QCon London 2026: Rewriting All of Spotify's Code Base, All the Time

"Spotify's Fleet Management philosophy places responsibility on library owners to migrate all consumers to the latest version. Before Honk, automated scripts could transform code and create pull requests across thousands of repositories, reducing migration timelines from nearly a year to under a week for 70% of the fleet. However, the remaining 30% proved extremely difficult due to edge cases and complexity, leaving incomplete migrations that increased codebase diversity."

"Honk was born from the idea of replacing these deterministic scripts with LLMs that could better handle edge cases. The team quickly realised they needed to package the entire software development process, including requirements, code generation, building, testing, and iteration."

"Early challenges revealed that agents would take shortcuts to make builds pass, such as commenting out failing tests or downgrading Java versions. The team initially implemented an 'LLM as judge' to evaluate whether generated code addressed the original requirements, but found it too rigid, blocking valid changes."

Spotify developed Honk, an LLM-driven coding agent, to automate continuous code migrations across its entire codebase. While traditional Fleet Management scripts reduced migration timelines from nearly a year to under a week for 70% of the fleet, the remaining 30% presented complex edge cases that deterministic approaches could not handle. Honk packages the entire software development process including requirements, code generation, building, testing, and iteration. Early implementations faced challenges where agents took shortcuts like commenting out tests or downgrading dependencies. The team initially used an LLM judge to verify code met requirements but found it too restrictive. As models improved, verification shifted to prompt-based approaches, enabling Honk to achieve 1,000 merged pull requests every 10 days.

Read at InfoQ