Siri In A Box

I’ve been thinking lately about the inevitable Apple based AI assistant, which will obviously be some kind of Siri based box. This comes from having used Siri for a number of years and only recently having used Amazon Echo’s Alexa. Having set up HomeKit and Alexa and getting it to work with a number of systems at home such as Harmony Hub and Sonos and August involved a number of glue systems such as Yonomi and Homebridge and Home-Assistant and Haaska. None of these are simple systems to set up, but all seem like something that should already be in both the HomeKit and Alexa ecosystems. They are not connected because of feuding between Apple and Amazon, and while that conflict is natural given the early stages of home AIs and the aggressive competitive nature of these two companies, it is frustrating for home users that just want to use a bunch of cool products together.

What sort of things would we want from an Apple based home assistant? Taking the best features from the existing services, I imagine that it would look something like this:

  • Disembodied assistant – the benefit of this is that I can just talk to it without deciphering the visual feedback, which is where something like Alexa is a win. No screen but maybe some LEDs to quickly see some kind of visual feedback.
  • Compact array of speakers and a microphone- the Echo devices are an amazing always-on tool dedicated to listening for you and speaking back to you, another Alexa innovation.
  • No battery necessary – it is always plugged in. Maybe if surviving a power blip is important something small can be included to keep the logic running, but having a battery means a shorter lifespan. Like Apple TV, this is something that won’t be swapped out yearly. BTW, the Evo is a neat battery for Amazon Echo Dot.
  • Needs to be cheap – the Echo Dot is all I need in a device and is only $50. Amazing.
  • App and web status – the Alexa app lets you get a view of all the devices that the system knows about. Home-assistant gives a very nice web based interface for showing all your devices and their status. The Home app, although it needs more automation complexity, gives a good view of rooms and devices and automations.

The Apple product that currently comes closest in form and function to this description is the Apple Watch. Given a tear-down of the Apple Watch, you can ditch the battery, the compact form, the display, the taptic engine, the rugged shell, the ingress protection, gps capability, barometric pressure sensor, the health gathering (heart rate sensor and motion sensors) – you are just left with the S2 SiP (a very thin and tiny board) and a speaker. Technically, you could even just go with the S1P SiP since the onboard GPS is not needed. That also still leaves you with very good bluetooth and wifi in a very tiny package. While this wafer thin system is not a significant electronic component, it is the brains of the whole operation that  is capable of taking on the task of just being present to serve as a Siri proxy. Given the S2 SiP as a starting point, what would you then need to add to it to make it a real product? My list would be:

  • Improved bluetooth – the W1 chips in the AirPods have amazing range, I’d expect the same for interacting with many HomeKit devices
  • improved wifi – this device needs to be ever present and ready, the wifi in the Apple Watch may not be strong enough
  • Improved speakers and microphones – the Apple Watch speaker and microphone are great for personal use and the few dings and bings that it does, but I would not use that to properly interact from across the room
  • larger case – appearance is everything in these devices. If you sell a device that looks like a poker chip and expect to charge more than $25 for it, that could be a value perception problem. The size of the Apple TV is just right and would differentiate it from the Echo devices. There should be some LED type display that gives visual feedback that it heard you, but not something you need to decipher to figure out the response – what about a touchbar style display that just shows the Siri response wave.

Given that Apple has promoted Siri as the assistant that is always with you, it would be quite a turn for Apple to instead go with a stationary device, but it’s not the first time they have pivoted on a successful product. There are also some challenges with this.

  • privacy – what can Siri be asked on this device that someone else should not know? What is being missed if not locking it down by not having any interface other than voice? Does any data live on this at all? What is allowed to be shared? This may be where circles of friendship (circles of privacy) could come into play and all “open” contacts lists and calendars and communication are centralized in this device.
  • multiple users – will everyone in the family log into this? would you just have another iCloud account that joins Family Sharing as an AI member? Does everyone join this as a bluetooth device which will track them? What about guests (would be creepy to ask them to register)? This would provide presence. This would also require voice recognition as to who is issuing commands such as if you are sending a message – this is much easier with personal devices, but this is more of a device available to all.
  • storage – is this going to eat into the little itty bitty 5GB total that Apple currently doles out to iCloud users?
  • management interface – there should be some kind of interface to see what the device knows. I’m not asking for a full content graph diagram but even a static list in would be good. The voice interface is not enough for everything. The Alexa app provides some post-mortem of what the device heard which you provide feedback to rate how it’s doing as Amazon continues to develop the system.
  • Better Siri – the Siri in Apple Watch is somewhat watered down version of what is in the iPhone. Given the bigger volume available in a bigger shell, Siri could potentially be more powerful in this device than in the iPhone itself. A lot of the jokes people make about Siri apply to Alexa and any other assistants as well, none are perfect. If you turn your head momentarily while talking, they all hiccup in interpreting your wishes. As humans, we just reply “huh?” but with AIs we seem to expect perfection even from our imperfect input. The technology still needs to evolve.
  • HomeKit is currently in its infancy – Open HomeKit to more than just hardware manufacturers. All their apps just look like the Home app with a different skin and are no more capable. Why can’t HomeKit and Alexa just get along? Talk to non-HomeKit devices. Heck, talk to IFTTT and Workflow! Give us plugins. Some very capable apps like Yonomi and Stringify are waiting to trigger your HomeKit devices!

That sounds like a complete product to me, but Apple has a tendency to premium price its products, however to compete with a $50 full featured hockey puck, it cannot be more than $99. Even still, that price will be tough to stay at. Consider the AirPods which are compact and difficult to manufacture, are not a stand-alone product, and are still $159. At $99 I would not only buy one, but probably 3 or 4 of them to have them around the home for the family to use.