[detaro]

I suspect shorter delivery loops killed it. Manual QA is alive and well in projects with long release/delivery times, like embedded products. Also things like partial rollouts aren't possible there, but can be done if you can update deployed software very quickly.

[hitsurume]

As a former QA Eng, this is pretty spot on. We use to have a lot more lead time in doing manual testing and catching things during feature testing and then adding to our regression suites. Nowadays the mindset of continuous delivery is to just fail forward and rollback or feature flag something off if it doesn't work. There is still manual testing in teams that are more waterfall and probably dealing with more impactful issues that make manual testing a priority, like payments.

[UK-Al05]

I work in finance and payments and manual testing is only confined to exploratory testing which we only do occasionally.