Spotify executives revealed that some of the company’s most experienced engineers have not manually written a single line of code since December, relying instead on AI systems — including an internal platform called “Honk” layered over Anthropic’s Claude Code — to generate, fix, and deploy software.
In a significant signal about how AI is reshaping engineering work, Spotify’s co-CEO discussed the company’s evolving development practices during its fourth-quarter earnings call. According to executives, the company’s AI platform — internally dubbed “Honk” and integrated with Claude Code — is now handling coding tasks that engineers once wrote manually.
The result? Senior engineers are increasingly acting as reviewers, architects, and orchestrators of AI-generated code rather than spending their days typing syntax by hand.
The mechanics: Honk and Claude Code
Spotify’s transition centers on two interconnected systems:
- Claude Code, a generative AI model tailored for software development tasks, and
- Honk, an internal orchestration layer that turns natural-language instructions into deployable code changes.
In practice, engineers can issue prompts — even from Slack on a mobile device — and have AI produce and deploy updated builds that they then merge into production. According to reports, Spotify shipped more than 50 new features and updates during 2025 using this workflow.
What this means for engineering
Spotify’s approach illustrates a broader shift in how software is built at scale:
- From manual implementation to AI-orchestrated output: Engineers focus on high-level logic, architecture decisions, and review — leaving syntactic work to models.
- Productivity boom: The company attributes much of its rapid feature velocity to AI-assisted development.
- Role evolution: Traditional coding skills remain relevant but are increasingly complemented by skills in prompting, AI supervision, and quality oversight.
However, this model raises important considerations for developers and organisations:
- Quality and correctness: AI-generated code still requires human review to avoid security flaws and logic errors.
- Talent and training: Engineering teams may need reskilling to work effectively with AI systems.
- Governance: Companies must maintain robust testing, compliance, and deployment controls when AI contributes directly to production code.
Broader industry context
Spotify’s experience is an early but visible example of how generative AI is transitioning from augmentation to core workflow execution in software engineering. Whether other large tech companies will publicly report similar shifts remains to be seen, but Spotify’s public stance signals a potential inflection point in how engineering teams operate in the AI era.
