Real numbers first, because this niche is drowning in screenshots nobody can verify. A business called Dentology, built to sell AI services into dental practices (going by the name, anyway, since that part was never spelled out), was doing about $7,500 a month roughly a year ago. Four months later, $50,000 a month. Well under a year after that, $150,000 a month. Those numbers came from the guy who built it, on his own channel, and we're not rounding any of them up to sound better. What actually got him there is mostly a lesson in how to get clients for AI services, and almost none of it was about the AI itself.
The honest math here is thinner than I would like, and I will say that plainly instead of padding it out. He didn't break down which client paid what, or how many clients it took to hit each milestone. What he did share, in enough detail to actually use, was what changed about who he pitched and what he promised. That's the part worth stealing, so that's what this post is about, not a fake per-client spreadsheet I'd have to invent to fill space.
Here's the part that actually stopped me for a minute. A Stanford psychologist named Albert Bandura spent decades studying why people who have every resource they need still don't act, and his answer wasn't motivation or laziness. It was self-efficacy: whether you believe you personally can produce the outcome, not whether you know the steps. Bandura's original research found that people talk themselves out of trying long before any client ever actually says no to them. My honest read is that's the real reason most posts about getting clients for AI services get bookmarked and never acted on. Not a skills gap. A belief gap.
The same guy mentioned, almost as an aside, that he ran his own small agency back in 2017 and hit a wall that had nothing to do with talent. He didn't dwell on it, and I won't either, since I only have the outline he gave and not the whole story. But it tracks with the self-efficacy point: the wall wasn't a missing tool. It was the part where you have to believe a stranger will pay you before you've proven it to yourself.
One thing before the two mistakes. I lost close to a month once trying to piece together an AI-services pitch from scattered YouTube videos, half of which contradicted the other half on pricing alone. A structured course cut through that so cleanly it was almost embarrassing how much faster the second attempt went. YouTube is fine once you already know what you're looking for. It is a maze when you do not.
One more thing first. The second mistake he named is the one that actually moved the revenue number, not the first one. I'm saving it for the end of that pair on purpose. Read both, but pay attention when you get to the second.
Mistake one: selling the AI instead of the outcome
Nearly everyone trying to land AI clients pitches the technology. "I can build you an AI agent." "I can automate your intake with AI." A dental practice owner doesn't want an AI agent. He wants the phone answered and the schedule full, and he doesn't care what's doing the answering. Once Dentology's pitch stopped opening with the word AI and started naming the outcome instead (more booked appointments, fewer no-shows, whatever the specific promise was), the replies he got started matching what he could actually deliver. Selling the outcome isn't a copywriting trick. It changes who replies, because the person who wants "more booked appointments" and the person who wants "an AI agent" are almost never the same buyer.
Mistake two: picking problems too far from revenue
This is the one saved for last, and here's why. Two AI fixes can take the exact same afternoon to build and pay completely different amounts, and the gap isn't difficulty. It's how close the fix sits to money actually changing hands. An AI tool that cleans up internal notes is nice to have. An AI system tied directly to booked appointments, inside a business that lives or dies on booked appointments, is something the owner will pay real money to protect. The closer the problem sits to revenue, the more it's worth, and the easier it becomes to charge for. That's most of the actual jump from $7,500 to $150,000: not more clients doing the same small fix, but bigger fixes on problems that were already costing real money.
If any of this depends on skills you don't have yet (writing the actual automation, or just getting comfortable enough with Claude to build a working demo before you pitch anyone), that gap is worth closing before the first cold message, not after. I picked a short daily-lesson course on Mindwand because the format matched an actual work schedule better than a 40-hour library I would have abandoned by week two. Skool communities and Coursera cover a lot of the same ground if that format fits you better. Pick one. Finish it.
Back to Bandura for a second, because it explains something that bugged me. If selling outcomes instead of AI, and picking revenue-adjacent problems, is this straightforward, why doesn't everyone already do it? My honest answer is I don't think it's information most people are missing. It's belief. Sending the first pitch means accepting, out loud, to a stranger who can just say no, that you might be wrong about your own value. Bandura's research says people avoid that exposure long before they run out of ideas. The fix he pointed to, and this is the one part of his work that felt immediately useful, is small completed tasks first, in order, each one slightly harder than the last. Not confidence pep talks.
That's also, not coincidentally, the whole argument for a structured course over a stack of bookmarked videos: you finish something small on day one instead of collecting seventeen open tabs. The lesson-a-day format on Mindwand is built around exactly that, and again, Coursera or Skool will get you a version of the same structure if that's what you already use. Format matters more than brand.
None of this is a guarantee, and I'd rather be honest about that than close on something inspirational. Dentology is one business, with one owner who apparently already knew how to build and sell, and his run proves the system works at exactly one data point. Most people who try this will spend weeks not landing anything, same as every real income story on this site. What the two mistakes above buy you is a pitch that at least reaches the right person, and a problem worth what you're charging for it. The hard part isn't finding the two mistakes. The hard part is sending the pitch while some part of you still thinks you might be wrong.
If AI-services work for local businesses isn't the specific lane that fits you, our broader AI side hustle breakdown covers a few other starting points worth comparing before you pick one.