Simon Willison shares insights from his appearance on Lenny Rachitsky's podcast, focusing on the recent advancements in AI and coding tools. He identifies a significant shift that occurred in November with the release of GPT 5.1 and Claude Opus 4.5. These models marked an inflection point in coding efficiency, drastically improving the reliability of generated code. Willison notes that tasks that once required careful oversight now yield results that are often functional, allowing developers to focus on higher-level problem-solving rather than micromanaging code output. He highlights the evolving role of software engineers as indicators for other information workers, emphasizing that while coding is straightforward—either it works or it doesn’t—evaluating complex outputs like essays or legal documents remains challenging. Willison describes his shift to writing code on his phone, leveraging AI tools to streamline his work, and introduces the concept of “dark factories” in software development, where automation reduces human involvement in coding. He suggests that the bottleneck in software development has now shifted from coding to testing, as rapid prototyping becomes the norm, allowing for quick iterations and more efficient product design. Willison also addresses the mental toll of adapting to these new tools, finding that using multiple coding agents can be exhausting. He warns against the temptation to overextend oneself by constantly utilizing these AI agents, which can lead to burnout. Lastly, he points out that interruptions are less disruptive than before, as the new tools enable programmers to recover quickly and maintain productivity despite breaks in focus.