In October, I’ve asked my AI agent a question:
Based on all you know, how likely it is that I’ll own $1M in investing assets in 5 years from now?
It went through my notes. What my job is. What’s my current assets collected. Then it gave me this one line:
This hits hard, because it’s what this whole brand is about - a public journey to $1M without sacrificing family nor job I like. I made a bold claim, and now I’m proving it being correct.
However, even though I love planning and aiming shiny goals, what I learned is at least as important, is doing the reviews.
I dislike reviews of any kind, and I procrastinate on them. Today I finally did what I always avoid. I wrote it down whether AI is right.
What I made
August: Posts 3; Notes and quotes 146; Replies 244; Commits on products 37+
September: Posts 3; Notes and quotes 175; Replies 374; Commits on products 102+
October: Posts 5, Replies: 136; Notes and quotes 182 Commits on products 405+
That’s 11 posts, 457 notes and 800 replies in three months, next to a full-time job and three kids in homeschooling.
And what are those commits?
Newsletter Triage. My 5,042 unread newsletters finally have a reader. It agreed with my own verdicts 13 times out of 14, saving me over a week of reading stuff I’d throw away anyway. That resulted in finding some real gems like this one!
A visual brand for Engineer of Wealth. Canonical characters and locations, a measured 3D room built from real sizes that ended up with interactive personal website that shows real progress in a comic-like way, a validator that catches image drifts. I wrote about it more on: I forced AI to generate precise images.
A comic. Issue 1, Issue 2, four more scheduled, and its own website releasing every Saturday at 07:00.
My agents and AI workflows, pulled out into their own repos. The ghostwriter, the executor, the visual critic. Those are the next products I’m going to add to my substack paid tier next!
Then in the first days of November I built a metrics collector, because every agent I have was planning from a September picture, and I was too busy building to update it.
My Claude Code usage tells the same story. 48M tokens in the week. 336M in the next one. 687M in four weeks.
7x. In one month.
I’m an engineer. Give me 90 minutes a night and I will fill it with code.
What I sold
Ship & Sell Kit sales: 1
Paid subscribers: 1
And about subscribers? 237 on September 19. 260 yesterday.
My activity on Substack dropped in October, and that one was on purpose. On September 19 I measured that 77.4% of my notes’ reach goes to people who already follow me, so I stopped treating notes as the way to get new readers.
I am going more collaborations and lives instead.
It was a good decision, but look at the table again.
Notes went down by 39.
Replies by 192.
Commits went up by 303.
I didn’t move that time to selling. I moved it to building.
The test I keep failing
I have a test for this in my own notes already. One question, every single month: had anything become banger? It could mean, someone made a published case study. Or I got a testimonial from someone real.
If not, in my own words: “you ran the treadmill again, regardless of how much got shipped.”
So what’s the October? Mostly treadmill, yes.
But ONE piece did fire, to be fair. My delivery got faster because of that. A single comic page ate 6M tokens on its first draft, and the three redraws after I fixed the setup cost about 750k. Each new issue eats less as the system get optimized.
This is of course the arc that feels great and pays nothing!
Now, a bit of surprise.
The comic started patly because my daughter asked: “Daddy, can we make a comics?”
There was no funnel, not even a plan. I had no time to create comics that has nothing to do with my online branding, so I merged both together… Then the article about building it became my best post of the last three months: 17 reactions, 15 comments.
Post reactions went from 16 in September to 44 in October. 36 of those 44 came from the four comic posts.
Yes, I know. Small numbers, one month, no conclusions - but I’ll take it regardless.
You can’t compete with someone having fun. On my website you can feel it that I do have the fun. By checking Quackmera on the shelf. Or by realizing, that the Comic rack is real.
It also explains the problem. I have too much fun building. Selling is the one thing I do only when a deadline forces me.
So is 7% wrong?
I wanted to find out that my AI is too pessimistic. But it’s not. It just concluded, that if I will continue the same habits, I’ll be the best AI builder in the world, and as far as possible from the goal.
The honest reading of the last three months is: building is not my constraint. I know, because I told you so. On July 30 I published Building is solved, selling is not.
Then I went and built more.
So, for November & December:
I pack what I built, into perks for my subscribers and sellable standalone products.
A new row in my Sunday recap: people who bought something and weren’t already paying me. It will probably say 0 for a while. Good. Let it be embarrassing!
The comic stays. It’s scheduled, it’s cheap now, and it’s the fun that keeps me here. Also, the BulpenStudio, the factory I use to make it, will be released too. And yes, my subscribers will get there in their standard premium subscription too.
The comic stays, because it already gave value to people.
On January 8 I’ll run the same prompt again and publish whatever it says - because I disagree. The data is a feedback I use to make the claim happen!
And because a whole article about not selling without an ask would be a bit too on the nose: Newsletter Triage is $49, or free for paid subscribers. If you have thousands of unread newsletters, it reads them for you.
Now, what are you building right now so that you don’t have to sell?
By the way, really good hook. It grabbed me right away.
We're pretty similar, man. I just want my own business and to be able to travel more with my family around Europe. We're both dreaming, working our asses off, and just starting out on this journey. Let's keep turning dreams into reality.