Smart In,
Smart Out

Over about two weeks this June, I had more than a dozen real conversations with Idaho business leaders. Not surveys, actual sit-downs about how they use AI, where it helps, and where it falls on its face. I'd just graduated from Boise State and taken a job at an AI company, and I wanted to hear how people who've run businesses through decades of change actually think about AI. Solo founders, regulated firms, a billion-dollar brokerage. Same question for all of them: how do you really use this stuff?

I expected the answers to split along the obvious line: big companies racing ahead, small ones falling behind. That's not what I found. The tools are free, or close to it. Everyone I talked to has the same access to the models I do. And yet the distance between them was enormous. One person is doing the work of a ten-person agency. Another is quietly drowning in confident, polished, wrong output and doesn't know it. Same town, same technology, completely different trajectories.

We were told this technology would flatten things out, that the little guy could finally compete with the big one. The great equalizer. But equalizers close gaps, and what I heard over and over was the sound of a gap getting wider. Because AI isn't an equalizer. It's a multiplier. Put a great marketer on it and they do the work of ten. Put a sloppy one on it and they produce ten times the garbage, just as fast. It doesn't care which direction it points, it just turns up the volume on whatever you already bring. Hand two people the same multiplier and they don't meet in the middle. They fly apart.

Dumb things, faster

The clearest description of AI I heard all month didn't come from a technologist. It came from the founder of a 160-agent brokerage that sells over a billion dollars of real estate a year. He started as a firefighter. He's not impressed by software. But he nailed it in one line:

He'd put real money behind that line. He swapped a $500K marketing agency for two sharp contractors and AI, at a fifth of the cost. The contractors didn't win because they had a better tool. They won because they knew how to aim it.

An executive coach I talked to later in the month had watched the same thing from the other side. One of her clients had to let someone go whose AI-assisted work kept coming back fast, polished, and completely useless. "Hot garbage," she called it. When I repeated the brokerage founder's line to her, she didn't hesitate: yeah, that's exactly it. AI "helps dumb people do dumb things faster."

Two people, two industries, the same verdict: the tool adds the speed, but nothing that tells it whether the work is any good. It multiplies what you bring. And most of the AI conversation, all the noise about models and features and prompts, misses the point. The tool isn't what makes the difference. You are.

It multiplies your judgment

Here's where I expected the answer to be "prompting." Learn to write better prompts, get better results, everybody's heard it. But that's not what the people pulling ahead were actually doing. The thing separating them wasn't a clever way of talking to the tool. It was something they brought to the table before they ever opened it.

A mining engineer I spoke with, now moving into commercial real estate, uses AI to do work he'd otherwise pay a lawyer for. Lease addendums, contract review, comparing insurance quotes, you name it. He does it safely for exactly one reason: he has an MBA with a business-law background, so he can catch the mistakes. "I just love the tool to save me thousands of dollars on legal fees," he said. He trusts AI precisely as far as his own expertise lets him verify it, and not one inch further.

If AI multiplies what good people can do, the obvious question is what happens to everyone else. At a lot of companies, the answer has been to cut them. A product marketer I talked to, someone who'd co-founded a software company himself, watched his employer fire the entire QA team, convinced AI could cover it. "I don't think that's worked out very well," he told me. They let the design team go too, figuring AI could turn their project managers into designers. Same logic, same result. The mistake wasn't using AI. It was believing the tool could replace the judgment, when the tool only works if someone with judgment is aiming it. He had a name for it: "an overeager intern that just makes stuff up." Useful, fast, and only safe in the hands of someone senior enough to catch it when it's confidently wrong.

It was never the prompting. What AI multiplies is judgment, your taste, your standards, your ability to look at a polished draft and know whether it's actually any good. Prompting is a skill you can pick up in an afternoon. Judgment is a skill that takes years to build, and it turns out to be the only input that matters.

Which leads somewhere genuinely uncomfortable. We were told AI would make expertise less necessary, that anyone could now do anything, that the playing field would flatten. The people I interviewed point to the opposite. AI made expertise more valuable, not less, because expertise is now the only thing standing between good work and confident garbage. And in a world where everyone can produce, the entire value shifts to whoever can judge.

And here's the catch: you can't buy that. You can buy every employee a subscription by Friday, but you can't buy them taste. It lives in the heads of your best people, and the multiplier compounds. Every day, the ones who can aim it pull a little further ahead, while everyone else produces polished nonsense faster and mistakes the speed for progress. The gap widens on its own.

A talent decision in a technology costume

Here's the part that took me a dozen conversations to see, and I keep coming back to it. When AI hands you back garbage, the problem was never the AI. It multiplied exactly what you gave it: your standards, your taste, your judgment. Every confident, polished, useless thing it produces is a reflection of a gap that was already there, in the person, the team, the company. AI didn't create that gap. It just made it visible, and fast. The useless output isn't the AI failing. It's the AI showing you, fast and without flattery, exactly where your standards actually are. Mine included. Which means your AI strategy was never really a tools decision, which platform, which license. It's a talent decision wearing a technology costume.

I'll be honest, that could read as bleak, but I don't think it is. Because the people I watched pull ahead weren't geniuses. They weren't fearless and they didn't have secret tools. They just knew their craft well enough to catch the mistakes, and they put the work in to point a fast tool at the right problems. That points to the actual work, and it isn't buying better tools. It's getting AI into the hands of the people who can aim it, and building that judgment into everyone else. That's the part that gives me hope, especially as someone just starting out: judgment can be learned, standards can be built, and the mirror that shows you the gap also shows you exactly what to work on. Smart in, smart out. The only question left is what you put in.

— W.S., Boise, June 2026