
If you own a business and know AI matters but don't know where to start, this is for you. I'm not a coder or a developer. I've spent enough time inside small businesses now, my own and a family one, to see the same mistakes over and over.
Somebody gets pitched an AI receptionist that never misses a call, or an AI visibility package, and they buy it because it sounds like doing something. It runs for a month. Nobody on the team uses it. Now the owner is telling a friend that AI automation is a scam.
The tools aren't fake. The problem is that buying the loudest pitch is answering the wrong question first.
The pitch usually sounds like this: "We'll put an AI receptionist on your phones, it never misses a call, sign here." Sounds great on the call. Nobody on that call asked what your team is actually missing calls about, or whether the problem is your phones at all. That's the pattern: a generic tool sold to a business that was never asked what it specifically needs.
The bottleneck was never whether the AI is capable enough. It's that nobody decided in plain terms what problem they were trying to solve before they bought something.
The fix isn't a smarter tool. It's picking one specific, annoying thing your business does by hand every week and starting there. Not the impressive thing. The annoying thing.
I've watched this play out with real examples behind it: a vendor invoice rebuilt by hand every week, and a spreadsheet formula nobody had checked in months. Small things. None of them are what you'd put in a pitch deck, and all of them are what cost you. (If you want the worked examples, read Claude Found the Excel Mistake That Overbilled a Customer.)
The impressive thing is picked by whoever is selling. The annoying thing is picked by the people doing the work. It is already known, because they complain about it. It is already measurable, because it happens every week. And it is already small enough to check: you can look at the output and know whether it is right.
That is the same pattern behind the mistakes I found in a small business I operate in. A vendor invoice that gets rebuilt by hand each week. A formula nobody had checked in months. Nobody would put either in a pitch deck. Both were costing money, and both were the right place to start.
Write down one sentence:
Every week someone on my team spends time doing ______ by hand.
That sentence is worth more than any pitch deck that lands in your inbox this month, because it's yours, it's specific, and it's true. Then take it through the steps under "What to do this week" below.
You do not need a meeting. Ask each person one question: "What is the one thing you do every week that you wish you did not have to?" Write down the answers. If three people name the same task, that is your starting point. If everyone names something different, pick the one that is repeated most often and easiest to check.
It took me nine months to get good at this, and it wasn't a weekend. I don't think it takes nine months for a business owner to get one real thing working, especially with someone helping who has already made the mistakes. But I won't hand you a number I can't back up.
If somebody tells you 30 days and a guaranteed result, ask them two things:
"Do you still use it?" is the question I'd want you to ask everyone who pitches you anything in this space. Not "does it work in the demo?" Whether the person who built it, or the business it was built for, is still opening it a month later.
If you want this looked at in your own business, comment with the one task you'd start with. If you'd rather do it yourself, the process I use to scope and build the first task is Forge on GitHub.
To see how I decide what's worth building and what to leave alone, read Best Mileage Tracker App? I Asked AI and It Said Buy One, where my own AI process found a free tool that already did most of the job, so I built a small piece around it instead.
I send what I'm building, what worked, and what broke. Free, no fluff.
Free. Unsubscribe anytime.I send what I'm building, what worked, and what broke. Free, no fluff.
Free. Unsubscribe anytime.I'm a marketer who builds his own tools. More at adamgarceau.com.