I’m curious to know why. This is first I’m hearing of it but it doesn’t look too bad, is it just that it’s less mature than other workflow automations?

Having used it pretty heavily in a containerized desktop environment, I wouldn't use it in production for anything business critical. It's a mix of "there's too much unknown" under the hood, and "there's weird things that I can't reproduce". It hadn't really reached critical mass as of yet, so support from the community is lackluster.

I'd sooner pay a bucket for something like Tines to offload critical mechanisms, and then do the dirt-cheap stuff on n8n.