Not GP, but one of my previous jobs used to have a deploy method of "ping one of the ops team on Slack to release a certain commit SHA to prod/staging/etc", which might easily take an hour or more if they were busy and/or you asked around lunchtime. Even if they were available immediately it would still take 20 minutes.

After we moved to k8s, the release model went to "every dev can use kubectl to deploy" and it took a minute or ten to release something to prod. Needless to say, we all loved it much more than the previous setup. Even more so after someone whipped up a script to automate all the kubectl invocations.

One of my former colleagues wrote an article about it here: https://wetransfer.com/engineering/migrating-frontend-to-kub...