
Quality, Maintainability, and Vibe Coding Risks

Data Beats Hype: Measuring Your AI Adoption Impact - Laura Tacho
Topic 9 of 14
Published Sep 29, 2025 · Clip 38:30–46:10 (7:40)
Laura discusses the paradox of AI code quality and maintainability: it's trivial to produce lots of code now, but it's unclear how AI-authored codebases will hold up over time since we lack longitudinal data. She muses on whether AI is a paradigm shift analogous to moving from Assembly to higher-level languages, possibly leading to AI code meant to be read by other AI agents. She describes her experiment asking Cursor to dockerize an app twice and getting totally different (though both technically correct) results, illustrating non-determinism. Teams with resilient, observable systems are best positioned for AI gains, and vibe coding is fine for proof-of-concepts but a recipe for disaster at enterprise scale without constraints, linting, and code standards.