AI for Small Business: The Smartest Person in the World Can't Tell You What to Ask

Picture the smartest person in the world sitting in the chair next to you, and you can ask them anything you want, all day, for twenty dollars a month. That's roughly what happened.

The part nobody mentions is that everyone else got the same chair, the same person, and the same twenty dollars.

What can AI actually do for a small business?

AI can draft, summarize, research, analyze and answer questions faster than anyone you could hire. What it can't do is tell you which question matters. That judgment comes from having carried a payroll, and it's the part AI hasn't made cheaper. Start with a job you already do badly.

Everyone just got handed the same brilliant assistant

I use it the way you'd use tools in a garden. You need a shovel to dig a hole and a rake to move leaves. Specific tools, specific jobs. It's not a magic wand and it won't fix a business that's broken for ordinary reasons.

But the assistant is real. It reads faster than you, it doesn't get tired, and it will explain the same thing four different ways until one lands. That's not nothing.

Here's the problem with it being real. Your competitor down the road has the identical assistant. So does the guy who just started last Tuesday. Whatever edge you think you're buying, you're buying it off a shelf that everyone can reach.

The questions are the scarce thing now

AI made answers free. It didn't make questions free.

Asking a good question about your own business requires knowing which number is about to become a problem. That knowledge doesn't come from reading about business. It comes from having been responsible for one.

What thirty percent means when payroll is Friday

My first company was in the late eighties. We manufactured rollerblade wheels, clothing and accessories, with distributors across Asia and Europe and retailers across the United States.

We were the small guy. There was a much larger wheel manufacturer out of Huntington Beach going after an OEM deal with K2 Skates, who were owned by K2 Skis up on Vashon Island. We won it. Took the whole thing, including all their European and German distributors, because we had street credibility and they didn't.

Great story. Here's the one underneath it.

I couldn't read a profit and loss statement. I couldn't read a balance sheet. I did not understand my own margins on products I had designed and priced myself.

What I eventually worked out, the hard way, is that the floor is thirty percent. Below that there isn't enough left to pay your people or your rent, and it doesn't matter how good the product is or how big the logo on the purchase order is.

We made payroll. The K2 business put us on the map, got the attention of some national companies, and the company was ultimately bought by First Team Sports, a publicly held outfit in Anoka, Minnesota. It worked out.

But I didn't learn thirty percent from a book. I learned it because the alternative was telling people there was no check this week. Nobody can hand you that. Not a course, not a consultant, and not the smartest person in the world sitting in the chair next to you.

Three kinds of people are writing about AI right now

People who play with the tools. People who know marketing. People who have run a business and carried the number.

Nearly every article you'll read about AI for small business comes from the first group. A few come from the first two. Almost none come from all three.

That's a filter you can use on everything you read this month, including this page. Before you take advice about where AI belongs in your company, work out whether the person writing it has ever been responsible for the thing they're pointing AI at.

Somebody who has never made payroll can tell you what a tool does. They cannot tell you whether it's worth doing.

AI didn't change the order of operations

Audience. Strategy. Plan. Test. Then tactics.

That order hasn't moved in decades and AI didn't move it. What changed is that tactics got cheap, so people skip the first four and start at the fifth. You can now produce forty blog posts in a weekend for a market you haven't defined, with a message you haven't tested.

Most wasted AI spend is an old mistake with a new invoice. It's the same mistake companies made buying a website they didn't need or a campaign aimed at nobody in particular. If the strategy underneath it is missing, faster execution just gets you to the wrong place sooner.

That's the bright shiny object problem, and it's older than I am.

Where AI for small business actually starts, and it isn't a tool

Pick one job you already do badly and expensively. Not the most exciting one. The one that costs you and that you keep putting off.

For most owners that's follow-up. Somebody inquired eleven days ago and nobody called them back. Or it's reading things you don't have time to read: contracts, reports, the sixty-page RFP that came in Friday.

Then ask better questions about it than you would have asked a year ago, and verify what you get back. The five jobs it genuinely does well are a decent starting list, and if you want the comparison, the honest answer on which one to use is that it barely matters.

Measure one number for thirty days. Not activity. A number that shows up in the business.

What it still can't do for you

It can't know your margin.

It can't tell you which customer is worth keeping and which one is quietly costing you money every month.

It can't make a call that costs somebody their job, and it can't sit with that afterward.

I spent a year putting AI to work across four brands at a $50M manufacturer. Content production time dropped thirty-eight percent, which was the number everyone wanted. The thing I remember is that three months in, all four brands had started sounding like the same company. That whole story is here. No tool caught it. A person reading the output caught it.

So the first question isn't which AI to buy. It's what you already know that the machine doesn't, and whether anyone in your company is asking it. If nobody is, that's the gap, and it's the one a fractional AI advisor is actually for. If you're still working out what that role even is, start here instead.

If you want to see what that looks like taken all the way, I built a working CRM in eight days for thirty dollars, and the interesting part is what happened after it was built.

Go find the job you keep putting off. Ask it a better question this week.

• AI made answers free. It did not make questions free, and question quality comes from having been responsible for a business.
• Everyone got the same assistant at the same price, so the tool itself is not an advantage.
• The order of operations did not change: audience, strategy, plan, test, then tactics. Most wasted AI spend is skipping to tactics because they got cheap.
• Start with one job you already do badly and expensively, verify what the model tells you, and measure one real number for thirty days.