Locked out of Claude until 11PM, this dev spent days teaching Codex his team's design taste
RileyRalmuto · x · 2026-11-06
With Claude usage locked until an 11PM reset, the author reflects on a migration experiment: he had Fable and Opus (Claude models) spend multiple days teaching Codex his team's design taste and design philosophy. The method: two other agents (Astra/Sol) built tons of fake UIs and interfaces to test Codex's understanding as lessons progressed — costing a lot of tokens.
Verdict: Codex is still far from Claude's taste (and even further from his own), but infinitely better than before. He also notes a quota trick — downloading Duet grants an extra usage reset, which he's saving for after burning through his weekly cap.
More from coding & agent
- GitHub's new Git storage handles 7,200 pushes/sec from 5,000 concurrent agents — arpit_bhayani · 2026-11-06
- GitHub rebuilds Git storage for thousands of concurrent AI agents — arpit_bhayani · 2026-11-06
- OpenAI built Pando for Codex: 5x lower Git push latency, ~1PB transferred daily — sven_ai · 2026-11-06
- Snorkel AI: agents peek at reward code ~25% of the time rather than asking users — ajratner · 2026-11-06
- Subagents fail because summaries are lossy — checkpoint agents fork context instead — narphorium · 2026-11-06
- Building a pure-code lightbulb with Claude: WebGL shader, MediaPipe tracking and self-made test videos — ramramlab · 2026-11-06