10 print "Hello World"

Wow.

This post is all about the future that starts with my past.

I was seven. Maybe eight.

My dad had a Commodore 64, where I occasionally played games on, and a book about QBasic, a computer language, on the shelf.
I don’t know why he had that book, and I’ll never get to ask him anymore, but I did pick it up one day.

And typed the first line in the book and hit enter:

PRINT “HELLO WORLD”

And the screen said HELLO WORLD back.

Wow.

Then I became a real hacker after reading just one page of that book, when I learned about line numbers and the GOTO command.

10 PRINT “HELLO WORLD”
20 GOTO 10

Hit enter.

And the screen filled with HELLO WORLD in an infinite loop.

I didn’t know at the time what a loop was.

I just knew that I had told this machine to do something, and it did it, forever.

I was seven, and I was a literal wizard.
A really powerful tech wizard.

With a QBasic wand.

Wow.

I remember very specific, weird things about that time. My dad taught me how to start a game when you inserted the 5-inch disks, and it was something like

“LOAD name ,8,1”

That number sequence is something I just cannot forget. The syntax might be wrong, but I’m 100% sure that the 8,1 is somewhere in there.

I can see it in front of me, on that typical blue C64 screen, today, over 35 years later.

Back when I loved a guy called Pascal

I was 17.

In Switzerland, a “6” is the highest grade you can get in school. I got exactly one “6” in my entire school career. Informatics. I don’t know how to translate this term into English. It just meant stuff with computers.

My final project for that class was a village simulation, written in Turbo Pascal (or was it Think Pascal?). Little pixel villagers are collecting wood, finding food, and walking back to their houses. You had to place the forest, the well, and the houses as efficiently as possible. More efficient placement, more points over time.

It sounds more impressive than it was. Everything was circles and squares. But I was incredibly proud. I would have felt like Neo from the Matrix, but the movie wasn’t out yet.

I was going to be a dank programmer.

Wow.

Memory pointers

The year 1999. 18 or 19 years old.

My first real job was in C++ and API development. A project called HIT Messaging System. I was building a message handler between Lotus Notes and Excel. If that sentence means nothing to you, trust me, it meant nothing to me either.

I mean wtf?

The “skills” I had from sending squares around to get wood didn’t translate that well, and my ADHD was confused that a single function could have more lines than my dumb village project combined. I hit a wall at memory pointers.

My boss tried to explain them. I went home that night so overwhelmed, so nervous, that I couldn’t sleep. I didn’t understand anything about them.
This was my first real job with real programmers. People who just understand this stuff. They built digital houses out of bits and pieces, and I was busy drowning.

I came back the next morning on zero sleep. Sat down in front of my boss. And fell asleep while he lit a cigarette in the office (yep, kids, that was a thing back then) while trying to explain it to me again.

He looked at me and said, “Maybe this job is not for you.”

That sentence stayed with me for 20 years. It was l

Wow :(

Things that disappear

They let me stay at the company. Not as a developer. I moved into UX, attended a course from Nielsen where I learned about why links are supposed to be blue and not pink, working with Flash, and learning the cursed skill of creating animated GIFs.

Google happened. Those heretics thought they would be better than Altavista with their “efficiency” filled to the brim category websites.

I learned ActionScript inside Flash and animated websites. Got decent at it (as much as you can get decent at a language where you just click around). Started to feel competent again.

I was finally a super hacker again by creating HTML websites with CSS and embedded Flash animations.

I was... a Web Designer. And an Information Architect.

Wow.

(Many years later flash was discontinued completely. The entire tech, gone. And with it, ActionScript and all these weird things I’ve learned about Flash, Dreamweaver, and whatever that other weird piece of software was that came with it.

That was a strange thing to experience. Your surface to talk to can just... disappear. The technology moves on, and you’re left with… memories?

Will I become obsolete?)

The 20-year drift

I got fascinated by UX research, interaction design, and the why behind how people use things. I was still building, just not in code. 10 years later, Growth systems. PLG frameworks. Product strategy. I wrote about it for years. Linkedin, Substack. The PLG Wizard was here, 31,000 of you subscribed, right here.

But if someone asked? “No, I’m not a programmer. I used to code, but that was a long time ago.”

But I never stopped being a builder. I just stopped getting credit for it, because I built weird frameworks around organizations and other great 1st data topics.

I tried JavaScript about ten years ago again seriously, because - god knows why. It didn’t stick. My work didn’t need it. Except for this one game called Screeps, where you write scripts to manage a colony. Little autonomous agents gathering resources, optimizing paths, building structures.

It was the village sim. Again. Circles and squares all over again, fifteen years later. Same brain, same pull. I just didn’t connect the dots.

The thing about complexity

There’s something I’ve learned about my brain. It’s not that I can’t handle complexity. It’s that I can’t handle boring complexity.

An API between Lotus Notes and Excel? Plumbing. My brain checks out.

