Our top engineers haven’t written code in months, says Spotify’s co-CEO
Shortly after Spotify announced a subscription price increase, the company’s CEO shed light on its new operating model, with AI front and center in coding and engineering workflows.
During the company’s Q4 earnings call, co-CEO Gustav Söderström said that since December, the company’s top developers have written no code manually at all, instead relying on an internal AI system called “Honk.” He praised the tool for dramatically accelerating software development in the organization.
Honk appears to be primarily based on the AI tool Claude Code, developed by Anthropic and widely regarded as the best coding tool among programmers and developers. According to Söderström, Honk enables engineers to deploy and modify code in real time, even remotely.
Söderström gave the example of a developer using Slack on their phone during their morning commute to ask the AI to fix a bug or introduce a new feature to the iOS app. By the time they arrive at work, the updated build is ready on their device and can be merged into production immediately.
He characterized the impact as significant, noting that the system has greatly accelerated development speed, allowing the company to ship 50 new products and features in the past year alone. Söderström also suggested that Spotify is building a large, proprietary AI training dataset through this process and expects continued improvements as the models are retrained.
Although praised by investors, Söderström’s statements faced pushback from other sides. Users expressed dissatisfaction with ever-increasing subscription prices, even as AI reduces costs. Cybersecurity experts also warned of threats and vulnerabilities, noting that prior cases have demonstrated that AI code can pose serious security risks.
While the company has introduced dozens of new features and AI-driven tools over the past year, the growing role of automation in writing and deploying code raises broader questions about how increased AI adoption could affect the future demand for software engineers.
