We have a long-running and ongoing conversation about the future here at UBIO. We see rich opportunities for a technology that can connect and unite the distant corners of the web. This is the scope of our project and our business. This is obviously crucial and important to us, but it also coincides with the general technology picture of tomorrow. And that’ll be an exciting time which involves much more than us. It almost certainly involves you too. It’s about the future and reach of automation technology and its applications. Most notably of assistant technology.
Let me take you on a ride into the future of assistants, how it all might pan out and hang together.
Yes, Alexa, Siri, Google Assistant, Cortana and all the other speaker based or mobile voice tools are probably what you think of as an assistant. This is fair, but it doesn’t stop at voice and while these technologies will inevitably develop we think these are merely the thin end of the wedge.
In 2020 Google announced a project called Duplex on the Web. It’s an extension of their core Duplex project, some fascinating R&D work in AI and telephony. Duplex is capable of holding a live, machine to human telephone conversation to perform tasks like appointment booking or event scheduling. It’s well worth checking out the recorded examples if you haven’t seen them. Can you tell that it’s a computer speaking?
Duplex on the Web extends the project and brings a general, assistive layer to everyday mobile web browsing. A layer on top of the browser can offset all the frustration and clumsy tapping when you perform chores like online booking. Duplex on the Web takes over for some of the session. As a reluctant soft-keyboard typist myself I’d certainly welcome some of this assistance inside Chrome when it’s released.
Incidentally, UBIO technology can already power something similar to Duplex on the Web, but hold that thought for a moment, I want to come on to where we fit in a little later.
Photo by Josh Hild from Pexels
Assistant technology is currently emerging. So where might they be heading in the not-so-distant future?
We expect there to be a ubiquitous, always on, always there set of services that ambiently help you with whatever task you are undertaking. And the interface used to access this assistant need not just be your phone or a particular hardware device. It can inhabit any interface.
It might start on your phone. Your calendar might make sure the alarm is set for a trip you’re making. Knowing you’re bound to be a little fragile this early your entertainment system will slowly raise the volume on some gentle music or a feed of news. You get in the car and its system takes over, guiding you to the airport and booking your parking. You have just enough time before boarding that you can spend some time in duty free. The airport’s POS system identifies a great price on the fragrance you wear. The screens above the headphones you’re looking at remind you of that. On the plane your media preferences are synced. And while you’re collecting your bags a taxi is ordered for you. While you’re there, you can quickly book the excursion by adding a likely date in your calendar. The hotel TV keeps you up to date with the box set you’re watching. And all of this was booked in the first place behind the scenes, with all your family’s details known and input on your behalf with the travel company.
The assistive technology of the future will be a thin layer across diverse systems and devices, all there for you. Based on your rich profile it’ll know what you like and want, and it’ll be ready to spring into life on whichever interface happens to be closest. It’ll be your ever present helping hand.
This all makes sense, I’m sure. Assistants on top of every little chore and admin detail so you get on with your life. Profiles, calendars and services all working together to sort all this out for you.
Conceptually this makes perfect sense, but as a prediction this is where we run into a little trouble. Big companies and a handful of developers can actively create the skills or support required for an assistant to provide a particular service. That’s great. But you’ll only have partial coverage of all possible services. For everything else it’s a different matter entirely.
Services are all channeled via the web. We buy, book, transact and connect on the web. Though they’re served to us by machines, web pages were written for humans. In general, and unless there’s an API available, machines can’t make sense of them. And those APIs are few and far between and, despite some efforts from champions of the ‘semantic web,’ there remains no standard way for machines to understand the content of websites.
Sadly most of the web will be dark out there for tomorrow’s assistants.
To have meaningful coverage assistants need to work around this service limitation. There are probably some options here.
1. Wait for every site, shop and booking engine to publish an API or re-deploy their services so machines can interpret or act upon them. We think this is practically impossible. At best such services would take an awfully long time to come online. Ubiquity of service is pretty unlikely.
2. Develop an AI to machine interpret and interact with any website. This is the Duplex on the Web approach, and it’s good. It might work well for situations where partial interaction is required. But an AI that could handle any old website would be a huge challenge. Websites are a wilderness of treacherous markup and varying quality. UIs are hugely divergent and technologies myriad. And for cases where no interaction is possible — say in the car, or the airport POS in the example I made earlier — it might all fall down.
3. Deploy intelligent robots. If machines can be trained to interact with any website, just as a human can then there need be no underlying code or cultural changes to the way the web currently works. And here the AI approach makes more sense, since it could provide a hybrid approach to augment the robots’ powers.
You’ve probably guessed already that our money is on Option 3. Let me tell you about the capabilities of UBIO’s web automation platform, Automation Cloud.
Photo by Naveen Annam from Pexels
We have one of the world’s most capable web automation platform. Our technology can interact with any website just as a human can. The Automation Cloud effectively turns any particular website, or segment of the web landscape, into an API. With it machines can connect, interact and transact with anything.
This means it becomes possible to incorporate any and all surfaces of the web into the assistant layer of the future. It might be Siri, Alexa or the next technology, though they won’t be limited to a few big box skills. They’ll be able to interact on your behalf with the rich and varied landscape of web services you labour over on your phones and laptops today.
Above I described a seamless, ambient layer that intervenes on your behalf and pops up wherever you need it. There’s still a lot of prediction in that vision. Its execution would require a lot of engineering and reach. But the connectivity question, as we comprehend it, is solved. If you can interact with it manually today, then you can automate it, also today.
You might be yelling commands at your phone, or at Alexa on the bookshelf. Maybe you have an alpha version of the new Pixel phone and Duplex booked your holiday rental car before you scrolled through this article. But it won’t stop there. In our view of the future there’s a pre-emptive, ever present and personal assistant coming. It’ll be wherever you are. And it’ll be able to do anything in the wild west of the web, where all the services are.