But a DCS flight simulator from the 1980s with a Russian cockpit where every switch matters? I’ll spend weeks learning that. Dwarf Fortress, a game that looks like absolute garbage but hides an impossibly deep economic simulation? Give me all of it.

Yeah. That’s a game. And it’s glorious.

I know how to scale products and optimize growth flows for millions of users. But filing my own taxes on time? Too many email accounts, Swiss financial terms I can’t keep straight (BVG, AHV, 2. Säule, 3. Säule), and the anxiety of even opening a message from my accountant. Administrative stuff that seems to come so naturally to everyone else.

I can simplify complex systems for others. I just couldn’t do it for myself.

And then AI finally clicked for me.

10 Print “Hello World”

A couple of months ago, something pulled me back.

I started working with Claude Code and building what I now call my vault.

A structured, text-markdown context layer for everything: my thinking, my work, my messy life. Nobody told me how to do it and I avoided on purpose any tutorial about it, I wanted to understand it 100%, not 99%.

From my experience leading the product at Jua.AI of how large language models work and my tech wizard background in QBasic I knew, that context is the real value in this world. Not the model. The context you feed it.

(No, context is not the same as “prompt engineering with 50 lines”. That should never have been a thing, and it will go away. You’ll see)

A persistent, hyper-efficient human memory context that exists before the model even reads the first word of what you say.

A memory context that is structured in a way where an AI only extracts from it what it needs without searching in plain text to deliver an accurate answer, not some answer.

Good context makes a simple “Help me with my taxes” build something amazing without having to go back and forth a 100 times before you give up. It knows who you are, what you want to know without asking for it, and how you want it.

Something that not only looks good but is, above all, correct. Something that can be trusted from the get-go. Something you understand.

And I built it. And it works. Right now. I use it with my clients, I use it with my own business.

I didn’t follow a tutorial. I didn’t read a best-practices guide. I just started building like I used to. Tinkering. And the old instincts woke up, and this time I didn’t get stopped by my attention span not being able to fix syntax.

This wasn’t unfamiliar to me. I’m not someone who follows manuals in “what” you should build. I just always learned enough to build something that naturally evolves from what makes sense. It has to be logical, above all.

And the thing that makes simple solutions for complex problems work is the context. And context is curated by hand, over time. AI will get better at automating it, but it will never be better than a manually curated one that is AI-Assisted. (AI suggest, you approve or reject)

The value of you as a knowledge worker or someone who wants to be a better person is the delta between what an AI can infer and what you carefully curate to be the memory above the AI.

You have to roll your sleeves up and do the work.

Generic context is generated. Correct context is earned.

The more correct context you have, the more likely you will improve the system you use it in. The more likely you are to spot bad data and decisions before anyone else.

You could think of your customers as “Average 35-year-old white male that likes music and wants to buy a CD-Player” or you could understand their context through a Job lens, “I want to listen to music on the go.”

That context allows you to explore different solutions, not just CD-Players. Job theory is nothing new in that regard, but building a memory around a challenge instead of a single job statement is.

I knew this years ago. I just didn’t realize I’d end up proving it with my own life by building a system around it.

And I will tell you all about it soon enough. In a way that I wish other people would have explained it to me.

If you know, you know. Yes, I’m one of those hipsters that uses Obsidian.

Why I’ve been quiet

Some of you noticed I haven’t been writing much. Here’s the honest reason: I didn’t know what version of me was supposed to show up. I was tired. And I didn’t really understand where all of this was going. This entire... AI thing, the industry.

PLG Leah felt generic. I had the frameworks, the experience, the data. But the world was shifting, and writing about growth mechanics without acknowledging what AI changes felt incomplete. I even started a separate AI newsletter, which didn’t really work either, as if these were two different people.

They’re not.

PLG, trust, and accuracy of information in the context of AI is really fucking wild. And I finally understand where my voice (and I think the industry) is going. The builder in me is back. Not as a developer who craps out the next 20 variations of a fitness app.

As someone who spent her life trying to understand how complex systems work, and thinks we’re all gonna be busy answering the same question in one form or another in our personal and private lives:

“Is what I’m looking at right now real or fake?”

We don’t need data analysts anymore to calculate a regression curve. AI can do that. We need analysts to know which context is the correct one before asking Claude to do it.

Helen from Marketing can now do a regression analysis with Claude. You bet she can. It’s just probably going to be about the wrong thing.

But hey, it looks dank!

Well…

The big misunderstanding that plagues this industry is coming from those who think that AI is for writing more efficiently. For mimicking voices and styles. Pretending to be human when you’re not. (Looking at you, synthetic Garys from sales)

It really should be about getting to the truth faster - the correct destination with fewer distractions. Whatever that means for you.

Anyways...

I’m not going back to C++. I’m not going to pretend memory pointers suddenly make sense.

But that seven-year-old who filled a screen with HELLO WORLD?

She’s still here.

And she’s having so much fun again. <3

20 GOTO 10

Next up

• “Why the bottom of SaaS is completely falling out.”