Developers Stop Writing Code, Direct AI Teams Instead

Software developers are experiencing a workplace revolution that's turning traditional coding on its head, and the change is happening faster than anyone expected.

Following the recent launch of advanced AI coding models from OpenAI and Anthropic, engineers at companies like Spotify and Anthropic itself have fundamentally changed how they work. Instead of writing code line by line, they're now directing AI assistants that handle the technical details while they focus on bigger picture solutions.

The transformation is already showing real results. Spotify's best developers haven't written a single line of code since December, according to co-CEO Gustav Söderström. Engineers now send instructions to AI via Slack during their morning commute, and completed features are ready to deploy before they reach the office.

The streaming company shipped over 50 new features in 2025 using these AI-powered workflows. At Anthropic, the company creating one of these AI tools, about 70% to 90% of all code is now AI-generated.

The shift has been gradual over the past year as AI models became capable enough to handle complex tasks. But recent releases made the trend impossible to ignore, sparking conversations across the tech industry about what this means for software development careers.

Boris Cherny, who leads development at Anthropic, hasn't written code himself in over two months. His team built an entire new product in just a week and a half, with 90% of the code written by their own AI tool.

Perhaps most remarkable is that these AI systems are now helping build better versions of themselves. OpenAI says its latest model "was instrumental in creating itself," marking a new milestone in AI development.

The Ripple Effect

This transformation is creating opportunities beyond just faster coding. Developers are evolving from typists into architects, focusing on creative problem-solving and strategic thinking rather than syntax and debugging. The skill shift mirrors other professions throughout history that have elevated human workers when automation handled routine tasks.

The productivity gains are substantial, with some engineers effectively producing ten times their previous output. Companies can build features faster, respond to user needs more quickly, and bring innovations to market at unprecedented speed.

However, developers are also flagging an unexpected challenge: the risk of overwork. Veteran engineer Steve Yegge warns that the addictive nature of AI coding tools can push developers to take on unsustainable workloads, even as their output soars. Some teams are considering installing nap pods at their offices to help manage the intensity.

The tech industry is learning to balance these powerful new capabilities with human wellbeing, ensuring that increased productivity doesn't come at the cost of developer health.

For now, software engineers are pioneering what knowledge work might look like across other industries as AI tools become more sophisticated and widely available.