“I do like Thai.”
You are on a trip in a town you visit in frequently. Quick with just your thumbs and your favorite smartphone:
- open Google, Yelp or whatever
- shift and sort
- find a restaurant near you
- shift and sort
- try to remember if you have ever been there
- find the menu
- sort through the menu
- try to figure out what you ordered
- figure out if they have a wine that pairs well with your order
- place the order by assembling the various parts in a shopping cart
- input your credit card details, address and contact details
- confirm it all
- wait, not quite knowing when everything will be delivered
Estimated time ~ 10-25 minutes
I have been researching empirically how the Voice First revolution will impact the fundamental way we interact with our technology and our machines. I have been doing this since the late 1980s. Over the decades and most especially in the last few years there is an ongoing debate from even some of the most informed observers that Voice First systems will never really be more than a smart speaker, egg timer or alarm clock. Some will concede that perhaps a Voice First system may offer limited use in Voice Commerce with re-ordering of commodity products, “Alexa, re-order paper towels”.
Many of these Voice First deniers focus on many invalid points of data along with using the past mobile revolution as a format of how the future will present itself. The folly in this is the future of tech is non linear and to most non obvious until we are ex post facto.
I have been working on systems that have proactivity, continuity and deep context. This element alone, with little improvement in intent extraction AI, will lift the utility of Voice First systems to new a level. Google agrees and they have been experimenting with simple context and continuity for the next generation of Voice First systems.
In the video below demonstrated by Nandini Stocker and James Giangola Voice First leaders at Google, we are introduced to Amanda. She is more closer to a Personal Assistant, as I specifically define it than the general purpose Voice First systems like Alexa, Google Home and to some degree Siri.
The flow works out to be like this:
- Human: Amanda.
- Amanda: Hi how can I help.
- Human: Help me find a restaurant.
- Amanda: Sure anything in particular.
- Human: I do like Thai!
- Amanda: Actually the last time you stayed in this area you ate at Thai Pepper, it’s on Montrose in Lakeview. Feel like going back?
- Human: Sure why not.
- Do you know what you’d like to order or do you need some help?
- Human: I need some help.
- Amanda: Okay let’s start with the main course you usually order chicken and you usually like it spicy they’ve got three dishes like that. I just sent them to your phone so which would you like?
- Human: I’ll go with the Thai Green Curry again.
- Amanda: Now you’ve got a choice of appetizers soups and salads, there too. Which would you like?
- Human: hmm
- Amanda: Which would you like or say skip?
- Human: I like the Spring Rolls.
- Amanda: Got it.
- Human: I’d also like a wine please wine.
- Amanda: Right since you always order white here’s a list of those. Can I recommend the Marlborough Sauvignon Blanc from New Zealand.
- Human: Actually can I see what the Creostore ownpace looks like…
- Amanda: There you go.
- Human: Yeah that’s the one. The one with the hand on the label. I’ll take the bottle.
- Amanda: Sorry did you want a bottle of the Creostore ownpace?
- Human: Yes please.
- Amanda: Anything else like say a desert?
- Human: Tempting but no.
- Amanda: Okay so I have the Green Curry Chicken in order of Spring Rolls and a bottle of the Creostore ownpace. Did I get that right?
- Human: Yep
- Amanda: The total comes to $93.12 put it on your usual card the one ending in 1367?
- Human: Yeah.
- Amanda: Include a fifteen percent tip?
- Human: Yes please.
- Amanda: Great that should be delivered in 20 minutes to a half hour from now. Did you need help with something else? If not I’ll check back in when the food’s on its way.
- Human: Nope.
- Amanda: Bye, talk to you soon then.
Note the flow of context and continuity. There is an easy interaction that is far more helpful than a website, Yelp page, or merchant app. The entire process was guided by the personal assistant in a way that is nearly identical to standard human to human interactions and thus there are all of the natural cooperative principles of communication.
In item #6 we can see that Amanda has presented a continuity based on a prior visit to the area. By item #10 Amanda is creating a personal context based on prior eating habits. The interaction of Wine selection is a wonderful example of how the user, once in the flow of the conversation felt at easy to interact with questions. The visual aid of situational, on demand image of the Wine bottle promoted a memory cue with the user. From item #26 to item #32 Amanda we see the order confirmed, paid for with a potential of a status update when it is nearing delivery.
Remember those 13 steps I presented for you above? Amanda does it all with no need for you to use your thumbs, no mechanical loads and very low cognitive loads, in about 2 minutes. Do this a few times, you will never want to use the “old” way of using a smart phone app.
Amanda has been demonstrated privately and in a limited public way by Nandini Stocker, a Voice First designer at Google. We can see in this video a classic example of how the flow of a typical face to face server/patron interaction would track. The fundamental difference is the context related to the patron. Amanda not only remembers that you have visited or tried the food at the Thai restaurant, she remembers the details of your prior order and even suggests a pairing of wine with the meal.
Make note this is not the experience of the phone tree IVR experience even some experts in AI have presented for food ordering. This is a simple yet robust dialogue that is informative and delightful. More closer to speaking with a friend than the rote Q and A we see in the first generation of Voice First devices.
I have stated that over 50% of quick service food ordering will be via Voice First AI.
— Brian Roemmele (@BrianRoemmele) April 14, 2017
In issue number 4 of Multiplex Magazine we will explore the basis of how Amanda-like systems will bring about a rapid and wide acceleration of Voice First systems. We surface huge opportunities for startups and legacy companies to lead or at the very least ride the wave. Many missed the computer, internet and mobile revolution, don’t miss the Voice First revolution—subscribe and stay miles ahead of the cutting edge.