Spotify Says Its Best Developers Haven’t Written a Line of Code Since December, Thanks to AI

Has AI-driven coding reached a major inflexion point? At Spotify, executives suggest it may have. During the company’s fourth-quarter earnings call, co-CEO Gustav Söderström revealed that some of Spotify’s top developers “have not written a single line of code since December.” The comment came as part of a broader discussion about how artificial intelligence is transforming the company’s product development process.

Spotify noted that it delivered more than 50 new features and updates to its streaming platform throughout 2025. In recent weeks, the company has launched additional AI-powered features, including Prompted Playlists, Page Match for audiobooks, and About This Song.

To support faster engineering cycles, Spotify’s teams are using an internal AI system known as “Honk.” The platform leverages generative AI, including Claude Code, to accelerate coding workflows and enable rapid product deployment.

Söderström shared a specific example during the call. An engineer commuting in the morning can send a request via Slack from their phone, instructing Claude to fix a bug or implement a feature in the iOS app. After the AI completes the task, an updated version of the app is delivered back to the engineer through Slack. The engineer can then review and merge the changes into production — all before reaching the office.

Spotify told analysts that this system has dramatically increased coding speed and deployment efficiency.

Söderström added that the company sees this as just the beginning of AI’s role in development. He suggested that the impact of AI tools on productivity and product velocity is expected to grow further over time.

The executive also highlighted Spotify’s efforts to build a distinctive dataset that cannot easily be replicated or commoditised by other large language models. Unlike general information sources such as Wikipedia, music-related data often does not have one definitive answer. Preferences are shaped by personal taste, culture, and geography.

For example, workout music varies widely by region and individual preference. In the United States, hip-hop is broadly popular, though millions of listeners prefer genres like death metal. In many European countries, EDM is common for exercise, while in Scandinavia, heavy metal frequently tops workout playlists.

“This is a dataset that we are building right now that no one else is really building. It does not exist at this scale. And we see it improving every time we retrain our models,” Söderström said.

During the earnings call, analysts also asked about Spotify’s approach to AI-generated music. The company explained that artists and record labels may indicate in a track’s metadata how a song was created, including whether AI tools were used. At the same time, Spotify said it continues to monitor the platform closely to prevent spam and maintain quality standards.