Google engineers are shifting from coding to leading software decisions

Industry reporting and a Google Cloud research program say AI now writes a large share of code, and engineers are moving from hands-on coding toward design, management and overseeing AI tools while companies provide training and embedded support.

AI has taken over much of the routine work of writing code, and that change is prompting a reassessment of engineers' day-to-day roles. Research and company statements show widespread AI use in software development, and companies such as Google are updating how teams learn and operate. Experts describe a shift from syntax and manual coding toward higher-level design, judgment and managing AI agents. This transition is driving new internal support systems and training across engineering teams. Key facts: - Google leadership has said AI agents are writing half of all code, and Ryan J. Salva, senior director of product management at Google, said the figure is "much, much higher now." - A 2025 Dora report from Google Cloud surveyed 5,000 tech professionals and found 90% of software development workers were using AI at work as of September, a 14% increase year over year. - The article notes examples from the industry: Spotify senior engineers reportedly have not written code since December, and Anthropic has been reported to use AI to produce 70–90% of its code. - Experts including NYU professor Julian Togelius describe developers shifting to roles that emphasize design, architecture, people management and oversight of multiple AI coding agents. - Dora's report found 80% of software development professionals feel AI has increased productivity, while observers warn that managing many agents and frequent context switching can contribute to burnout and a sense of reduced control over time. - Google is responding by embedding employees within engineering teams to track tools and hold workshops or office hours, and by focusing work on using AI for maintenance and scaling of deployed applications. Summary: The rise of AI-generated code is changing what many firms expect from software engineers, who are increasingly focused on higher-level decisions and managing AI tools rather than manual code authoring. Companies are creating internal training and support roles to help teams adapt, and work this year includes a stronger emphasis on using AI for maintenance and operations of deployed software.