The problem with voice assistants at the moment is the same as command line tools.
You have to know what specific prompts work and don’t work etc. I often see my parents make simple queries like “turn off the white lights and turn on the night lights” only to get the response that it can only do one thing at a time.
There will be a future where this gets more and more sophisticated. But in the meantime, it’s not surprising it’s losing money.
A customer of mine has a product based on DragonNaturallySpeaking that medical doctors use for dictation. They are doing a brisk business. It's all software.
No fancy hardware. For this use case, it's almost flawless. And way faster than typing. The only commands you give are related to the recording -- stop/start/playback-type commands. Not sure how sophisticated it gets beyond that. But the speech-to-text technology is pretty damn good now.
At this point in the SOTA, Whisper [0], can probably be a drop in replacement for the for-profit Nuance versions of Dragon, etc. It's even open source.
Even if it were more sophisticated, I don't really see how it makes money. Even if it could absolutely nail that slightly more complicated command, and manage your home automation with natural language processing... that doesn't make money either!
I agree completely with your assessment, that they behave more like a verbal CLI than a proper assistant, but solving that well doesn't seem like it offers a great opportunity for monetization.
My hope is that the open source voice assistants get more popular, and it gets easier to flash the Amazon Echo to be a satellite device for the open source voice assistant. I've looked at running the open source voice assistants locally, but the biggest barrier I had was the cost for each of the microphone/speakers. Amazon's really got the cost of those things way way down.
I could see it as a replacement for UberEats if it could get to a high level of convenience. E.g.
"Alexa, order two medium cheese pizzas from the closest Domino's for delivery."
"Okay, ordering two medium cheese pizzas from a Domino's 2.7 miles away for delivery as soon as possible, for $25. Is that correct?"
"Yes."
"Okay, ordering pizzas. It'll take around 40 minutes."
But you'd need a very high level of AI intelligence/integration to get to that level. As it stands you'd probably go back and forth a bunch of times as the AI quizzes you on exactly what you want, when you want it, from which restaurant, etc. And that would just kill the usability, to the point where it's better to just use your phone.
Yea, I think there is a revenue stream there. I just question how big it is. First of all, you need to take a pretty-small slice of the transaction since they also need to make and deliver the pizza. If I know my Alexa order is always $1 more than calling it in myself... I'll probably just call.
And then, there's a weird thing where this probably only works for the simplest (and then cheapest) transactions. If I'm placing a complex order because I've got 7 friends over, I'd rather use an online ordering system or just make a phone call.
So you're taking a small cut, and probably only getting in on the smaller orders. There'd need to be a lot of volume to make it work.
HomePods aren’t sold for a loss (Apple doesn’t do that, plus price makes it seem even less likely). Macs/iPhones/iPads/etc also pay for its development.
But it doesn’t exist to make money like Alexa (why would I buy random stuff with it?). It exists to make Apple devices more sticky. And it works well for what I use it for.
The Alexa business model never makes sense to me, and that’s even if the devices were never sold at a loss, which they frequently were.
fwiw google understands "and" but you have to use full sentences on either side of the and because it processes them as independent queries.
so what you wrote ought to work but "turn off office and living room lights" wouldn't
The problem with voice assistants at the moment is the same as command line tools.
You have to know what specific prompts work and don’t work etc. I often see my parents make simple queries like “turn off the white lights and turn on the night lights” only to get the response that it can only do one thing at a time.
There will be a future where this gets more and more sophisticated. But in the meantime, it’s not surprising it’s losing money.