I spent a month writing 33,459 lines of code to create a stable video translation function, and you say you can do it with a single skill? Tell me, am I dreaming or are you dreaming? For certainty, use engineering; for tolerating uncertainty, use AI.

Using Codex also brings new pain points: Slow and cumbersome architecture migrations are a common problem in every AI system, and this is particularly evident in GPT 5.4, with various obscure compatibility layers. Furthermore, Codex requires adding tests every time it's modified, resulting in more test files than implementations! While this seems stable, it's incredibly tedious. The result is that refactoring is extremely slow! And every time, I have to say "continue," even after sitting in front of my computer for two whole days, Codex is still doing the same thing! The image below is Claude's sharp review of Codex:

The assembly process of a quantum computer:

Being surrounded by people smarter than you is an opportunity, not a competition.

A terrifying thing: In the future, people who use different AIs will have completely different worldviews! You'll deeply understand this once you've actually used two different AIs.

Current AI Cost-Effectiveness Ranking: 1. GPT team (3.3 parking spaces on Xianyu) 2. Google Pro Family Plan ($20 with 6 accounts) 3. MiniMax 2.7 (29 RMB per month, 600 model calls in 5 hours)

Feynman Learning Technique

A terrifying thing: In the future, people who use different AIs will have completely different worldviews! You'll deeply understand this once you've actually used two different AIs.

I spent a month writing 33,459 lines of code to create a stable video translation function, and you say you can do it with a single skill? Tell me, am I dreaming or are you dreaming? For certainty, use engineering; for tolerating uncertainty, use AI.

There are things you need to know before you use OpenClaw.