One good way I've found is to use Home Assistant's integrations page.

Let's say you're looking for a smart doorbell. You'd go here: https://www.home-assistant.io/integrations/#doorbell

You can then go to each integration and on the right if it says "IoT Class: Local Push" you can be fairly certain it can be locally controlled without the cloud and that it reports state without the cloud. In this scenario, it'd be DoorBird. You can be fairly certain you can block internet access at your router for this device and it'll work just fine. Don't quote me on that for Doorbird though.

This can be a good first place to look, but don't rely on it.

Many integrations have their local access via reverse engineered APIs, and they often go away when they get popular enough.

I wrote the Samsung Family Hub smart fridge integration for home assistant and one day Samsung removed the local access in a fridge update and now there is no way to access it without their android or iOS app.

That should be illegal...

I wonder if you could return the fridge once they update claiming it stopped working for your purposes.

Ah yes I agree, I've had the most luck with the above method if I block network connectivity to prevent automatic updates. But thats not always possible / you have to give up many other features potentially.

Yes - would really like a good doorbell camera that I can host at home. Seems like the Amazon and Google are the best unfortunately.

Check out doorbird https://www.doorbird.com

Can confirm. The doorbird is great. I use it with home assistant

See my sibling comment!

Best bet is get it up & running using a local HA server and then firewall it off so that it can't phone home