While it's debatable if Apple would release something outright as encompassing and complete as OpenClaw, they should have helped developers and builders to build something similar themselves.

This could have come in any form, a platform as the author points out for instance.

I have a couple of ideas, how about a permissions kit? Something where before or during you sign off on permissions. Or how about locked down execution sandboxes specifically for agentic loops? Also - why is there not yet (or ever?) a model trained on their development code/forums/manuals/data?

Before OpenClaw, I could see the writing on the wall. The ai ecosystem is not congruent to Apple's walled garden. In many ways because they have turned their backs on those 'misfits' their early ad-copy praised.

This 'misfit' mentality is what I like so much about the OpenClaw community. It was visible from it's very beginning with the devil-may-care disregard for privacy and security.