AI Coding Surge Exposes CI Infrastructure Bottlenecks
AI-generated image · Bay Street Wire
Linear re-engineers its continuous integration pipeline to handle the volume of AI-generated code commits.
The acceleration of software development via AI agents is creating a performance gap in validation pipelines. As Linear first reported, the company found that while agents have made shipping code exponentially faster, continuous integration (CI) has become a bottleneck that increases infrastructure costs and delays developer feedback.
To address this, Linear CTO Tuomas tasked the team with reducing CI costs and increasing speed. The company implemented four primary optimizations to its TypeScript-based codebase:
* **Infrastructure Upgrades:** Linear migrated workloads from GitHub Actions to third-party runners with higher-performance storage and CPUs. This resulted in jobs running 34% faster on average, with some workloads decreasing in time by 52%. * **Toolchain Modernization:** The team utilized tsgo, the native TypeScript compiler, which reduced the weekly median of the tsc check by 73%. Additionally, rewriting custom lint rules to use static analysis instead of TypeScript type information reduced full-repository lint time by 55% and API lint time by 68%. * **Job Optimization:** Linear focused on critical-path jobs, such as change-detection. By capping fetch depth and removing unnecessary working trees, the median duration of change-detection jobs dropped from 26 to 8 seconds. * **Resiliency Improvements:** To combat network instability between third-party runners and GitHub, Linear replaced the standard checkout action with a custom composite action featuring retries and backoff.
Despite test suites nearly quadrupling in size since the start of the year, Linear reduced pull request wait times from over six minutes to just over five, while cutting runner time per test by approximately half.