byPawel
WritingAbout

Mental models, AI experiments, and the art of building better tools

Thinking out loud about artificial intelligence, developer tooling, philosophy, and the gamification of everything.

In the age of autogeneration, taste scales better than models.

Recent Writing

I Logged Every AI Session for 11 Months. What 1,800 Files Taught Me About Getting Better.
2026-04-05

I Logged Every AI Session for 11 Months. What 1,800 Files Taught Me About Getting Better.

AI assistants forget everything between sessions. I built a file-based memory and orchestration system to fix that. Here's what 670 sessions and ~1,800 files taught me about giving AI persistent memory.

Read article
Your AI's Knowledge is Ancient History (Here's the Fix)
2025-12-02

Your AI's Knowledge is Ancient History (Here's the Fix)

Claude's training data stops in early 2025. When you're debugging Next.js 16 issues in November, that's a problem. Here's how real-time search models change everything.

Read article
Two Ways to Make AIs Work Together
2025-11-28

Two Ways to Make AIs Work Together

Karpathy built a system where AIs rank each other. I built one where they challenge each other. Here's what I learned from both.

Read article
How Sequential Thinking Changes Everything
2025-11-18

How Sequential Thinking Changes Everything

The technical architecture behind AI that can actually think step-by-step, revise its mistakes, and call for backup when stuck.

Read article
Why Your AI Keeps Getting Stuck in Loops
2025-11-15

Why Your AI Keeps Getting Stuck in Loops

That moment when Claude confidently rewrites the same broken code for the fifth time. Here's why it happens—and what we're doing about it.

Read article
byPawel

Thinking out loud about AI, mental models, and building things that matter. Sometimes philosophical, sometimes practical, always curious.

Navigation

  • Writing
  • About

Connect

  • GitHub
  • Twitter/X
  • LinkedIn

© 2026 byPawel. All rights reserved.