A few simple use cases:

Remote heating control. Go on holiday, leave the heating in frost-protection maintenance mode in the winter, turn it on again at the airport. Come back to a warm house, but plenty of energy saved.

Sunrise/sunset lighting. I have the lights in my office set up to turn on around dusk and turn off around 3am if I haven't already gone to bed. This works so well I hardly ever think about it now. It also means the house looks occupied when I'm away.

Motion-sensing light switches. I experimented with these, but it's a harder problem than it seems. I set up a PIR switch in the kitchen, and being able to walk in/out while carrying plates/food/things without having to hit the lights manually is amazingly useful.

Problem is PIR sensors have a limited range and are easily confused by various domestic heat sources. I was going to experiment with multiple sensors and more reliable software, but got distracted by a house move. So that hasn't happened - yet.

I still think user tracking combined with light level sensors would be amazingly useful. Switches on walls are probably going to be redundant when automation really arrives.

Colours and such are much less interesting than having smart lights that work so well you only have to think about them when you want them to do something unusual.

Having said that, I have mixed feelings about voice control. I have Alexa controlling some of the lights, and I don't find it very natural. This may be an unusual view, but I think speech is a channel of emotional communication, and tone carries a lot of information.

It's second nature when talking to humans, but talking to machines falls into some kind of social uncanny valley. It's impossible to do it without being reminded that you usually think about tone while talking, but now you don't have to - so of course you still do, but think you shouldn't, even though you can't stop.

That's a significant time, cost, and effort you're spending on what amounts to micro-optimizations though. Using a standard light switch is maybe half a second of effort. (Not an improvement over getting your light-control ap out.) It only makes some sense if you live in a huge space where you have to walk for 5 seconds to get to the switch or your heating takes 2 hours to warm a room.

Color me cynical, but it seems like a big hassle to solve a non-problem.

Those are all good use cases.

But the tech to implement is costly, fragile, cumbersome.

'Not worth it yet' for me, and frankly, I don't see it happening for a while yet.

> I think speech is a channel of emotional communication, and tone carries a lot of information.

"Aziz, light!"