The businesses fAIr Advantage Ohio exists to help are the same businesses everybody else is targeting right now. Some of those people mean you well. Some of them don't. Most of them can't tell the difference themselves.
I've been writing software for about forty years across seven industries. I've watched a lot of technology waves come through. Every one of them brought a crowd of people who showed up the week after the wave hit, bought a domain name, and started selling. AI is no different in that respect. It's different in one important way, which I'll get to at the end.
One thing before I start. I'm going to use myself as the example a lot in this post. That's because I'm the example I know best, not because I'm trying to sell you something. I'm not. You won't find a service or a product of mine named anywhere in here, and that's on purpose. The point isn't that you should do business with me. The point is that people like me exist, and you should be doing business with somebody like that. Take everything that follows as a peek behind the curtain, not a pitch.
Here's how I sort them.
1. Are they asking you for money?
Be suspicious. Not disqualified. Suspicious.
People have to get paid. I have to get paid. You have to get paid. Nobody should be embarrassed about that. The question isn't whether money changes hands. It's when, and for what.
If someone wants a meaningful chunk of money up front, before you've seen anything work, that's a yellow flag. You're putting cash in at the point of maximum uncertainty. The right time to pay is after value has been demonstrated, or at minimum after a small, well-defined piece of it has been demonstrated.
There are plenty of ways to structure a first step so that nobody's taking a big risk. A short paid pilot with a clear outcome. A demo built on your actual data instead of theirs. A first month you can walk away from. If the only path on offer is a big check and then we'll see, keep your hand on your wallet.
2. What does this do that I can't do myself?
This is the question. Ask it exactly like this:
"What does this do that I can't do myself with ChatGPT or Claude?"
Then stop talking and watch.
Anyone who has actually built something useful can answer this in a couple of sentences, with no tap dancing. To hear the answer clearly, you need to know what the twenty dollar a month version already does. It does more than most people selling to you want you to know.
"It connects to your email, your calendar, your other systems." Big whoop. ChatGPT and Claude both do that today, out of the box. They can read your inbox. They can see your calendar. They can plug into your files, your CRM, your project tracker, and a long list of other tools. The connector ecosystem for these products is massive and growing every month. Connecting things together is not a product anymore. It's a checkbox.
"It has my coaching built into it." They're overvaluing their coaching. That model has read their book. It has also read every other coaching book, every framework, every methodology, and every blog post those people ever wrote. You can type "coach me in the style of X" and get a remarkably good result. If the whole pitch is that their expertise lives inside the tool, ask yourself how much of that expertise was already in there before they showed up.
So what's left? What's actually worth paying for?
Look for the ways they raise the quality floor. Anyone can get one good result out of these tools. The hard part is getting a consistently good result, every time, from anyone on your team, without you standing over it. That's the difference between a trick and a process. Ask them:
- How do I get high quality results out of this every single time, not just when I'm paying close attention?
- How does this scale past me? What happens when my two employees use it, or my ten?
- How much of what you're selling depends on my instructions? If the answer is "most of it," then I'm the product and you're the wrapper.
- What happens when the model underneath changes next month? Because it will.
That last one deserves a little more room. Here's the peek behind the curtain. When I'm designing something, one of the first questions I ask myself is whether it gets better as the models improve, or whether the heart of it is a workaround for something the current model can't do yet. If it's the first kind, every new release makes it stronger without anyone lifting a finger. If it's the second kind, it has a shelf life, and the next release can end it. New models are landing about every six weeks right now. That's a very short shelf life. Ask the person selling to you which kind they've built. If they've never thought about it, they've built the second kind.
Real value shows up as guardrails, checks, structure, and things that run correctly when nobody's watching. Someone who has done this will talk about failure modes and how they catch them. Someone who hasn't will talk about features.
If the answer is a wall of words about transformation and unlocking potential, what you're buying is a subscription with a markup and a logo on it. A lot of what's being sold right now is exactly that. Ask the question and you'll hear it.
3. What have you built that survived contact with the real world?
Demos are easy. I've built a lot of demos. Anyone can make a thing look good for eleven minutes in a conference room. Ask anybody who's ever been on a first date. The first date is the demo for the relationship. The first date and year three don't generally look the same.
The real world is where it gets interesting. A customer types something nobody expected. The data is messier than the sample. Somebody in the back office has been doing the job a slightly different way for nine years and never told anybody. This is where most of what's being sold falls over.
So ask: what have you built, for whom, that's still running? How long has it been running? What broke, and what did you do about it?
If someone can't name a single thing that's been through that, they're selling you a demo.
4. Ask for references
Boring. Old. Still works.
Ask for names of people they've done this for. Then actually call them. Not the testimonial on the website. A human being on the phone. Ask that person if they feel like they got value, what it does today, and what they wish they knew going in.
If the references are all "we can't share client names," that's an answer.
5. A few more that don't fit on the poster
Can they explain it in plain English? If you leave the meeting less clear than when you walked in, that's on them, not you. Jargon is where inexperience hides.
Do they ask about your business first, or pitch first? Anyone who starts with the product before they understand what you do is selling a product, not solving your problem.
Do they ever say no? Anyone who says AI can do everything you ask has never watched it fail. It fails. I use it all day and it fails all day. Someone who tells you where it breaks is someone who has used it.
Who owns your data when this is over? Where does it go, who can see it, and can you take it with you if you leave? Get that in writing. If they can't answer it, they haven't thought about it, and that's its own problem.
Are they teaching you or trapping you? The good ones leave you more capable than they found you. The other kind builds something only they can touch and calls it a partnership.
Are they using the word "unprecedented" to rush you? It's true that this is moving fast. I say it myself, probably too often. It is not true that you'll be left behind if you don't sign this week. Urgency is a sales tactic. Speed is a fact. Don't confuse them.
Not everyone who fails these tests is a con artist
I want to be clear about that. Some of them are just new. They read the same articles you did, got excited, and started offering before they'd finished learning. That's not a crime. Some of them will turn out to be very good.
If you can look at one of these people and decide their inexperience is worth taking a chance on, do it. That's your call and it's a legitimate one.
But be honest about your position. Most of the small business owners I've talked to are hoping somebody takes a chance on them. That usually means there isn't a pile of spare money sitting around to take a chance on someone else. Wanting to help someone get started is a good instinct. It's not a business plan. Better options are out there. You don't have to take the unnecessary risk.
Why this isn't the Wild West anymore
Everything above would have been harder to say two years ago, because two years ago almost nobody had a track record. Everyone was new. The only honest thing anyone could offer was a guess.
That's changed, barely. There's been just enough time for some people to build things, ship them, watch them break, fix them, and keep them running. Those people exist now. There aren't many of them yet, and they're outnumbered by the noise, but they're out there, and they can answer every question in this post without flinching. Your job is to find one.
So here's the short version. Ask what it does that you can't do yourself. Ask what they've built that's still standing. Ask who you can call. And don't hand over the money until you've seen something work.
Most of what you'll come across won't survive contact with the real world. Your job is to find the small part that will.