Spotify uses AI to write most of its code, CEO reveals

Speaking on the call, co-CEO Gustav Söderström said that Spotify engineers “have not written a single line of code since December,” attributing the rapid development of new features to the company’s internal AI system, “Honk.”

The platform has introduced over 50 new features and updates to its mobile app in the past year, including AI-powered Prompted Playlists, audiobook Page Match, and the About This Song feature.

Söderström said Honk enables engineers to deploy code remotely and in real time using generative AI powered by Claude Code. “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 they can then merge it to production, all before they even arrive at the office.”

He added that Spotify is building a unique music-related dataset that other large language models cannot replicate, drawing from the platform’s extensive and globally diverse user base. Söderström cited regional differences in musical preferences as an example: “Americans might prefer hip-hop, many Europeans might prefer EDM, while Scandinavians might prefer to work out to heavy metal. This is the type of dataset that Spotify is building, which sets it apart, because the company has the largest swath of users from which to pull data.”

“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 said.

Spotify also outlined its approach to AI-generated music. Artists and labels can specify in a track’s metadata the extent to which AI contributed to its creation. The platform continues to monitor for low-quality or “slop” AI-generated tracks, according to the company.