Are you narrow or general intelligence, human?.

What I cannot create, I do not understand.

Good Product Management boils down to:. Consistently high success rate on value-driving experiments. Great search heuristics (PMs are basically RL agents optimizing feature-value space). Balancing sharp product taste with rapid iteration to refine it.

First principles thinking is slower but accurate - grounded in fundamental truths. Pattern matching is faster but risks errors when reality diverges from past patterns.

It’s weird how we carve continuous spectrums into distinct chunks; like “red” or “loud.” Neural nets think in floating points; maybe all concepts are just fuzzy ranges? Are distinctions just artifacts of our wetware?.

engineering is about building, and science is about understanding.

White-collar jobs are first to be disrupted by general intelligence - cloud works fine at desks. Blue-collar roles need AI at the edge to control robots in real-time. We’ll have to wait a few years for GPUs powerful enough to shrink into mobile form factors.

With junior hiring slowing down, schools need to adapt. Teach AI-assisted coding so students can leapfrog straight into the hard stuff: distributed systems, data architecture, latency. The stuff only seniors used to learn on the job.

Saying neural nets are “just matrix multiplications” is like saying your favorite movie is “just pixels on a screen.” Technically true, entirely misses the magic.

AI might reshape distribution by shifting discovery from marketing to truth-seeking. Will autonomous AI buying agents tilt the game toward PLG startups? If product finding becomes about truth, authentic value wins.

Top advice for young people: learn to use AI more.

When I said “I don’t code” - It was as odd to them as if I told them that I couldn’t walk or dance, even though I have both my legs. Paraphrase From.This is Your Brain on Music.Daniel Levitin.

Keeping up with tech change is today’s most critical skill. It demands intuition, exponential thinking, knowing AI’s strengths & limits, and mastering AI tools - essentially becoming a cyborg.

Lowering the entry barrier to coding doesn’t diminish the jaw-dropping, frontier-pushing work happening at the edges - it just means more people get to experience the magic.

Feeling like Remy from Ratatouille - “Anyone can code!”.

I love it when an AI argues with me.

The cost of coding is trending toward zero - but building great software isn’t. Thoughtful design, polished UX, care, attention, and human judgment remain as valuable as ever.

Customers - not you - decide if your product is compelling. Your job: Rapidly explore the product space with smart heuristics. Success is measured by hit rate over time - don’t mistake a few samples for statistical significance.

I just want to build.

There’s a big difference between asking AI to speak for you vs. asking it to translate your ideas into a form that resonates broadly. One bypasses thinking; the other clarifies it.