Aiden Bai vs Sentry CEO: is human code review the real bottleneck for AI coding?

aidenybai · x · 2026-11-22

Aiden Bai argues human review is fundamentally a bottleneck in AI-generated code, with 90% of it performative; if frontier tokens got 100x cheaper, humans should shift to higher-leverage verification loops. Sentry CEO David Cramer pushes back: it's not about token budget — agents get one-sentence prompts and must mind-read intent, producing unpredictable outcomes. Unless developers write precise specs instead of precise code, human review and iteration can't be bypassed.

Related event: Aiden Bai and Sentry CEO Debate Whether Human Code Review Is the AI Coding Bottleneck(3 posts)→

Original post →

More from coding & agent

coding & agent channel →