๐ค Curiosity: Why Does More Context Sometimes Make an Agent Worse? Eight years shipping AI systems into production games taught me to distrust a very seductive idea: that if an agent just had a lon...
๐ค Curiosity: What If Your Coding Agent Came with Its Own Skill Tree? Eight years shipping AI games at NC SOFT and COM2US taught me one thing: the bottleneck is never raw intelligence โ itโs the ha...
๐ค Curiosity: If Every Tool Needs Memory, Who Should Own It? I now run four coding agents on any given week. Claude Code for the long refactors, Codex when I want a second opinion, Cursor when I am...
๐ค Curiosity: What Happens When You Show a Model Its Own Answer Key? Eight years of shipping AI systems at NC SOFT and COM2US taught me to distrust a passing test I did not write carefully. Not bec...
๐ค Curiosity: Why Do We Still Read Logs By Hand? I have shipped AI into games for eight years. We let neural networks pick matchmaking brackets, tune difficulty curves, and drive digital humans โ a...
๐ค Curiosity: The Game Build Problem We Never Solved Hereโs what keeps game development teams awake at night: Itโs 3 AM. Your studio just shipped a critical bug fix to production. But before the p...
๐ค Curiosity: Why Does Every Agent Framework Rot the Same Way? Iโve now built or inherited enough agent scaffolding to recognize the decay pattern, and itโs always architectural rather than technic...
๐ค Curiosity: What Happens When an Agent Writes to Your Files? I keep running the same experiment across every AI-agent project I ship: give an agent write access to a folder, let it run for a few ...
๐ค Curiosity: What if the Unity Editor were a dependable production surface? The most expensive part of an automated game workflow is often not the build itself. It is the human handoff around it: ...
๐ค Curiosity: Why Do We Use Days-Long Models in 5-Minute Sessions? Hereโs a pattern I keep seeing, and Iโve been guilty of it myself: a genuinely new class of model ships, and we use it exactly lik...