We built a system that investigates production incidents automatically

The First 30 mins of an incident is still manual. I usually end up jumping across Kubernetes, logs,...