During Spotify’s fourth-quarter earnings call, co-CEO Gustav Söderström stated that the company’s best developers have not written a single line of code since December, attributing this to AI tools that accelerate development processes.
Söderström made the remark while discussing Spotify’s integration of artificial intelligence into its engineering workflows. The company reported shipping more than 50 new features and changes to its streaming app throughout 2025. These updates encompassed a range of enhancements designed to improve user experience across the platform.
Among the most recent additions, Spotify launched AI-powered Prompted Playlists, which generate customized playlists based on user prompts. Page Match for audiobooks provides functionality that matches specific pages or sections within audiobook content. About This Song offers detailed information on individual tracks. All three features became available to users within the past few weeks.
Spotify engineers rely on an internal system named Honk to enhance coding efficiency and product deployment speed. Honk facilitates remote, real-time code deployment through generative AI, with specific use of Claude Code. This setup allows engineers to manage development tasks without direct coding involvement.
Söderström provided a specific example of Honk’s operation. “As a concrete example, an engineer at Spotify on their morning commute from Slack on their cell phone can tell Claude to fix a bug or add a new feature to the iOS app,” he said. “And once Claude finishes that work, the engineer then gets a new version of the app, pushed to them on Slack on their phone, so that he can then merge it to production, all before they even arrive at the office.”
The company attributes significant improvements in coding and deployment velocity to Honk. Executives described the speedup as tremendous during the earnings call discussions with analysts. Söderström emphasized the early stage of these advancements, stating, “We foresee this not being the end of the line in terms of AI development, just the beginning.”
Beyond deployment tools, Spotify leverages a proprietary dataset tailored to music-related queries. This dataset stands out because responses to such queries lack universal factual answers and differ based on individual preferences and geographic locations. For example, preferences for workout music vary: Americans tend to prefer hip-hop overall, though millions favor death metal. Many Europeans select EDM for workouts, while numerous Scandinavians choose heavy metal.
Söderström highlighted the dataset’s exclusivity and growth. “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,” he noted. Spotify constructs this resource through ongoing data collection, enabling model refinements with each retraining cycle.
Regarding AI-generated music, Spotify permits artists and record labels to specify production methods in a track’s metadata. This includes details on whether AI contributed to the song’s creation. Simultaneously, the platform maintains active monitoring to detect and remove AI-generated spam content.
