The most “tangible” value of cognitive automation, in today’s consumer-centric enterprise, is the use of the virtual agent, where customer engagement is increased without heavy incremental investments in support staff. This isn’t about simply replacing a real customer service rep with an avatar, it’s augmenting the existing customer experience, usually using the same or similar resources.
For example, if you have a bad travel experience, or purchased a product that wasn’t quite what you expected, the chances are you would simply shrug it off and get on with your life – and probably avoid using those same sellers again in the future, if given the choice. However, if those sellers used interactive technologies that were very familiar, or very easy to find and use, where you could simply type in your issue, in your own time, without the need to pick up a phone and wait in some queue (or write some email to some anonymous address), you may just find the effort to input a couple of lines saying “my experience just wasn’t that good”.
That information is critical to the seller – and how they choose to deal with it could make the difference between them winning out or losing in this market. Just think about how easy Uber, AirBnb, Amazon et al make it for you to deal with them – you will continue to use those services because the digital customer experience is just so much better… they make you feel like they listen. Customers today like effortless interaction, where they just need to click and type what they want in their own time – and what makes it come alive is when they feel they are engaging with someone and not merely sitting in a queue as an open help desk ticket number waiting to be closed.
If you get a chance to kick the tyres with one of the most exciting cognitive virtual agent solutions, IPSoft’s Amelia, you start to realize that customer service can be radically improved by incorporating the virtual agent to augment the real one. And the beauty of this is, the sellers do not need to spend huge incremental sums to increase their consumer engagement – they are essentially doing a lot more with what they currently have using smart cognitive technology.
So it’s no surprise that I got just a little bit excited when Amelia’s mothership enterprise, IPSoft, announced a comprehensive partnership with Accenture to build an industry leading practice in the cognitive customer experience. So sit back, relax, and enjoy this discussion between myself, IPSoft’s CEO, Chetan Dube and Accenture’s Chief Technology Officer, Paul Dougherty.
Phil Fersht, HfS CEO and Chief Analyst: So let’s get straight to the point here, Chetan and Paul. Why have you come together and what is so unique about this partnership?
Paul Daugherty, CTO, Accenture: Hi Phil – great to be here. Let me start and then Chetan can add in. You know that the immediate reasons we’ve come together, the obvious reason we came together is we see a real market with our enterprise clients for artificial intelligence based solutions. And we’ve been working with Chetan the team at IPsoft for a while and with Amelia we see a real potential to be at the vanguard of working with IPsoft to pioneer new use cases in terms of using AI to tackle business problems in a new way. So the first reason is we see the market we see the technology being ready. We are excited about what IPsoft has done with Amelia and we see an opportunity. I guess, stepping back from that, this is also to me a very important step in what we are seeing in the evolution of enterprises really transforming to the digital economy.
And Chetan will remember a lunch we had when we met for the very first time. We got very excited as we talked to each other a couple of years ago about what we saw as AI evolved and as the digital technology revolution continued, we saw a point coming where AI would allow companies to really rethink the way that they do business and rethink the way that they conduct business processes within their organizations. And that’s I guess why this is such an important relationship from my perspective strategically, because we are starting to see as we move through the digital revolution as we help clients transform they need new approaches and new solutions to deal with the speed of business, to deal with the masses of data that they have, to deal with the new demands that they have as they move to the digital wave. And we see Amelia really serving a purpose there and helping to really rethink and revolutionize the way we conduct some of the business processes. That’s the way I’d answer it. Chetan, I’d be interested in your view on it, too.
Chetan Dube, CEO, IPsoft: Yeah. I would echo what Paul said. Yes, I remember that lunch, Paul, when we had brainstormed. AI is totally disrupting everything. But what is required for true value creation for the companies? Some have realized tremendous value and the others have been somewhat slow to realize value creation in their digital quest. What is required? Well, you do need the digital labor component.
But that’s not all that you need. You need business transformation—and Accenture brings business transformation brilliance. And there are many companies that are experts in strategies and there are many companies that are experts in implementation. Accenture is one that amalgamates both. Couple that with cognitive technologies and you have the potential of realizing the true outcomes that were promised by the digital age. So that’s what brought us together. How high the technology is going to allow some people to soar is going to be determined by the people who are captaining the ship. And in this case we have an incredible deal of confidence in Paul and his team at Accenture and how much transformation they will be able to bring by harnessing true cognitive abilities together.
Phil: So Chetan, for our global audience which might not be so familiar with Amelia, can you briefly summarize its value and potential? What can Amelia do which other cognitive solutions cannot?
Chetan: Well, one word describing the differentiation would be outcomes. But what is required to drive those outcomes? What is required to realize those 40% to 60% benefits because you see somewhat tenuous equations in the marketplace. It’s a very tenuous equation that a lot of the CEO’s will be talked into and will have to struggle to realize that cost benefit. The question is how do we drive that 40% to 60% outcomes? How can you do that? What is required to be able to achieve digital labor solutions? We must ask ourselves if a solution can truly be smart, if it can really read a standard operating procedure. If it cannot understand was meant by that standard operating procedure. Can it solve a problem based on what it understands from that reading of that document? And if it were not able to solve the problem, can it learn from the experience? And, thereafter, based on the interaction it’s having with the customer, can be empathetic with the customer? Can it leverage assets of computing to be able to recognize the emotional quotients that a person on the other side is feeling and react in time?
