[beenthere]

Since warm air rises. Even if the stats knew the temps in the other areas. The lower units, would be the ones to do most of the heating. Software integration will have a hard time beating physics.

 

[beenthere]

Yes, there is no denying this logic but then 20 years ago we didn't think self-driving cars would be real!

Here's my thoughts. This is so simple that I am going to describe that when applied to say, AI or machine learning, it isn't much more complex.

Program in your air con systems (all of them). Add in sensors about temperature, time to temperature, floor it's on, live airflow, seer etc., then you program you preferred temp, maybe per "system" or maybe whole house. Then add in room sizes (maybe including heights). The rules (or AI) simply learn what I do manually already knowing how science works but I am not experienced in the systems, and how they change based upon age, freon, outside temp etc.,. I know the lowest cost system to run (Carrier), takes x minutes to heat or cool. Same approx for other systems. Add it all together and a computer could compute the "simple" and get smarter over time.

And after all that. Warm air will still rise to the higher level, and cool air will sink to the lower level. Making the lower units do the most work heating in the winter. And as always, the unit in the higher level will still be doing the most cooling. Don't need AI to control it. Most people learned a long time ago, to set their thermostats appropriately for each season.

Highest efficiency A/C for the higher level, highest efficiency furnace on the lower level.