For those deep on AWS: What works for resilience beyond the marketing diagrams?

I have been in an AWS only shop for years now. We have done the Well Architected reviews, drawn the multi availability zone and multi region pictures and talked about pilot light versus warm standby until the whiteboards wore out. Some of that held up, some did not. Resilience on AWS is supposed to be straightforward: spread across availability zones, add a second region, lean on managed services, drive everything through IaC, test with chaos. What I have