Rahim Rahemtulla: Hello and welcome to Silicon Valley Innovation Center. Here at SVIC, we promote discussion on technology and business innovation through our Executive Immersion Programs and our online events, just like this one that we have today. My name is Rahim Rahemtulla. I’m a SVIC Brand Ambassador. Today, I’m talking with Kunal Contractor about the role of conversational AI in digital transformation. Don’t forget, if you have a question for Kunal, I should say, you can send that to us in real time. Just do that through the comment section of whichever platform you’re listening us through.
So our guest today, let me tell you a little bit more about him. Kunal Contractor is Global Director at Avaamo, which is a deep-learning software company which specializes in conversational AI for enterprise. And Kunal joins us now, I hope. Kunal, are you there?
Kunal Contractor: Yes, thanks, Rahim. How are you?
Rahim Rahemtulla: Yes, thank you Kunal. I’m very well. Okay, you’re coming to us from Silicon Valley, I believe.
Kunal Contractor: I am, indeed.
Rahim Rahemtulla: Very good. And I want to ask you, so AI chatbots. That is a concept, I think, we’re all somewhat familiar with at this point. But looking at Avaamo’s website, reading about the company, I feel like what you do is perhaps a little bit different, or at least you describe it in slightly different terms because you call it AI-driven conversational computing. I see that phrase used quite a bit in relation to Avaamo. So perhaps, just to start, unpack that for us a little bit. Tell us a little bit more about it.
Kunal Contractor: Yeah, absolutely. It’s a great question. So 1.36 things like chatbots as, you know, Satya Nadella said back in 2016, as the CEO of Microsoft, “We’re going to fundamentally revolutionize how computing is experienced by everybody.” But chatbots are great to help fill out forms or, in a way, perform what we call the first generation of conversational interfaces. Things such as, “Play Taylor Swift on my Amazon Echo”. But to truly deliver what we would call the promise of conversational AI, fundamentally new technology had to be built to perform what we then call multi-turn conversation and execute judgment-intensive tasks, just like humans would.
So what we do, in a way, is deliver a second generation of conversational AI that is able to execute those rich multi-turn dynamic conversations, so it’s capable of actually handling a lot of queries and aspects like customer services, generating quotes in insurance, answering claims for inquiries within healthcare, so it’s a pretty rapidly evolving technology and it does promise to change the way that we work and how businesses would really interact with their customers or employees and then their stakeholders.
And our world is about five different classes, actually, of conversational agents, from those that are slightly more simple to slightly more complex. And they go from things like chatbots, as you mentioned, which are quite narrowly focused and they execute a simple workflow using natural language. Or you can have a virtual personal assistant, which is good for first, second, third-type of service and you usually see it in devices like Facebook Messenger. Then a virtual customer assistant is kind of similar and they act on behalf of the enterprise to really stimulate a conversation that can deliver information and take care of actions on behalf of that customer. The employee assistant, which is very similar in simplifying engagement within the organization, and, finally, things like Avaamo, which is more of a four stack platform from where we do offer the full tooling, machine learning, data-science automation and integrations into all the enterprise system. And that is really how we see one of the biggest differences between conversational computing and things like chatbots.
Rahim Rahemtulla: So I’m going to play the role here, Kunal, of, say, a slightly skeptical business executive. That’s what we do here at SVIC, we like to connect executives and legacy companies with startups, with innovative technology companies because we want them to connect and to jumpstart innovation at their own company. So I might be looking at Avaamo and I might think, “Well, this sounds pretty great. You’re basically saying to me I can save money and by sort of having less human customer support, I can use those customer support agents in more creative roles and get more out of them and help them to achieve their potential and yet, I can still serve my customers, answer their queries just as well as I always could.” That sounds fantastic.
Kunal Contractor: Yes. Absolutely. So, as a part of that, when users are speaking to our bots- like we actually found with one of our insurance companies – they prefer speaking to a bot because, unlike sometimes human sales guides, they don’t lie. So in this age of fake news, that must count for something. But also bots don’t judge you. I could bet that you’ve probably Googled something before that you probably would never dream of asking another human being because, again, Google’s not going to judge you.
So even though bots are and can in some ways replace the activities of human beings, they very much also work in a way to supplement the activities of those humans. So we’re working right now with one of the countries top wealth management divisions of one of the country’s largest banks and they use our bots for their financial advisers. So even though their experienced agents are almost 99% certain of providing a response to an answer for one of their customers, they still use our bots just to make sure on the off chance they get the that 1% incorrect. But how can they ask that question to a colleague? That’s kind of embarrassing for them. Somebody who’s been doing this job for 13 to 15 or 20 years can’t ask somebody who’s been there for 3 years who probably remembers more of the training, “Hey, was the tax-free rate up to 5500 or was it up to 6,000?”Because then they’re just going to look a bit silly. But what they’re able to do now is, using our bots, all of their agents regardless of experience are able to reduce the amount of time they’re on the phone with a customer by over 50% and they’ve reduced reiterated new applications by 70%. So that’s really driving a substantial difference within the organization, impacting that customer experience and using the bots to supplement their existing workforce without necessarily replacing it.
Rahim Rahemtulla: That all sounds very good, indeed. But what if I have a workforce and I trust them and I know that they’re capable and I know that they know all about my business. But the idea is not going over to an AI platform to speak to my customers. Customers must, at some sense, prefer to speak to real people. They want people to pick up the phone at some point, don’t they?
Kunal Contractor: Some of them do and some of them don’t, so that kind of ties back into that insurance example where people actually prefer to speak to a bot because the humans would lie to them. Or, I won’t name names, but if you think of some of the country’s largest cable providers, you still go through some of their automated systems and spend about 10 minutes doing that. Then you get through to a human being and they ask you all of these questions all over again.
So, in today’s world as well, where people seek instant gratification, bots are able to answer user’s complex questions pretty much immediately instead of having to wait for ages to get an agent on the phone, and even when you’re on the line with an agent, some of them have about 50% of those calls as deadtime while the agent is looking up information.
Rahim Rahemtulla: Indeed.
Kunal Contractor: What our bots are actually able to do is predict what the user’s going to say next in the conversation, using our knowledge that we’ve learned from the deep-machine learning, the data science and the knowledge grasp. So it’s already pre-processing and calling the APIs and performing the transactions before the user even asks the question. So when they do, the user has an immediate response instead of even having to wait for about 400th of a second for the bot to then run those API integrations.
We’re actually doing something similar for the world’s largest appliance manufacturer right now, so we can actually even provide the user with information that’s relevant to them before they even ask that question.
Rahim Rahemtulla: And how do I know, though, that the bot, the AI platform is giving the right answers? How does it work in terms of oversight on these bots and what’s the process for them? When we change something in our company, how do we communicate that to the bot and how do we know that it’s always really giving answers to people that they actually want?
Kunal Contractor: Sure, absolutely. So that comes through extensive levels of analytics and business intelligence, but effectively using our data science automation we’re able to train it. Let’s take a simple, say, service desk or customer support type example as you’re alluding to. So we’re able to use previously asked and answered questions that the company has gone through so they know, “These are all the questions that people are going to ask us and these are all the different ways they’re asking us these questions and here’s how we have successfully responded to this questions, which especially ties in with how the users then rated that piece of customer experience or given a high NPS score.” But for a lot of companies, they’re really driven by maintaining and improving their NPS and providing, what I at least like to call, customer delight. So we use their existing data sets to train the bots. The bot immediately, once it goes live, knows what intent people are looking for when they ask those questions and how exactly to answer that question.
But we do know we’re not going to solve world hunger in one full swoop. Our bots can and always will escalate to human being if needed, to a live agent needed through the phone or on live chat or if the user just wants to do and wants to speak to a live human being. Our goal is not to force everybody to always use the bot. The goal is to drive customer experience and help the organization reduce cost at the same time. So if there are still some users who want to have a conversation on the phone, that’s perfectly acceptable.
And, actually, one of our latest products is what we call “conversational IVR”, where we’re replacing some of those large complex tri-based IVR systems, so users can actually have a phone-based conversation with the bot, including things like what we call human prosody and human discourse, which is the ability for the bot to add in things like how we’re having a conversation such as the role of pauses, a laugh, a giggle, literally even saying “um” while it’s delivering a response instead of it just sounding like a full automated system. So we’re trying to address that aspect as well.
Rahim Rahemtulla: It sounds like it’s really very, very sophisticated, indeed. You wonder, is this going towards a place where one day we’re not actually going to know if we’re talking to a bot or a human?
Kunal Contractor: Yeah, absolutely. So we actually get that sometimes right now, especially in the text and some of our voice customers, where the customer may not realize if they’re speaking to a bot or a human. And, in a lot of cases, depending on the client and why their customers are coming in or why their employers are working with them or with the bot, in a way – excuse my language – but quite frankly, they don’t really care as long as they can get their issue resolved as soon as possible.
Rahim Rahemtulla: Sure.
Kunal Contractor: So there are some cases where we work with certain healthcare providers or an online therapy site, actually, where for certain situations you do want to speak to a certified therapist but the bot is still there to help collect a bunch of information. So it saves human valuable resources, especially in people like therapists or medical professionals. So the bot can handle, let’s say, even if it’s up to 60% of collecting information, so that allows humans to spend more time actually providing better healthcare to patients instead of wasting time collecting information that can be automatized.
Rahim Rahemtulla: Absolutely. And I want to ask you, Kunal, what about digital transformation in all of this, inside corporations? When they’re trying to change their business processes and make use of data, better use of technology but to make the whole organization itself more efficient and empower it, really, to innovate more actively. And I’m thinking here how, particularly, I know that you were at SVIC fairly recently and you talked to an oil & gas major company who came in and joined us for an Executive Immersion Program and in that presentation that you gave to them, you didn’t talk so much about using Avaamo to talk to customers, but more about how using AI within a business can really transform how the business is ran. And so it sounds like it is – well, this would be for you to tell me – using AI to see where and how processes work, being able to collect data on those business processes, how employees interact with them, and then using that to analyze that information and being able to iterate on it.
Kunal Contractor: Yeah, absolutely. Great memory from that presentation. So this is a part of that digital transformation. It’s also in a way – we’ve a number clients who are calling it – what’s known as “industry 4.0”. That’s the name given to the current trend of automation and data exchange, including things like cyber-physical systems, IoT, cloud computing and especially what we’re doing as cognitive computing. But as we’re talking about addressing the internal employee experience, being a fellow Brit I’m sure you’ll appreciate the quote from Richard Branson, the founder and CEO of the Virgin Group. He said, “If you look after your stuff, they’ll look after your customers. It’s that simple.” So the core of a great customer experience is a greater employee experience. And a lot of organizations are failing at actually offering a proactive and innovative service that can really help with employee productivity. There are a huge number of factors that shape that employee experience, including the design and collaboration capabilities within their work environment, the tools and platforms they use to accomplish their work, but also that overall unique system that enables a productive workforce.
So we’re working with one of the largest tech companies – actually, quite a few of the largest tech companies – here in Silicon Valley, one of the world’s largest airline manufacturers, one of the world’s largest FMCGs, electronics company and some of the world’s top healthcare organizations, especially in the US. Some of them, actually, have been Fortune 10 companies. All of them are working with us for a simple IT helpdesk bot, and this is really powerful in terms of investing in that level of an employee self-service. Especially if you look at something so simple like a password reset. One of our customers gets 15,000 passwords reset tickets coming in every single month and it takes the employee around 20 minutes to get their password reset. So if you think about that and do the math as to how much time and money and frustration goes into that loss of productivity and we’re able to actually to resolve that and reset that password in under 27 seconds, that’s driving a major impact to the organization.
Rahim Rahemtulla: Yeah, obviously it’s a huge time and productivity savings to be made there. So, Kunal, is this a technology for any company, any industry? Do you guys target yourselves to some particular sectors?
Kunal Contractor: We do. We’ve got a particular focus on a number of core industries that range from banking, financial services, insurance, healthcare, telco and retail, but we do have a number of other customers in a number of other industries that range from FMCGs, manufacturing, supply chain, logistics, etc. as well.
Rahim Rahemtulla: So potentially there are applications of this across different businesses. It’s not just that a particular type business is particularly well-suited to the way that AI works and AI is able to deliver solutions in particular domains.
Kunal Contractor: It is absolutely applicable to practically every organization of any size in any industry. We tend to focus on slightly larger enterprises and not as much the smaller companies. But if you recall, quite a few years ago, we were in the world of companies having to adopt a digital restructuring; if you didn’t have a website, no one’s really going to do business with you. Then we moved into a mobile-first strategy; if you didn’t really have a solid mobile app, you weren’t going to get much done. Today, we’re in the world of an AI-first strategy. That’s becoming ever so important with analysts houses like Gartner themselves predicting that by 2020, which is literally just around the corner, nearly 80% of interactions people would have with a brand is going to be automated. So with that in mind, it doesn’t matter how big or how small the organization is. I’m working through some of our other partners with some startup organizations that are building apps to help pregnant women being able to ask and answer questions to reduce the amount of time that they have to go in and see a physician. So this is a very small startup company that’s just raising seed money, but I’m working with them because it’s a wonderful use case to, as I mentioned earlier, some of the Fortune 10 companies. So a great deal is happening.
Rahim Rahemtulla: Yeah, that’s very interesting, Kunal. It definitely sounds like AI really does have a very, very bright future. And I’m glad that you mentioned that, the 2020 number, because we’re almost running out of time here today, unfortunately, and so I want to ask you as a last question, a last thought here, about something that your CEO and co-founder of Avaamo said, actually, earlier this year. Ram Menon, is that right? I’ve got the name right, have I?
Kunal Contractor: Yes.
Rahim Rahemtulla: So he was at the World Mobile Congress and he said that conversational AI is really still at day one and is only actually going to get more sophisticated. And so I’m wondering, with that in mind, if you were a company now who’s looking at this technology and you’re thinking that, what do you do? I presume you want to say that you should get on board with it now and you don’t need to wait until it becomes more mature. So just tell me, sum it up for me, what is for you the key driving reason that any enterprise needs now to get onboard with conversational AI and not wait around?
Kunal Contractor: It’s a good question. Yes, within the world of AI, it is still very early days. Just even five years ago, the prospect of the kind of things we’re able to do now seemed far-fetched, futuristic and stuff you might have just seen in movies. So we’re not, also, nowhere near the levels of Skynet just yet. They are predicting that bots will go to perform full surgeries probably in about 50 or so years from now, so we’re still quite some time away, even though bots are still helping with some elements of things like surgery.
But one of those core reasons, to specifically answer your question, Rahim, is conversational computing is a very rapidly evolving technology and it does promise to change the way that we work and how a lot of business would interact with their customers, with their employees, stakeholders, and with a massive capability of impacting both the customer experience and reducing cost at the same time. So that one reason is what we call “last mile automation”, because if you’re able to address the core aspect of a business to allow it to grow, remain competitive and not be at the detriment of their customers. And especially as we know from history, any and every early adopter to get technology always gets to the top. Whether it’s a military aspect or a commercial or business aspect, that’s always been the case. So that’s a final note. Quite simply, that’s the biggest reason.
Rahim Rahemtulla: Fantastic. Yeahs, well, early adoption and how that is going to lead to good things. So, Kunal, I thank you very, very much for joining us today. I think we’ll have to end our discussion there but I’d like to, yeah, thank you very much for taking the time to meet with us.
Kunal Contractor: Not a problem and thank you very much for the opportunity.
Rahim Rahemtulla: It’s been a pleasure. Thank you to our listeners as well out there for tuning in. I want to say, if you’ve enjoyed my interview with Kunal, don’t forget to join us on Thursday, this Thursday September 13th. We’re going to be hosting a webinar on AI and the future of the workplace, so I think my conversation with Kunal has set us up very well for that and we’re going to go into more detail. We’re going to have a panel of three excellent speakers, so please don’t miss that. And visit our website, siliconvalley.center for more details and to register. I do hope you will be able to join us then. Unfortunately, that’s where we will have to sign off today. From me, from my guest, Kunal Contractor of Avaamo, goodbye.