I think Kubernetes config is a perfect target for AI. Conceptually k8s is not that complex but setting up and managing all YAML files with tons of boilerplate is very hard if you don’t do it every day. An AI should be the perfect tool to analyze the files and make changes.

Busy using a fair bit of A.I. to produce yams manifests. It still hallucinates a fair bit and doesn’t connect the dots well between the various parts of you colour outside the lines even slightly

Works but feels substantially weaker than on coding tasks. Not sure whether that’s lack of training material or the lack of execution flow code has.

Hard no.

There is already people who uses k8s that does not understand how it works or even know how the app works inside k8s.

Adding AI to it is just asking for a disaster down the line.