Simon Willison admitted this week he skips line-by-line reviews in production code. A 25-year veteran programmer no longer reads every line AI writes—we believe this is more alarming than any tech breakthrough.

What this is

Willison previously drew a line: vibe coding (coding by feel: ignoring the code, accepting AI outputs, asking it to fix breaks) suits personal tools; agentic engineering (professional engineers leveraging AI while controlling quality) is the proper path for production software. The core distinction is whether you care about code quality, security, and maintainability. But now, when Claude Code writes a JSON API endpoint with near-zero errors, even veterans skip line-by-line checks—the two paths are quietly converging.

Industry view

Optimists see this as the natural result of tool maturity, like programmers no longer reviewing machine instructions after high-level languages replaced assembly. Willison himself says AI lets him tackle larger projects, with 25 years of experience remaining the quality baseline. But we note a clear slippery slope: once review habits loosen, "lower quality but faster" quietly replaces "high quality and faster." Willison described his own feelings as "quite upsetting." More concerning, inexperienced developers may not even realize they are vibe coding—without 25 years of experience as a calibrator, the gap between "appears functional" and "truly reliable" is systematically underestimated.

Impact on regular people

For enterprise IT: The barrier to internal tool development drops, but vibe coding is quietly infiltrating production systems, potentially accelerating technical debt accumulation.

For individual careers: Non-programmers can build prototypes via prompts, but the chasm between "can prototype" and "can deliver a reliable product" hasn't disappeared; it's just ignored by more people because the tools are so handy.

For the consumer market: SaaS iteration accelerates. Whether software quality drops is hard to perceive short-term, but long-term it may surface as data breaches or service outages.