Where is all the new, good software?

The article explores why highly-capable AI coding hasn't produced an obvious surge in new, excellent software despite expectations. It offers multiple explanations: improvements exist but are incremental (fewer bugs, better features), adoption rates are low, non-technical bottlenecks in development processes persist, AI is being used for ephemeral solutions rather than permanent products, and much AI-generated software remains private or internal to organizations rather than publicly visible. Additionally, teams may be using AI efficiency gains for leisure rather than output, and AI sometimes encourages less software use overall.