While working on a notification delivery system, we came up with different versions of the services. The services are broken up into several microservices, and most of the time we need to work on optimizing them (rewrite in other languages etc.). But a lot of time is wasted carrying out performance testing to identify which version is better (higher request rate, lower latency etc.). The idea is to take the concepts of genetic programming to discover the strongest candidate.
I wanted to design an architecture called Perfect Architecture before - only to realize there is no such thing as perfect. It is always it depends-architecture and should be designed based on the use cases. But evolutionary architecture is something that I would like to advocate - it mimics the human process and has a brain (machine learning/deep learning) and has a pipeline (whatever organs in the human body has), has self-healing and functions as a whole.