Has AI coding reached a tipping point? That’s the case for Spotify at least, which was shared this week its fourth quarter earnings call that the company’s best developers “haven’t written a line of code since December.” That statement, from Spotify co-CEO Gustav Söderström, along with other comments about how the company is using AI to accelerate development.

In the note, Spotify points out that it will ship more than 50 new features and changes to its streaming app throughout 2025. And, most recently, it has launched more features, such as AI-powered Prompted Playlists, Match Page for audiobooks, and About This Songall of which were launched within the last few weeks.

At Spotify, engineers use an internal system called “Honk” to speed up coding and product speed, the company told analysts on the call. This system allows for things like remote, real-time code deployment using generative AI, and Claude Code in particular.

“As a concrete example, a Spotify engineer 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,” Söderström said. “And when Claude has finished that work, the engineer can then get a new version of the app, pushed to them in Slack on their phone, so he can merge it into production, all before they get to the office.”

Spotify credits the system with helping speed up coding and deployment “tremendously.”

“We see that this is not the end of the line in terms of AI development, it’s just the beginning,” Söderström said.

The exec also affirmed Spotify’s ability to create a unique dataset that cannot be commoditized by other LLMs, the way they can with other online resources, such as Wikipedia. That’s because there isn’t always a real answer for music-related questions, he said.

For example, if you ask what exercise music is, you will get different answers from different people, sometimes based on their geography. Americans prefer hip-hop in general, although millions prefer death metal. And while many Europeans work in EDM, many Scandinavians prefer heavy metal.

“This is a dataset that we are building today that no one else has built. It is not on this scale. And we see that it is improving every time we retrain our models,” Söderström said.

Analysts on the call also asked about Spotify’s approach to AI-produced music. The company explained that it allows artists and labels will be shown in the metadata of a track how the song is made but the platform is still monitored for spam.