Spotify Engineers Rely on AI, Stop Writing Code

Share


Spotify’s most experienced software engineers haven’t written a single line of code in months, as artificial intelligence (AI) fundamentally reshapes how the streaming giant develops its products.

Speaking during the company’s fourth-quarter earnings call, co-CEO Gustav Söderström revealed that AI is now deeply integrated into Spotify’s engineering workflow, generating and deploying software with minimal human coding. “Our most experienced developers have not written a single line of code since December,” he said.

Rather than manually programming, engineers now supervise an internal AI system called “Honk,” which combines generative AI—including Anthropic Claude Code—with real-time development tools. This setup allows engineers to direct, review, and refine AI-generated code instead of producing it themselves. For instance, an engineer can use Slack on their morning commute to instruct Claude to fix a bug or add a new feature to the iOS app.

Spotify says this AI-driven approach has accelerated development and deployment. In 2025 alone, the company shipped over 50 app features and updates, including AI-powered tools like Prompted Playlists, Page Match for syncing physical books with audiobooks, and “About This Song,” which offers contextual storytelling.

Söderström emphasized that Spotify’s vast user data is a key competitive advantage, enabling the company to train and refine AI models at a scale unmatched by competitors. “This is a dataset that no one else is really building. It improves every time we retrain our models,” he said.

While the changes are significant, Söderström described them as an early stage of a broader transformation, signaling deeper AI integration across Spotify’s software engineering in the future.


Recent Random Post: