One of my favorite conferences of the year is this week’s Oracle Openworld, mainly because Oracle is a company that competes in a variety of business application categories. And they do have apps for the higher end of the SMB space. Unfortunately I was unable to attend this year, but very fortunate to speak with Suhas Uliyar, Oracle’s VP of Bots, AI and Mobile Strategy.
Conversational Interface Technologies
With artificial intelligence, bots and voice-first devices being three of the hottest topics today, I was glad Suhas took some time to share his thoughts on how customer engagement is being shaped in real time by these technologies. Below is an edited transcript of our conversation. To hear the full interview, please click on the embedded player below.
* * * * *
Suhas Uliyar: I’ve been in the mobile industry for the last 21 years and every single year has been an excellent and interesting year. Mobile continues to change every year, but I’ve been with Oracle now for four years. Prior to Oracle, I was lead VP at SAP for their Internet of Things and prior to that, I was with Motorola Solutions as the Chief Solutions Architect for Enterprise Applications.
So I’ve been in this industry for 20 plus years, and it’s been a really exciting time the last decade plus. Looking at evolution of mobile and now Tech Bots and conversational AI.
Small Business Trends: I had a conversation a little over a year ago with the CTO and co-founder of Hubspot. He said something that really fascinated me. He thought bots and conversational interfaces may be the most important technology in the last couple of decades. Since you’ve been around and been involved in this, is it that important? Is what we’re seeing with bots and voice-first devices … Is it at that level of importance in your mind?
Suhas Uliyar: Absolutely, Brent. I describe it as the next coming of the browser. In fact, chatbots and AI are going to do to all our applications, and the way we interact with data, what browsers did to clients or apps if you can remember that transformation.
It is going to be the next, I would say, the next decade is all about a conversational AI-first approach that is a much higher magnitude than what we saw from mobile-first approach.
Small Business Trends: So we’re hearing a lot of talk about AI and rightfully so. There’s so much going on in the area. But what impact does conversational interfaces have on AI?
Suhas Uliyar: So AI is a broad subject, Brent. I think there are several flavors of AI, and the way we tend to categorize AI is this five buckets. First, it’s all about machine learning, deep learning technologies. One particular, natural language understanding is absolutely critical and vital to the success of chatbots.
The other categories are equally important. The next one is dialogue and context. The third is knowledge services. The fourth is data and insight. One I’m missing … cognitive services around speech and image recognition, so on and so forth.
Now, why is natural language understanding very important? What is really driving the adoption of this tremendous momentum we’re seeing in conversations interface. It is a fact that we can converse in a language that is most natural to us and the one we are most comfortable in. And when you look at things like Facebook Messenger or ReachOut, WhatsApp, Telegram, Skype- all those different chat-based messaging. The other aspect to communication is that we expect and get near real-time instantaneous response from the other side. Today, a lot of that was human to human, but the two aspects of a natural conversation and getting instantaneous responses has made that channel very attractive compared to using email or any other communication channel where it is “fire and forget” and wait if the other person responds or not.
In order for that to scale for an enterprise and be able to use chatbots to really enable that adoption, you need something, a computer program i.e., a chatbot to handle that scale. And to handle that scale means to be able to process that natural language that’s coming in from the end user and be able to then extract relevant information from that communication, and then integrate to all the systems of record and be able to get information back. Whether that is, for example, a financial institution where you’re trying to check the balance of your accounts or your airline where you find out if there’s an aisle seat available for you to reserve, etc., etc.
In order to facilitate this, natural language understanding, deep learning, machine learning algorithms are very critical whether that is neural network based or spectrum based. So, without AI and, specifically, natural language understanding, chatbots, in my opinion, would not even take off. That’s number one.
The second aspect is what I mentioned as dialogue and context. Which is the ability to divine a conversation with the end user. You want to make sure that the bot is not human, but, at the same time, you want to have a decent conversation with an end user. And we all know, as humans, we don’t think quite linearly. We tend to go off on different tangents and based on a certain questions.
For example, let’s say that I was looking to transfer money from one account to another or I want to pay, send money to somebody. I say, “Send money to Mom.”, for example. And the bot responds with, “Yeah, that’s great. Which account you want money to transfer?” And I say, “checking account.” And it says, “How much do you want to send Mom?” And suddenly I realize actually I’m not sure I have enough money in my checking account. I’m just going to ask the bot a question which is, “What’s my balance?”. And I don’t want to give any other context than saying “What’s my balance?” And the bot needs to be smart, and the conversational context designer has to be smart enough to recognize that you’re switching context now, and you’re going down a different path asking about balances, perhaps follow-up questions could be around, what are the transactions, what are the deposits or checks that have cleared. And then you can come back to the original transaction that you had. So this is what I mean by being non-linear.
And the dialogue and the context also, from a machine-learning perspective, has to be smart enough to understand that, and work in conjunction with your natural language understanding to be able to process that.
One final comment I’ll make about this – I mentioned the word ‘extraction’. When I, as an end user, get information like what were my last five deposits in my checking account, the machine learning should be smart enough to understand the ‘last five’, and I don’t need to quote all that. The machine learning algorithm is smart enough to extract all that information, extract that information like my checking account, knows who I am, what I’m asking for, and construct the integration layer to go get all the information from the back end.
So, long answer, but AI is absolutely critical to the success to chatbots.
Small Business Trends: Great, and there has been so much diving in and companies are really starting their conversational interfaces, looking at chatbots and looking at how you can integrate into things like Facebook Messenger because that’s where all the people are and that’s where a lot conversations are taking place.
