Has the evolution of AI in coding reached a significant milestone? It appears so for Spotify, which revealed during its recent fourth-quarter earnings report that its top developers have not engaged in traditional coding since December. This statement, made by co-CEO Gustav Söderström, highlights the company's innovative use of AI to enhance development processes.
Remarkably, Spotify has introduced over 50 new features and updates to its streaming platform throughout 2025. Recently, the company launched several exciting functionalities, including AI-driven Prompted Playlists, Page Match for audiobooks, and About This Song, all rolled out in the past weeks.
Spotify's engineering team is utilizing an internal tool known as "Honk" to accelerate coding and product delivery. This system enables real-time, remote code deployment through generative AI, specifically leveraging Claude Code.
As an illustrative example, Söderström explained how an engineer can communicate with Claude via Slack on their mobile device during their commute to request bug fixes or new feature additions for the iOS app. Once completed, the updated app version is sent back to them on Slack, allowing the engineer to merge it into production before even reaching the office.
The company credits this system with significantly enhancing coding efficiency and deployment speed.
"We believe this is just the beginning of AI development, not the end," Söderström stated.
Additionally, he emphasized Spotify's unique ability to create a dataset that other large language models cannot replicate, unlike common online resources such as Wikipedia. This uniqueness arises from the subjective nature of music-related inquiries.
For example, the notion of workout music varies widely among individuals based on geographical preferences. While many Americans lean towards hip-hop, others might favor genres like death metal. In Europe, preferences can differ significantly, with some opting for EDM and others gravitating toward heavy metal.
"We are currently building a dataset of this nature that is unparalleled in scale, and we observe improvements with each model retraining," Söderström remarked.
During the earnings call, analysts also inquired about Spotify's stance on AI-generated music. The company clarified that it allows artists and labels to specify in a track's metadata how the music was created, while also maintaining strict oversight to prevent spam on the platform.