You're Using AI Wrong. Here's What Actually Makes Money.
+ Use my free "Promotion Brain" prompt doc
This is the written version of this week’s “sidequest” video. The video is the full, unedited, occasionally unhinged version (with a new camera setup!). This article is the good bits, organised. [Watch on YouTube] and copy/paste my Promotion Brain prompt!
Hey you,
Most of the CEOs I know who are using AI tools are not making more money with them. They’re making less. Which is actually kind of funny tbh.
Not because the tools are bad. The tools are insanely good, like anyone using Fable the last few days would be crazy to say they suck.
BUT… whats bad is
The tools keep pulling them into €50 an hour work when the job they’re supposed to be doing is worth €5,000 an hour.
I watched it happen at my own company. Early on, one of the people here who is responsible for making money spent ages building a vibe coded proposal machine. Client call goes in, transcript comes out, a standardised proposal comes out the other end. Really cool.
Except by the time a client asks us for a proposal, we’ve already closed the deal. The hard part, the part worth real money, is getting more of those calls in the first place. And the person working on that is now locked into an AI fucking dopamine fest. Great.
If you’ve spent an evening recently building some clever little internal tool for your business, you might already know exactly where this is going.
Before anything else: the thing I made for you. It’s called the “Promotion Brain” and it’s a free Google Doc, right here:
→ The Promotion Brain (free Google Doc)
Paste it into a Claude or ChatGPT project. It interviews you about your business, then makes every future AI session work on the €5,000 an hour job instead of the rabbit holes. You don’t need to give me your email address to get it. You can if you want to. If that’s all you came for, take it and go. If you want to know why it exists, keep reading.
The nuggets (if you only read one section, read this one)
Your job as a CEO is to be the promoter of your business. Most of your week should be spent getting more people to know your business exists. More awareness means more leads, more leads means demand beats supply, and demand beating supply means you can raise your prices. If you don’t like that job, you might need to hire a CEO.
AI tools are tricking you into €50 an hour work. Invoicing, scheduling, meeting summaries, internal systems, proposal generators. That work existed before AI, and you correctly ignored it or hired someone for it. Now the tools make it fun, so you’re doing it yourself.
Systems don’t correlate with revenue. You can leave your company an absolute chaotic mess and still make a lot of money. You can have the most structured, streamlined, calm company on the planet and be making nothing. I don’t see a correlation between those two things, at my company or any company I know.
Be suspicious of the perfect note-taking system. Including your own. When I meet a systems-obsessed founder, I assume one of three things: they’re not making much money, their company doesn’t need to make money, or they already made so much they’re just looking for stuff to do. If your meeting-notes setup is “so dialed,” the only question that matters is: how much money have you made?
AI-generated content is slop, and everyone can smell it. Everyone has access to the same tools, so everyone’s AI LinkedIn post looks like everyone else’s AI LinkedIn post. Use AI like a researcher on your team. Deliver the thing as a human. It’s the delivery that people connect with, and you cannot outsource that.
If you want to vibe code something, vibe code something wired to revenue. Not a meeting summariser. A quiz that collects emails from potential clients and ends, for the right people, in a call with your team. I built exactly that at Facilitator.com. It shows up in the P&L. The internal tool never does.
Steal the “Promotion Brain.” The free Google Doc linked above. It exists to catch you mid rabbit hole and point the tool back at the only work that pays €5,000 an hour: promotion. No email required.
Okay. The story.
€50 an hour work in a €5,000 an hour seat
Here’s how I keep it simple in my own head. There is €50 an hour work: invoicing, scheduling, summarising meetings, tidying documents, smoothing internal processes. Fine. Useful. Changes nothing.
And there is €5,000 an hour work: promotion. Getting people to know your business exists. Finding leads. Increasing the leverage of the whole company.
That proposal machine from the top of this article is the whole pattern in miniature. It was clever. It worked. And I had to say out loud: I don’t give a shit about the proposal process being crystal smooth. The deal is closed on the calls, so the clever person’s time should go to the actual hard part of the business: getting more of those calls to happen.
Here’s what changed with AI. The temptation to do low-value work always existed, but before these tools you had a kind of accidental focus: the boring stuff was boring, so you didn’t do it. Now people who have entire teams for admin are saying “no no, I’ll rebuild the proposal process myself.” And I know the counterargument, because I hear it every time: “but I build the system once, and then I never need someone to do it again.” Even that is wrong. The fact that you are the one building it at all is the problem. That was never the seat you’re paid to sit in.