But where do you see the voice-first devices starting to come in? Is this something that’s really too far out for a lot of companies to worry about or is it coming faster than some folks think it is?
Suhas Uliyar: I would say it is coming faster … You see from Alexa, but I’m not sure if you’ve picked up Google Home now has the capability to have multi-user authentication on voice. So, for example, if my daughter says, “Call Mom.”, it’ll call my wife, as opposed to calling my mom. And if I was to say, “Call Mom.”, it would say, “You don’t have a mother.”
So I think that sophistication is already there, and there is a lot of enterprise security features being added to it, so I would say voice is coming fast.
But before we get into these, what I call, VPAs like Alexa or Amazon Dot Echo, Google Home, Apple Pod. Brent, voice-first has been there through Siri, and Cortana and Google Voice. What I would say is, particularly in the enterprise world, especially in the CRM and service and so on so forth, a lot of customers want to be able to extend their existing mobile app or their existing web interface to include voice in it.
So for example, if you are using Internet Explorer or Chrome or your favorite browser, to be able to extend that on your MacBook or on your Surface devices and be able to use either Siri or Cortana to interface with the chatbot. I don’t think you have to wait until everyone embraces the Amazon Dot, Echo, Google Home. We’ve been seeing it for the last year plus where mobile devices have already been interacting using Siri, Cortana, Google Voice, etc. And so that’s been a part of our adoption that has happened over the last year. Where there’s mobile devices that had mobile apps, have now extensions to chat applications within the mobile app or within the browser itself. Within that itself, you could use those voice channels.
So those are already being adopted, and we are seeing a very natural adoption of these voice channels because people, again, like to converse and use voice. The only thing that’s, I would say, inhibiting the growth or adoption of devices like Dot Echo, Google Home in the enterprise is the security layer of it, which, I believe, is getting fixed very quickly. Today you can use Alexa, for example, for a home bank or integration with your bank, which is probably the most secure thing that all are concerned about. So, I see this progression very fast and very quickly we will see this level.
Now, the only thing that stops voice being adopted, in general, is the privacy, not necessarily to do the authentication, authorization of it, but what people can listen into. So, certainly, you’re not going to walk in your house with a bunch of guests and ask your bot, Alexa or Google Home, what’s your balance. You want to do it in privacy. The problem with that is if you’re in a public environment, no matter where you are, the response back in voice could be heard by most of the people. So I think it just comes down to the use case the environments where you are.
That’s why one of the areas that we’ve heavily focused on is gathering insights into the adoption of these channels. It’s not as simple as saying, “People don’t want to use voice versus chat versus that.” It’s also being able to understand what are the specific use cases where certain channels are better than the others. Are there certain environments, are there certain times where these channels are being used or adopted more than the other channels?
Small Business Trends: One of the folks over at Amazon said, “Now is the time to start designing experiences for the ear.” Does your customer base need to morph the way they traditionally do things in order to take full advantage of delivering experiences that customers are starting to expect with these interfaces? What are some of the biggest challenges facing companies when looking to implement these strategies?
Suhas Uliyar: Brent, totally. About a decade ago, the digital agencies were all transforming themselves to create delightful mobile app experiences. If you asked me this question 14 years ago, I would have said, “Yes, enterprises need to really look at bringing on designers, UI/UX designers for your mobile apps.” And now we are in the exact same stage a decade later for the conversational designer. And we are beginning to see certain companies bring on folks or start with research capabilities. At Oracle, we have a conversational designer team that helps our customers to really understand how do you best configure this so it does provide the delightful experience to the end user.
So the risks actually are in exactly that mode, which is having the right design to have the right experiences. And also understand the channel. As an example, if you’re designing for Facebook Messenger, a more visual channel is a little bit easier to handle so that when you send information back and send pictures or videos back, it’s easier to consume in Messenger, Line, Skype, etc. But you’re not going to see that in Alexa or Google Home or other types of channels. And so you have to really understand what channels your customers are coming in and have the proper design.
Small Business Trends: If we peer out a year or two from now, where are we going to be with conversational interface? Where are we going to be with the voice-first interfaces, do you think?
Suhas Uliyar: I believe we are going to see a convergence of immersive technologies and conversational interfaces. So things like augmented reality, virtual reality… One of the challenges of virtual reality today is that the type of the devices restrict what you can and cannot do. But I see that opening up. Google Glasses was an experiment, which showed it could potentially work- it needed some work. But, imagine a day where you don’t necessarily need to look at your mobile device variables in themselves are conversational-led. So now, you could be very – in a very natural mode- be able to look at the world in a three dimensional way and be able to discern information you need using conversational interfaces.
There’s a lot beyond that. In just in the world of AI, for example, image recognition, sentiment analysis, Q and A analysis. We’ve got knowledge graphs, so I am able to … CRM, for example, imagine the data the salesperson asks the CRM database, “Hey, it’s a new quarter. Tell me the 10 sales opportunities that helps me make my quota.” And the back-end, the bot, and the AI engine not only takes the leads that they have, but they are also able to marry that with information from other sources, whether it’s LinkedIn or Reuters or other things that calculates the propensity of the customer to buy your product and highlights that to you based on the question you asked. And that information could be sent to you in whichever way you want. The devices could be like, for example, HoloLens that Microsoft is working on. So I would say we are entering into the next few years where devices are going to take a very interesting shape through 3D virtual reality, but the interaction with that is all going to be voice-based.
This is part of the One-on-One Interview series with thought leaders. The transcript has been edited for publication. If it’s an audio or video interview, click on the embedded player above, or subscribe via iTunes or via Stitcher.