I've launched my first info product, and the launch changed how I look at something that's been bothering me for months.
There's a huge influx of info products about artificial intelligence right now. Courses, cohorts, books, paid communities, podcast tours. I've always been a bit perplexed by the info product world, and over the past months I've been properly disturbed by it, because most of the AI ones are made by people who have no clue about artificial intelligence. Some of what they teach is a lot of BS.
I've been working with technology for 35 years (I ran out of games on my Commodore 64 at six and started writing my own, and I never really stopped :) and with artificial intelligence for the last five to seven. So I get annoyed easily on this topic. Launching my own product made me stop being annoyed long enough to see the mechanism behind it. Once I saw it, I had to admit it works.
The procrastination guy with the Mac Mini
One example.
There's a guy who wrote a book about procrastination. He's now running a campaign promoting AI. He writes a lot about AI and agents, and he's making the rounds as a podcast guest.
His pitch goes roughly like this: he runs 420 agents, or a million of them, the number moves around, on a Mac Mini he bought. One agent runs his copywriting. Another writes code. A third does research. He tells it as a story, with a setup and a cast and a happy ending, and it sounds exciting.
I know it's bullshit. Anyone who has run more than a couple of agents on real work with real stakes knows it too.
What he's selling is super superficial, and he packages it in a super exciting way. Those two things go together. The less detail there is, the more room there is for excitement.
What the story leaves out
Three things are missing from the Mac Mini story, and each one matters more than the agent count.
The first is security. Agents that write your copy, write your code and do your research need access: to your accounts, your files, your inbox, sometimes your money. Every one of those is a door. The story never mentions who holds the keys, what each agent is allowed to touch, or what happens when one of them reads something it shouldn't and acts on it.
The second is brain atrophy. The way he teaches AI is "build a million agents and they'll do all the work for you while you chill out." If you hand off the copywriting, the code and the research, you stop practising copywriting, code and research. The part of you that knew how to judge the output gets weaker every month you don't use it. Then an agent gets something wrong and you can no longer tell.
The third is that it doesn't work. We all know the "agents do everything while you chill" setup doesn't hold up. Agents drift, they loop, they finish the wrong task with total confidence. Somebody has to check. Every AI setup worth paying for has a person making the calls and the tools removing the friction around those calls.
Then there's the comment section. Whoever replies with a counterargument, even in a trivial, polite way, gets deleted and banned immediately. That told me more than any of the podcast episodes. If your system holds up, a question about it costs you nothing. If it doesn't, every question is a threat to the launch.
Why he wins anyway
That guy can be successful with AI, and with virtually anything that comes next. The regular AI guy, the one who has been doing the actual work for years, mostly can't. I count myself in the second group.
The answer is distribution.
These people have huge mailing lists built on their previous topic. Say you have 100k subscribers from your procrastination or productivity work. AI becomes the hype, which it is right now. What do you need to do? Play around with it a little. Stay superficial, because people like easy ways and easy things. Then blast your list that you're pivoting to something new.
You already have a name. People already associate that name with "this person helped me once." So you sell your AI info product at a nice price, and a slice of those 100k people buys it. A small slice of 100k is still a lot of people.
The buyers don't care much that it's superficial. It's early, most of them know less than he does, and superficial reads as accessible when you have nothing to compare it with yet.
Depth wasn't what sold his first topic either. Pit even his productivity material against Huberman's research and it comes nowhere close. The list did the work then, and the list is doing the work now.
Playing the odds
What he's doing is playing the odds. He doesn't need everyone to be happy. He needs enough of them.
Some people buy it. Some of those people like it. A few are glad they opened themselves up to new horizons, which is great.
Then those happy people become testimonials. The testimonials get him into companies. Companies want training, so he teaches B2B. B2B work gets him onto conference stages. Conference stages make the name bigger, which makes the list bigger, which makes the next launch easier.
Damn, this is a beautiful mechanism.
I say that as someone who has spent a career building systems. Every stage feeds the next, and none of them depend on the product having any depth. The input is attention, the output is more attention, and the AI course is whatever happens to be passing through this year.
When the hype moves on
The problem shows up later.
The moment AI hype goes away, he moves on to the next shiny thing, now with a mailing list of 200k people instead of 100k. The mechanism doesn't care what the topic is. It worked for productivity, it works for AI, and it'll work for whatever comes next.
The people he taught stay behind.
When the dust settles, a lot of them will be disappointed that it didn't really work for them. The agents didn't do the work while they chilled out, and the person who promised they would is busy with a new topic.
The worse outcome is the other group: the ones who think they're experts now. They took the course, they've got a story of their own, and some of them will make their own info product. That's the next wave, even more superficial, and sometimes literally shitty. It sells to people who know even less, and some of those people will make the wave after that.
That's the part that disturbs me. Each wave teaches the next one that superficial is how it's done, and each wave has less to lose from being wrong.
Where that leaves the regular AI guy
So where does that leave someone who has done this work for years and has no 100k list?
Knowing the material doesn't carry it to anyone. The procrastination guy understood that before I did. He built the audience first and picked the topic later. Most of us who do the technical work did it the other way around: years of topic, no audience.
Launching my first info product made that gap concrete for me. I can be as right as I like about agents, security and brain atrophy, and it only matters to the people who hear it.
I've stopped being surprised that he's winning. He's playing a distribution game, and most of us have been polishing what we know and wondering why nobody's watching.
Before you buy the next AI course
Look at the seller's comment section first. Post one polite, specific question: how does your setup handle security, or what do you do when an agent gets it wrong? If the comment is still there tomorrow with a real answer, you've found someone worth learning from. If it's gone, keep your money.