Notice what Michael didn’t do.
He didn’t try a demo. He didn’t paste in a sample someone handed him.
He took real conversations with real companies he was working with, and he put those in.
That’s the test that matters. Not whether a tool looks good on someone else’s example. Whether it holds up on your own messy, real material.
His verdict was three words long.
Like magic, man.
The man with three jobs
Michael wasn’t a tech guy by trade.
He’d worked at Oracle. He knew what that kind of career looked like from the inside.
"...could get a job where you’re making. $300,000, $400,000 a year. But then you’re a slave to them. And you’re always anxious about losing your job once you get it and proving value and worth."
So he made a call.
"Ultimately, I said, I want to get out of that. I want to do something for myself."
And he did. When people asked what he did, he had a three-part answer.
"I’m a loan officer, I do insurance, and I’m a social security analyst."
He’d seen what that last one did to people approaching retirement.
"You should have saw their eyes open like saucers when I mentioned that."
He had something people wanted. His problem was reaching enough of them.
"At the end of the day, it’s a numbers game."
And numbers take time. Time Michael didn’t want to spend on outreach by hand, because, as he put it, "I do so many other things."
The transcript that came out wrong
Not every transcript Michael fed into a machine came out like magic.
On one of Andrew’s group calls, he shared his screen to show an outreach campaign he’d been building for LinkedIn.
He’d done his homework. He’d taken transcripts from a couple of calls, given them to ChatGPT, and had it fill in all the advanced fields in his outreach tool.
The messages came back. He wasn’t thrilled.
"I do see that some of that sounds very AI."
He was right. Andrew could see it too.
"Everything that I had previously seen looked good coming out of there, and what you’ve got is very AI-generated."
Andrew’s read was that the trouble came from the extra instructions layered on top. Michael had asked ChatGPT whether to fill in the advanced prompt fields, and it had said yes.
Andrew suggested he strip them out and see what came back.
Michael didn’t argue.
"So I’ll take that out. That’s a good idea."
Maybe you’ve been there. You give the machine more and more instructions, hoping each one makes it better, and somehow the output keeps getting worse.
More isn’t always more. Sometimes the fix is taking something away.
The technical brains
Michael wasn’t shy about telling Andrew the truth, either.
At one point he put it to Andrew this way:
"You may lack in webinar proficiency, but you have the technical brains, so don’t worry about it."
It’s a funny line.
It’s also an honest one. Michael’s point was simple. The polish wasn’t what mattered to him. What mattered was the part that comes after the slides, when someone has to make the thing actually work.
The corn
Nearly a year after that outreach campaign, Michael showed up on one of Andrew’s calls after some time away.
He’d been heads down on work.
And he started with a story.
A good friend of his was in the seed business. Corn, specifically. With cuts at the USDA and other agencies, there was nobody left keeping track of a lot of the information about different hybrids. The data was all out there on websites. His friend had been trying to scrape it with ChatGPT and struggling.
So Michael stepped in.
"I just put co-work to work on it and just, I mean, it got all of the information that he needed. I mean, I probably spent with personal time about an hour on it. Claude spent about a day and a half on it. And I gave him something that he would have never been able to finish himself."
An hour of his time.
A day and a half of the machine’s.
And something his friend could never have finished alone.
Here’s the line that tells you how far he’d come, though. He said it before he even got to the story.
"Helping people with AI isn’t my thing."
And yet there he was, doing exactly that.
Something Andrew had said got him thinking about asking his friend for a testimonial. Something he could put on his website about how he could use technology alongside the rest of his business.
The other side of the table
Think about where this started.
Michael feeding transcripts into software and calling the result magic.
And where it ended up.
Michael being the one who made the magic happen for someone else.
He still said AI wasn’t his thing.
This time, he was the one making the magic.