How I Learn Without Turning My Life Into Homework

I have two YouTube accounts.

One is personal — football, music, random things I enjoy. The other is entirely AI, automation, technology, and business. No crossover. No exceptions.

I didn't set it up this way from the beginning. I did it after my single account became a useless mix of both — where searching for a tutorial meant landing in my personal feed and suddenly learning felt like homework. The recommendations got confused. More importantly, my brain got confused. I'd open YouTube to relax and feel the low-grade pressure of things I should be watching instead. I'd open it to learn and get pulled somewhere completely unrelated.

The separation wasn't a productivity decision. It was a protection decision. I didn't want learning to contaminate rest, or rest to contaminate learning. Both needed to exist cleanly — in their own space, on their own terms.

That turned out to be the first layer of a learning system I didn't know I was building.

Why I Needed a Better System

The system became useful when I started moving between domains faster than a normal course-based approach could handle. At different points, I had to understand automation, AI agents, contact center systems, payments, business operations, and technical implementation well enough to have real conversations — not just consume content. I did not need academic mastery every time. I needed fluency. I needed to build a map quickly, test it against real problems, and avoid turning every free hour into another obligation.

How the System Works

Most people learn a new topic the same way they were taught in school: find a course, watch it in order, hope it sticks. That method is slow, passive, and increasingly unnecessary. It also assumes the course was designed for where you are — which it almost never is.

What I use instead is a deliberate sequence. Different tools for different cognitive jobs. Here is how it actually works.

Layer 1 — AI Research First

Before I watch a single video, I use AI to build a mental map of the topic.

This is not casual questioning. I use AI with web access to build a first map from current sources — then I check the parts where accuracy matters. For topics where numbers matter, I use NotebookLM to gather and organize sources so I can reference them properly. And I always look for real user discussion — Reddit, forums, practitioner communities — and summarize it into a document. That layer tells me what people who actually work in the space think, argue about, and get wrong. It is usually the most useful thing I read.

The prompt I start with depends on how new the domain is. If I have no background at all:

I'm learning [topic] from scratch. I have no background in this field.

Give me:
1. A high-level overview — what this field is, why it matters, how it's structured
2. The key concepts I need to understand first before anything else makes sense
3. The most important numbers, metrics, or facts I should know (with sources)
4. What practitioners actually debate or disagree about
5. What a beginner typically gets wrong

Use plain, simple English throughout. Where technical terms are unavoidable,
collect them all in a glossary at the end with a one-line definition for each.
Keep it dense and factual. I'll ask follow-up questions after.

If I already have adjacent knowledge and just need to map a new domain:

Explain [topic] by connecting it to what I already understand about [adjacent field].
Focus on:
1. The core mental models
2. The key terminology
3. The most important flows, metrics, or assumptions
4. What practitioners debate
5. What someone with my background is likely to misunderstand
Use proper terminology, but explain anything that is domain-specific.

The glossary instruction matters more than it sounds. When you're new to a domain, trying to absorb terminology and concepts at the same time is the fastest way to understand neither. Separate the language from the knowledge. Learn the map first.

Layer 2 — YouTube for the How

Once I have the mental model, I go to my learning YouTube account for the actual implementation.

Search first, before asking AI for recommendations. The search itself teaches you something — what's popular, what's recent, what practitioners are actually making videos about. AI recommendations come in as a backup when I can't find what I need or want to go deeper on a specific angle.

This layer is about watching people build, not just explain. The goal is to see how something actually works in practice — the steps, the decisions, the things that break.

Layer 3 — Podcasts and Industry Leaders While Walking

When I'm learning something new, I notice I actually want to go for a walk.

Not because I'm disciplined about morning walks — I'm working on that. But because the podcast becomes the reason to go. Two things that should happen, pulling each other forward.

I look for people who are thinking about where an industry is going — founders, operators, investors, practitioners who have been in a space long enough to have a point of view worth hearing. Some shows I follow consistently. Others I find specifically per topic, searching first and asking AI for recommendations when I can't find what I need.

The difference between this layer and the others is cognitive mode. AI research and YouTube tutorials require active attention — you are processing, mapping, building. Walking with a podcast is passive absorption. Ideas settle differently when you are moving. Connections happen without forcing them.

Layer 4 — AI Voice Conversations While Walking

Sometimes instead of a podcast, I talk to ChatGPT on my phone while walking.

Not for structured learning. For questions I want to think through out loud — the kind of conversational, exploratory Q&A that would feel too slow at a desk. Why does this work the way it does? What are the implications of that? How does this connect to what I already know from another field?

These conversations are not useful in the traditional sense. They don't produce notes or structured output. What they produce is fluency — the ability to think about a topic without looking anything up.

Layer 5 — Back to AI for Application

After the research and the watching and the ambient absorption, I go back to AI — this time with a real problem.

Not "explain this to me." Not "quiz me." I describe something I am actually trying to build or solve and ask for help applying what I have learned. And when something breaks — which it always does — I bring the exact error, the exact context, and ask for the specific fix.

This is where learning actually locks in. Debugging something that you built is the fastest way to understand why it works the way it does. The mistake is more educational than the tutorial.

Layer 6 — Build Something

Everything above is preparation. The actual learning happens here.

Build the simplest possible version of whatever you are trying to understand. It will be imperfect. That is the point. The gaps between what you thought you understood and what you actually understood become visible the moment you try to build — and those gaps are where the real learning lives.

The stack is not about learning faster. It is about learning in a way that does not drain the rest of your life.

A Real Example: Moving Into a New Domain

This also helped when I had to move into the payments domain. I was not starting from zero — I understood systems, operations, client workflows, and technical implementation — but payments had its own language, architecture, flows, and assumptions. The system helped me build enough fluency to understand the moving parts, ask better questions, follow the logic behind transaction flows, and hold credible conversations without pretending I had become an expert overnight.

What This Is Not

This stack is not for every topic at full depth. I always start with AI research — that layer is non-negotiable because it sets the foundation. Everything after depends on how deep I need to go. Sometimes Layer 1 and a few YouTube videos are enough. Sometimes I run the full sequence.

It is also not a system I follow rigidly. It is a mental model for how different tools serve different cognitive purposes — and once you understand that, you stop treating YouTube, AI, and podcasts as interchangeable and start using each one for what it actually does well.

How I Know I've Learned Something

Not when I've finished the stack. Not when I've watched all the videos.

When I can have a credible conversation with someone who knows the field — not by pretending to be an expert, but by asking better questions, understanding the language, and knowing where the real problems usually sit.

That's the test.

Not finishing more content.

Becoming useful faster.

---

Written from experience, not expertise. More on LinkedIn.