Spotify's top developers have stopped writing code by hand, thanks to AI

# Spotify abandons hand-written code: how AI reshaped the work of its top engineers

Imagine that your job, writing code, is suddenly almost entirely automated. This is not a dystopian scenario but the reality Spotify's best engineers have lived with since last December. The streaming giant has disclosed an unusual fact: its leading developers have practically stopped writing lines of code by hand. Instead, they use Anthropic's Claude Code and Honk, Spotify's own internal platform, which have turned building software from scratch into a process of curation and architectural design. The development shows that the industry has passed a turning point, not in the sense that machines are displacing people, but in that the very substance of engineering work is being rapidly redefined.

What is going on here? Spotify did not simply plug in yet another AI assistant. The company built an ecosystem in which human craft and artificial intelligence work in close tandem. Claude Code, Anthropic's advanced AI assistant for writing code, has become the main tool that does the heavy lifting of translating high-level ideas into working code. In parallel, Honk, an in-house Spotify development, tunes the process to the company's specifics by learning its architecture, conventions and development patterns. The result: engineers no longer get stuck on syntax and routine implementation details.

But how does this work in practice? An engineer states the requirement and describes the needed functionality and architectural considerations. The AI system takes that specification and generates working code, often on the first or second attempt. The human does not disappear; they move higher up the decision-making hierarchy. Instead of spending hours debugging, the engineer checks the logic, considers edge cases, ensures architectural quality and makes strategic decisions about how the system should evolve. It resembles a move from assembly-line worker to project engineer.

For Spotify, the shift has tangible consequences. Development speed has risen dramatically. Tasks that used to take a day or several days are now done in hours. Code quality has improved in parallel, because experienced engineers can focus on architectural problems and system design instead of fighting syntax errors. The company also reports that people have to hold less code in their heads, since the AI handles many implementation details.

However, the case raises broader questions for the industry. If Spotify's top engineers no longer write code by hand, what does that mean for mid-level and junior developers? Skills need to be reinterpreted: knowing how to use an AI assistant effectively is becoming more important than mechanical mastery of syntax. In addition, the premium on architectural thinking and system design is rising, which is exactly what machines do worst.

Spotify is demonstrating not the end of the programmer's era but the transformation of the role. Engineers are becoming directors of AI systems, architects and strategists. The question is not whether people can stay relevant, but whether they will be ready to rethink their skills. For those who adapt, there is a chance to work at a higher level of abstraction. For the rest, the story may be less comforting.