Spotify has revealed a significant shift in its software development process, stating that some of its top engineers have not manually written code since December. During a recent earnings call, the company detailed how it is using an internal AI system to accelerate product development, allowing engineers to deploy new features and fix bugs remotely, even from their mobile phones.
This move highlights a growing trend where generative AI is not just an assistant but a primary tool for creating and maintaining software. The company credits this AI-driven approach with a dramatic increase in its ability to ship new features, with over 50 updates released in 2025 alone.
Key Takeaways
- Spotify's top developers have reportedly stopped writing code manually since December, relying on AI tools instead.
- An internal AI system named "Honk" allows engineers to use generative AI to write and deploy code.
- The system enables developers to fix bugs or add features to the Spotify app from their phones using Slack.
- Spotify attributes a significant increase in feature rollouts, including over 50 in 2025, to this AI-powered workflow.
- The company is building a unique dataset for music recommendations that it believes cannot be easily replicated by other large language models.
A New Era of Software Development
The music streaming giant is pioneering a workflow that could redefine the role of a software engineer. According to Spotify co-CEO Gustav Söderström, the company's internal system, called "Honk," is at the center of this transformation. This platform integrates generative AI, specifically Claude Code, to handle complex coding tasks based on simple text commands.
Söderström provided a striking example of this new process during the company's fourth-quarter earnings call. He described a scenario where an engineer can issue a command through Slack on their phone while commuting.
"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. 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 theoffice."
This capability represents a major leap in development speed and flexibility. It effectively removes the traditional constraints of needing a dedicated workstation and physical presence to write, test, and deploy code. The company stated this new method has sped up coding and deployment "tremendously."
The 'Honk' System and Product Velocity
The internal system, "Honk," is more than just a coding assistant. It's a comprehensive platform for remote, real-time code deployment. By integrating AI directly into the developer workflow through familiar tools like Slack, Spotify has reduced friction and accelerated its entire product lifecycle.
This increased speed, or "product velocity," is evident in the number of new features Spotify has recently launched. In the past few weeks alone, the platform has introduced several AI-powered functionalities:
- AI-powered Prompted Playlists: Allows users to create playlists using descriptive text prompts.
- Page Match for audiobooks: A feature that enhances the audiobook listening experience.
- About This Song: Provides users with more context and information about the music they are listening to.
These launches, which followed more than 50 other new features and changes in 2025, underscore the practical impact of using AI in development. The company is not just experimenting with AI; it is using it to deliver tangible value to its users at an unprecedented rate.
From Commute to Code Deployment
Spotify's AI system allows an engineer to identify a bug, instruct the AI to write a fix, receive a testable new version of the app, and approve it for production release—all from a smartphone before reaching the office.
Building a Defensible Data Moat
While many companies are leveraging existing large language models (LLMs), Spotify is focused on creating a unique, proprietary dataset that it believes will provide a long-term competitive advantage. Söderström explained that unlike factual data that can be scraped from sources like Wikipedia, music preference is highly subjective and personal.
He pointed out that the definition of "workout music" varies dramatically across different cultures and individuals.
For example, while many Americans might prefer hip-hop for exercise, some may opt for death metal. In Europe, EDM is a popular choice for workouts, but in Scandinavia, heavy metal is also common.
This subjectivity creates a complex data challenge that generic AI models cannot easily solve. Spotify is actively building a dataset that captures these nuances of human taste at a massive scale.
The Subjectivity of Sound
Söderström emphasized the value of this approach. "This is a dataset that we are building right now that no one else is really building. It does not exist at this scale," he said. "And we see it improving every time we retrain our models."
By understanding the subtle, often unstated, connections between music, mood, activity, and culture, Spotify aims to deliver recommendations and user experiences that are far more personalized and accurate than competitors can offer. This unique data is a strategic asset that becomes more valuable as it grows, making it difficult for others to replicate.
The Future of AI in Music and Development
Spotify's executives see their current use of AI as just the beginning. The company is positioning itself not only as a music streaming service but as a technology leader pioneering new ways to build software and interact with content.
Handling AI-Generated Music
When asked about the rise of AI-generated music on the platform, Spotify explained its current strategy. The company allows artists and labels to use metadata to indicate if a track was created with AI. At the same time, it continues to actively police the platform for spam and fraudulent content to maintain quality standards.
The move to AI-assisted, and in some cases AI-led, coding suggests a future where the role of a developer shifts from writing lines of code to directing and supervising AI systems. This could lead to even faster innovation cycles and allow engineers to focus on higher-level problem-solving and creative product design.
As Spotify continues to refine its AI tools and expand its unique datasets, the line between technology and content curation will likely blur even further. The company's vision is clear: to use artificial intelligence to fundamentally change how software is made and how hundreds of millions of people discover and enjoy audio content.