I think in terms of dx they are going the route of watching file changes and deploying it to their infrastructure. It says “sub-5-second code synchronization”.
We had looked at this approach before creating SST and felt it would be too slow. If you are not blocking requests, then you have to wait for the deployment before invoking your functions. We wanted to make sure that after a code change you could simply invoke the function and be sure that the latest version was executing.
On the architecture point, just echoing what @thdxr is saying. We want something completely open source and self-hosted in your AWS account (for the reasons he outlined). With a lot of the recent Serverless Inc products (Components and now the Cloud), it’s not clear to me what’s running on our end vs theirs, what’s open source and what’s not.
That said, I’d love to try it out and check out the benefits to their way of doing things.