Right, redoing your entire workflow around a new corporate AI platform is signing up for a lot of throwaway work. If you like fiddling with things like that then great, but if the tooling is a means to an end and you're more interested in actually building stuff with it you're adding a lot of cost by chasing the latest fads.

Does anyone think that the brand new version of Antigravity will still exist in a recognizable form two years from now? Google will almost certainly have killed or "upgraded" it again to a new platform by then.