When AI writes almost all code, what happens to software engineering? (newsletter.pragmaticengineer.com)

Recent advancements in AI coding tools, particularly models like Opus 4.5 and GPT-5.2, are dramatically reshaping the landscape of software engineering, leading to a significant trend where AI is poised to write the majority of code. Developers have reported impressive outcomes while experimenting with these AI tools, producing substantial code outputs with minimal manual effort. Notably, one engineer managed to build and deploy software entirely from a mobile device using AI agents, marking a paradigm shift in the ease and accessibility of software development. This evolution suggests that traditional roles within software engineering may need to adapt, placing more emphasis on leadership and product management skills while potentially diminishing the value of niche technical expertise. As witnessed by both AI tool creators and independent engineers alike, the recent model releases have led to a tipping point in capabilities. Software engineers report "a-ha" moments where complex coding tasks have become manageable for AI, which now operates almost at a senior developer level. However, this shift raises concerns regarding software quality and the implications for work-life balance, as the influx of AI-generated code could exacerbate existing software engineering challenges. Overall, the AI's role in coding is developing rapidly and may redefine the future expectations and responsibilities of software engineers across the industry.