Infrastructure & deployment
Agentic coding is straining CI. Here’s how we scaled test impact analysis at Anthropic
Anthropic reports a 25x rise in CI jobs over six months as code output and review accelerated—exposing a test-selection service that could not scale horizontally.
Why it matters: Stale test history can make test selection less reliable even when CI has run. The article recommends planning for 25x load within two quarters, instrumenting services so agents can detect problems, and checking that incoming CI jobs match outgoing jobs.