Booking Travel with Artificial Intelligence
My name’s Ben Waymark, I’m Technical Architect with Project X. Specifically, what that means is that I get to play with all the really new, fun toys.
One of these toys, for those of you who are unfamiliar with it, is the Amazon Echo. Now what the Amazon Echo is is a speaker. Lots of speakers on the market, not very exciting in itself, except that you can talk to it. So what happens is you ask it to play music for you, you ask it to read a book, you ask it for a recipe.
Now we’ve been discussing a lot about artificial intelligence and some of the opportunities that has risen from artificial intelligence. One of the big areas that artificial intelligence has been making an improvement in, is voice recognition. It’s done great things for chatbot like Christian has just talked about. It does great things for propensity engines and big data like John’s talked about, but it’s also improving our ability to process language and to understand it. As John said earlier where 10 years ago it was about 20% accurate, now it is over 90% accurate. And that gives us an opportunity to use voice as an interface. What that means is that we can now start having our mobile phones, we can have our tablets, we can have our laptops, we can have our desktops, but we can also have something we just speak to. And as an interface, this creates huge opportunities for marketing.
The reason why, is that with a voice interface, there is an automated way that a retailer can ask, “Do you want fries with that? Do you want a drink with that?” In a way that a popup message, a flashing advert, or a pop under message just won’t do. For airlines, this could be asking if a passenger wants lounge access, if they want fast track, if they want parking. For hotels, this could mean, do you want to have a bottle of Prosecco waiting for you when you arrive, do you want a special meal. For the car rental industry, this could be asking if you want a GPS, if you want a car seat, or if you want to fill up your tank with petrol. Now in this example, I’m specifically looking at the flight selling and holiday discovery, but it’s good to remember that there’s a lot of opportunities where we can use this new interface.
Just to set the scene a little bit, imagine that it’s early morning, brushing my teeth, or maybe it’s in the evening, I’ve got guitar jazz playing, I’m cooking up a nice meal. Suddenly, I remember that I need to book my holiday to my imaginary office in Barcelona. I do this every week, I forget it happens. I don’t want to stop cooking, I don’t want to stop listening to the music, so all I need to do, turn to my lovely speaker. Let’s see if this works, and say, oh, let’s check to make sure this is, there we go, and say Alexa, ask OpenJaw Airways to book me a flight to Barcelona.
Alexa: Would you like to leave from your home airport of London Gatwick, Mr. Waymark?
Ben Waymark: Yes please, Alexa.
Alexa: When would you like to leave?
Ben Waymark: I’d like to leave next Tuesday.
Alexa: When would you like to return?
Ben Waymark: I’d like to return next Thursday. Here it comes.
Alexa: You could fly on OJ 2708 leaving London Gatwick on 30th May at 1500H, then return on OJ 0487 leaving Barcelona on 1st June at 2035. This will cost 550 Euros for an economy seat.
Ben Waymark: Okay, this is the good bit.
Alexa: Or for an extra 50 Euros you could get a premium fare with more legroom. Which fare would you like me to book for you?
Ben Waymark: Of course Alexa, book me the premium fare please.
Alexa: Okay. 600 Euro payment has been taken from your Visa card ending in 8765 under the name Mr. Ben Waymark. Your booking reference is SUWTE. Further details including conditions of carriage has been emailed to your registered address email@example.com. Would you like me to save the search for you?
Ben Waymark: Oh, yes please Alexa.
Alexa: You can search for a return trip from London to Barcelona leaving on Tuesday, returning on the following Thursday, by asking OpenJaw Airways to search for the saved trip to Barcelona.
Ben Waymark: So there you have it. I’ve managed to book a flight. I’ve managed to pay for the flight. I’ve managed to save the search, so next week it can make it even a quicker to, and a less painful way of booking a flight, it’s all there. The question is now, what have we done?
Well, we started off in the Alexa app, and we already have the passenger details and the credit card numbers. This can be linked to an Amazon account, or we could create our own app that would allow us to fill this in. We’ve also in the app identified the passenger’s local airport. Now this is important only because in this interface we want to be as frictionless as possible. It’s not somewhere where we can put in a lot of details, we want to make it as absolutely easy as possible, so if we can remember someone’s airport, we definitely want to do that.
Next what we’ve done is we’ve taken what really is a pretty dumb device, because all this can do is connect to wifi, and then it sends, it records your voice, and then it sends that as a voice file to a cloud API. Now the cloud API is a really clever bit where it determines what the key words are. This is where the speech recognition comes in, and really comes on its own. Amazon have a massive group of supercomputers, let’s call them, of its big cloud computing infrastructure, that will take the voice, and it will pull out the key bits of information. So superfluous words like the please and the thank yous gets knocked out, and the important words get put into a search string, which then will hit the t-Retail platform. t-Retail platform then does, as it would with a website, as it would with a mobile app, or any other interface, it does its search, it returns the results, sends them back into Amazon’s supercomputers, that then send it back to the Amazon Echo as a voice file.
For the Echo itself, it’s very simple, but for the big infrastructure there it’s a lot of work, but it’s a lot of work that can be done very quickly thanks to this massive infrastructure.
Now the nice thing is, once we’ve developed all this framework, we can start building on that. For example, I may find myself thinking, Alexa, where should I go on holiday?
Alexa: Are you taking Mrs. Waymark with you?
Ben Waymark: Yeah, I think I better. Actually, I’ll be in trouble if I don’t.
Alexa: Would you like a city break or a beach holiday?
Ben Waymark: I’m thinking beach holiday.
Alexa: Hotel OpenJaw has a promotional romantic trip for two to [Kofu 00:08:25] from London, staying in their executive spa suite and including two free spa treatments. Would you like me to email you the details?
Ben Waymark: Oh, yes please, Alexa.
Alexa: Those details have been sent.
Ben Waymark: Okay. That is the two main presentations, and those seems to have worked, so I’m feeling relaxed now, and I’ll tell you what we’ve done here.
We’ve now programed Alexa, it’s actually at the level of the supercomputers that we were looking at before, to assemble a set of information. In this case, specifically what we want to know, is we want to know what type of holiday it is, how many passengers are traveling, it’s myself and my wife, sorry, the holiday type, in this case we’ve decided to go on a beach holiday. And then what we do, and this is where it gets really clever, is we can send that search detail to the t-Retail cloud, which can then take your profile information, match that against the propensity engine that we were talking about earlier with the T data cloud, and it can find a very specific recommendation for you based on what it already knows about you. And then once that’s done, we can either book it as we did before right on the spot, or if it’s a more complex itinerary, if it’s something that needs something more than can be done on voice, we can either send an email or the Alexa comes with an app that also allows you to display extra details and links and that type of information.
The possibilities with this are endless. It’s good to think from a marketing point of view what we can do with this. It’s good to think about where we can go with this, and it’s also interesting to really think about all the stuff we’ve been talking to you today about our t-Retail cloud, about the T data, about the T marketing, for example there’s no reason why we couldn’t mix Christian’s chatbot that he was discussed there, so rather than interfacing on Facebook, we would be interfacing on the Alexa instead. There’s lots of opportunities here.