The common assumption, and why it is wrong
Many organisations only automate a process once it runs constantly. We see it differently, for a concrete reason.
A procedure that happens twice a year is the most error-prone one there is. It needs knowledge nobody has to hand, it gets looked up again every time, and it gets done differently every time. That is where automation returns most: not in time saved, but in repeatability.
What it covers
Provisioning virtual machines and containers, the processes around operating them, and connections to the usual public clouds: Azure, AWS, Google Cloud, IBM Cloud.
That gives one place to manage what otherwise lives in several consoles, keeps cost in view, and turns the question of where a workload runs most cheaply into something you can answer rather than estimate.
What to watch
Automation encodes the processes you have. Where no rule exists it makes the absence visible. That is the uncomfortable part of a rollout and also the most useful: the decisions belong to the operating model, not to the tool.