Gregorein used a single Anthropic Claude session to review front-end files, estimating commits at ~2K lines added/450 removed, arguing AI enables code volume exceeding human review capacity, degrading quality without scrutiny—likening it to "move fast and break things." Defenders like developer Elvis Sun (Medialyst.ai founder) countered that engineers should build self-improving AI systems monitoring analytics, shifting roles from gatekeepers to architects; Gregorein noted the site had months of data yet no fixes until manual inspection.[Input]
This stems from Tan's ongoing AI acceleration advocacy, including YC trends where ~25% of W25 cohort startups have 95% AI-generated codebases, enabling smaller teams and rapid output (e.g., Tan's prior 10K LOC/day claims at SXSW).[1][4][7] Newsworthy amid 2026 AI coding boom (e.g., "vibe coding," YC tools like Inspector), it spotlights quantity-vs-quality debate: AI boosts productivity but risks bloat/security issues without engineering oversight, as Tan dismissed critics with a "haters" post on April 1.[Input][2][3] Story published April 2, 2026.[Input]