Agents have made CI the bottleneck. Faster pipelines are the wrong fix.
Three posts landed in September that I think engineering leaders should read together. Anthropic’s engineering team wrote that their continuous The post Agents have made CI the bottleneck. Faster pipelines are the wrong fix. appeared first on The New Stack .
Three engineering leaders recently shared their experiences with continuous integration (CI) in September. Anthropic reported a 25x increase in CI job volume over six months, while Linear's test suite quadrupled since January. Depot's CEO emphasized the need for agents to validate code and maintain trust as they work. However, the solutions proposed seem to address the wrong layer.
CI was originally designed for human output, with developers submitting a few pull requests (PRs) per week. However, with the rise of agents, the volume of PRs has surged, causing CI pipelines to become a bottleneck. The reason for this is twofold: the increased volume and the delay in the verification process. Agents not only write code but also open PRs and wait for the results, which can take 20 minutes.
By the time the result comes back, the developer has lost their working context, leading to a full round trip for every failure.
The industry responded by making CI faster, using faster runners, smarter test selection, bigger caches, and pipelines that agents can call before committing. While these improvements are necessary, they do not address the underlying issue: CI is still verifying a repository instead of the entire distributed system. For cloud-native applications, a repository represents only one service out of many. Tests in a repository exercise that service but ignore the rest of the system.
When a change passes through various stages, from the agent to CI to staging, it is verified against a copy of the code rather than the entire system. This means that even if the verification process is faster, it does not provide comprehensive validation. The verification loop needs to move from the repository to the system, ensuring that the change works with all the interconnected services.
This shift is necessary to prevent breakage that occurs in distributed systems, where failures can cascade through multiple services.
Written by urgent.news from The New Stack's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
