Generative AI coding tools and agents do not work for me

4.5

“Generative AI coding tools and agents do not work for me” is a thoughtful, experience-driven essay that critically examines the current wave of AI coding assistants. Written by veteran engineer Miguel Grinberg and featured on Hacker News, the article goes beyond hype to explain why many developers feel less productive, less in control, and more frustrated when trying to integrate generative AI into their daily workflow. Instead of offering yet another tool or framework, this piece functions as a reality check and a practical guide to thinking clearly about AI-assisted programming. It covers real-world issues such as inaccurate code suggestions, hidden maintenance costs, diminished understanding of codebases, and the risk of outsourcing too much design thinking to opaque models. Grinberg explains how these factors impact debugging, long-term code quality, and a developer’s ability to reason about complex systems. This resource is especially valuable for engineers, tech leads, and engineering managers who are evaluating whether, when, and how to adopt AI coding tools. It can help teams set realistic expectations, design better workflows, and avoid common productivity traps. If you’re feeling pressure to “just use AI” for coding but your experience doesn’t match the marketing promises, this article gives you a clear vocabulary and framework to articulate those concerns and make more grounded decisions about AI in software development.