A while ago we were running a revenue challenge at AJ&Smart, and I was talking it through with a few people. One of them suggested a well-known operating-system-for-your-business framework, which they were in the middle of installing at their own company, because their company was having challenges. And my thought in that conversation was: is this a systems problem? Or is revenue not coming in because there’s a lead generation problem?
Systems are easier to work on. That’s the trap. You can’t cleanly measure whether the system did anything, so you get to kick the results down the road forever. It happens inside my company too: I ask a team lead how a product is selling, and the answer is “this month we focused on improving the internal way we work so we can be more efficient.” I can promise you, that has never once shown up as money.
The tools are whispering that efficiency is how you make money. It isn’t. You make money by being better at the CEO job, and the CEO job is promotion.
The slippidy sloppidy problemidy
The second way I see CEOs using AI wrong: the generic content machine.
“I don’t like doing content anyway, so I’ll get the AI to do it.” Here’s the problem: It fucking sucks. It’s soulless. Everyone has access to the same thing, so your post reads like the same slop as everybody else’s. When I get an email from another founder that was clearly written by AI, I find it disgusting. I don’t even want to read it.
I’m not saying don’t use AI in the process. I used Claude to help me build the document I gave you above, and to get my head straight before recording. That’s the right use: AI as a researcher on your team. But then I sat down and recorded the video with no script and no teleprompter, and I write my LinkedIn posts directly into LinkedIn, spelling mistakes included. The human delivers.
The more people hand their voice to the tools, the more vanilla everything gets. Which quietly makes the human stuff more valuable every single month. That’s not a moral position. It’s an arbitrage.
The math on this exact video
Let me show you the €5,000 an hour thing with the video you’re reading the companion to.
Recording it took about an hour, because it’s unedited and goes straight from my mouth to YouTube. Add the research beforehand and the upload, call the whole thing three hours of my time.
Now the other side. From a video like this, a couple of people will join the newsletter. And when I eventually announce something, one or two of the people who joined will come along to one of my in-person events, and those are 8K per ticket. Run that math and be aggressive about rounding down: this video is worth somewhere around €4,000 an hour to my business.
The same three hours spent vibe coding an internal meeting summariser? Cool tool. Genuinely. And it generates €0 this quarter, and probably €0 next quarter too.
Yes, I know about the famous exception: Basecamp built an internal tool that turned into a multi-million dollar product. Sure. That is not what’s happening in most small businesses. Most of us are just making cool stuff that leads nowhere, with a permission slip borrowed from a company that got lucky once.
What I do differently now
One rule: the money work happens first, and the fun stuff happens after.
I still use these tools to make silly things I shouldn’t be making as a CEO. Right after recording this video I’m going on Greg Isenberg’s show, because that’s promotion, and that’s the job. And tomorrow I want to spend time building a little website I have no business building myself. I’m a designer, it’s fun, and someone else should be doing it. The difference is I know which hours are which. The €5,000 hours got spent first.
The “Promotion Brain” is how I make the tool itself enforce that rule. From the document:
The €50 per hour work is invoicing, scheduling, summarising meetings, tidying documents. Fine. Useful. Changes nothing.
The €5k per hour work is promotion.
Same free Google Doc linked at the top. Take it.
If you’re mid rabbit hole right now
I want to say this to the specific person who spent last night connecting three tools together for a workflow that saves nobody any real time.
I’m not judging you. I fell into the rabbit hole too, and I’ll fall in again, probably tomorrow. The tools are built to be delicious.
But your company can stay a chaotic mess and still make a lot of money. Nobody is coming to give you a prize for the streamlined internal process. So close the tool-builder tab, take the Promotion Brain, and spend the first hour of tomorrow on the only question that pays €5,000 an hour: how do more people find out that your business exists?
It is NOT hard to get started. That’s the whole point of the document.
Cheers, Jonathan
P.S. The full video is the complete, unedited, occasionally chaotic version of all of this, including me breaking my own document live on camera. If you know a business owner who’s currently three weeks deep into building an AI system for a problem they don’t have, send them this. That’s genuinely the best way to help this show grow.