This page exists to answer once and for all. Not to justify — to clarify.
Dense AI content published regularly triggers a reflex in some: "slop." Without even reading. Without checking if there's sourcing, thinking, a real position behind it.
Here's what happens behind the scenes, unfiltered.
The "slop" reflex
Steven Bartlett (Diary of a CEO) calls it Law 3: when an idea challenges an existing belief, the brain goes defensive. It stops reading. It just looks for reasons to dismiss.
Result: no matter the nuance, the person responds to the caricature in their head. Those who drop "slop" in 2 seconds under a 2000-word text — that's exactly what's happening.
The man behind the username
Energy engineer. 29. Not an influencer, not a marketer. A technician who learned in the field before learning from books.
Non-linear path: vocational diploma → technical degree → prep school → engineering school → business school → field. Cleantech, international business development, industrial projects across 3 continents. The intersection of technical and commercial.
And that's where AI comes in. When you're managing multi-country projects, analyzing regulations, comparing technologies, sourcing, writing bids — you have two choices: hire a team, or give yourself the right tools.
AI is what lets me do the work of 10 without hiring a single one. A tool, not a crutch. And like any good tool — it's the person holding it that makes the difference.
Why "Supersocks"
It's not a random username. The sock is a daily reminder: time is finite.
I publish to structure my thinking, document what I build, and share what might be useful to others. My expertise is energy — not AI. I'm a tinkerer who might annoy the purists, and I'm perfectly fine with that.
Why now
You can run 30 billion parameter models on a $1,400 PC now. Two years ago that took a team, an insane cloud budget, and a PhD.
Open-source is catching up to closed APIs. Agents are doing real work, not demos. The barrier to entry has never been this low.
Those waiting for things to "stabilize" will wait a long time. I jumped on the train. If I crash, at least I'll have learned something.
The method
99% of the content comes from my own thinking. AI helps polish, not think. Typical workflow: reading → thinking → sourcing → writing → AI to challenge → rewriting → publishing.
My research started manually. Today it's automated — 105 sources, Grok analysis, zero noise. But the filter stays human.
Different format. Long reads, short takes, sharp positions. I reply to almost everything. It's a choice.
AI = tool, period
Using AI today is like blaming someone for using Google 15 years ago. What matters: the thinking, the critical eye, respect for authors.
Grok for analysis. Claude for code and writing. Gemini for research. Perplexity for sourcing. Each one has its job.
What I build
Wattselse — Vector mapping of energy companies, regulations, subsidies. Working MVP on Supabase.
AI Signal — Structured AI intelligence feed. 105 sources, automatic analysis, zero noise. You're reading it right now.
Cogefox — Sports betting algorithm. In development since 2016, fully automated.
The 5 buckets
Architectures, hardware, what's happening under the hood of models.
How it actually runs on your PC. Agents, configs, struggles, solutions.
Unfiltered critiques. Hypes that don't hold up. What's under the rug.
Fresh papers, Chinese trends, what's actually going to move the needle.
My builds, my energy expertise, what I test for real. Raw behind the scenes.
Final word
Skepticism is legitimate. Keep challenging — that's how content gets better.
I'm my own first reader. The day it sounds fake, I stop.
This is not personal branding. This is not filler. It's a guy who builds and shares what he learns.
The rest is noise. 🧦
