Developer Finds AI Coding Tools Cut Project Time but Warns Against Trusting Code Alone

A developer built two projects using different versions of OpenAI's GPT models — EngHub in roughly a year and AlbumMap in about two months — and noticed a significant difference in workflow efficiency. He attributes the faster timeline partly to GPT-5.5 feeling more capable, but also acknowledges that prior experience, project scope, and differing requirements played major roles. Rather than claiming a simple speed multiplier, he argues the more meaningful takeaway is that AI tools are narrowing the gap between an idea and a working product. However, he cautions that evaluating AI coding assistants solely on generated code quality can be misleading, since visually functional UI can still produce incorrect end-to-end behaviour. He emphasises that testing real product outcomes — such as whether approved content appears in a final video export — matters far more than whether the code looks clean.