As you know (do you?), I run Claude Code across every one of my side projects.
So when Andrej Karpathy’s four-rule CLAUDE.md file started showing up everywhere in May, I didn’t just read about it, I installed it and ran a week of real work through it.
And, then I added my own version to it, pulled four more rules and kept updating them accordingly.
And this has been my pattern!
Every single article of mine, that has made me good money is basically a product of me sitting on my laptop, testing out things.
Writing it barely takes effort.
My usual idea is to:
Pick a topic you’re already using, not one you’re researching to write about.
The gap between “I read about this” and “I tested this on my own five projects” is the entire difference in how an article performs, both in read-through and in whether people trust the numbers you’re citing.
Anyway, this one crossed 250 dollars.

techaiguild.aibucket.org
It picked up all of a sudden, so late. I wonder why! So, I started including it in all my articles footer!
There’s nothing super special about my writing, again.
The tech-part is useful though.
In the words of my imaginary girlfriend, “Your English, your grammar and your vocabulary is ordinary. Your money making skills aren’t.”
Things I’ve learned along the way, or what I can still give away for free xD.
Claps don’t matter, reads do.
If your write-up is getting views but the read time is low, the fix usually isn’t the topic, it’s the opening two paragraphs asking too much patience before the payoff.
I don’t know about SEO titles, but write good SEO descriptions. Long enough to include the keyword someone’s actually searching, short enough that Google doesn’t truncate it in results. Works for Fiction too, btw.
This is me going out of my niche, btw.
So, I’ll keep this article free.
Meow meow, Medium :)
In case we are meeting for the first time, come over here, it’ll be worth the roller coaster of articles that are gonna come up in the next few weeks.
And if you’re hunting for your next startup idea, GritGlean aggregates real demand signals and pain points from Reddit, X, HN, Stack Exchange, Quora and more, and it even finds existing apps for sale if you’d rather skip the build :)