Incident Summary — January 16, 2026

What happened: A small fraction of jobs failed during a routine deployment this morning. The issue was detected within one minute and fully resolved within two minutes.

Impact: 31 customers, 507 parse jobs

Root cause: We deployed a change to an internal API contract. While our end-to-end tests passed, our rolling deployment strategy meant there was a brief window where services were running mismatched versions, causing errors for a subset of in-flight jobs.

Resolution: We immediately reverted the change upon alerting.

Going forward: We're shipping additional automated checks today that will flag any changes to internal service contracts before merge, preventing this class of issue in the future.