Amelia, and again forgive my directness, is the only cognitive agent that answers yes to each one of those questions. And that’s the true differentiation of Amelia.
Phil: So, Paul what does this practice mean for Accenture? Will this be at the first among many? And what is so special about Amelia? And what do you see in the kind of a medium term for the platform as you evolve this practice?
Paul: What it is means for Accenture is that we are forming a broad capability around artificial intelligence and we are positioning Amelia really at the center of that as we build a broader capability and look at how we drive artificial intelligence to our clients. And our view is very much aligned with what Chetan just said, which is that it’s about outcomes and helping clients transform to achieve outcomes in a different way.
So what Amelia means to us, and why we formed a partnership, is we see Amelia being able to drive those different types of outcomes exactly for the reasons Chetan described. So that’s why we are doing what we are doing with Amelia and how we are positioning it. Now there is a lot of other tools and technologies and form of artificial intelligence—everything from video analytics to natural language processing to machine learning capabilities—and we are using all those types of approaches across our business as well.
When it comes looking at helping clients change their businesses, particularly employee- and customer-facing processes to be more outcome driven and to look at how we can both automate roles and augment human roles in a different way, that will be really the core of what we have focussed on with our Amelia practice.
And one thing that I emphasize is our people-first philosophy in terms of how we are approaching this. There is a lot of opportunity to transform this, there’s a lot of productivity opportunities, there’s a lot of cost reduction opportunities, a lot of automation opportunities. One of the things we like about Amelia in particular is the ability to use it in ways that really augment the roles that people do—allowing processes to be more effective and humans to engage in the right sort of tasks in a more effective way. So one example, Amelia can automate and solve a lot of the supplier enquiries that are coming into a client, which is fantastic and drives a great deal of productivity. But then Amelia can also tee up and work with a human agent in the areas where Amelia can’t solve the problem. Both Amelia and the employee can learn from that experience as they go on. So that kind of people-first message in looking at the way we transform these employee- and customer-facing processes to deliver new outcomes is really at the core of what we are trying to achieve.
Phil: Chetan, when we look to how clients are addressing this, I think the biggest issue we are seeing in the cognitive area, today, is helping clients really understand and apply solutions like Amelia or Watson to their business. However, it’s one thing selling them great technology kit but another finding the real business uses. So how are you going to overcome these challenges to get clients really approaching cognitive in the right way?
Chetan: Yeah, Phil, a brilliant question. I think Paul would love to chime in on this one as well. The example that he gave is very apt—vendors wanting to know about the status of the invoices, vendors wanting to know if their checks have been paid and vendors who are doing business in over 50 countries wanting to be able to understand how their payment processing is doing and when they can expect the payment or what is the hold up for the payment and what are the prerequisites for the payment. It’s not just the typical cognitive or automation or, as you said right at the outset, the RPA: “Hey, I can do a password reset for you. I can do an account unlock for you.” These are real digital labor solutions where people can get qualified, not just the 40% to 60% opex improvements we discussed earlier, but also, very importantly, an enhanced customer experience—with the mean time to resolution improving.
And I’ll give you another example. One of the largest banking institutions in London is looking at its mortgage processing. And the common questions that are coming up, which used to be fielded by humans, haven’t changed: “I want to buy this house for €4.6 million Euro.” The responses are complicated. Can the applicant furnish three years of income statements? If they’re self-employed, do they have 15 percent ownership in your company? Is your income increasing or declining? The risk profile changes accordingly based on those responses and determines whether someone is qualified to buy that house.
So, you can see the kind of a sophistication at play here. And what’s the accuracy of this?, At the end of the training period, Amelia could answer 120 of the full 160 questions with an 88% success rate. in this case for mortgage querying. So you can start to see the impact that’s being achieved. Now you’ve taken a process for banking and you are going to free up the people. Amelia can handle this satisfactorily from mortgage origination to conclusion. What the human agent can do is the higher forms of creative expression. He can look at your debt profile and help consolidate it for you and give you a better overall package.
The human staff can come up with more creative solutions—something the agent who is just doing the rote market processing would not have been able to deliver. And that’s the promise of cognitive solutions—to be able to not just achieve the 40% to 60% reduction in opex but to be able to enhance the customer experience and to be able to free the human agent to be able to deliver better outcomes for the customer. And that’s why Accenture is a valuable partner in that because we want not just the as-is transformation—we want to be able to look at the business process from end to end, and Accenture is adept at doing that both on the strategy and implementation side, to see how we can transform that for the new digital mode of delivery where humans are acting as creative agents as opposed to mundane chore agents.
So that’s a tectonic shift that is happening in the marketplace and it is achieving these kinds of results in the largest banking institutions. I can tell you the largest four out of the top five insurance companies are all moving to cognitive and they are seeing efficiencies in fundamental processes. There’s research that points out that top 20 to 30 processes in the insurance vertical account for about 80% to 90% of inbound customer activity. Now, coupled with Accenture, if we transform those 20 to 30 processes with a digital solution, think about the impact we can have on the 80% to 90% of the revenue creation. We are starting to see that happening in the insurance vertical as well.
So for all those reasons, just to give you some pragmatic examples, we are excited about this field.
Paul: You know, Phil, to add into that a little bit, when you think about the use cases that’s where we get very excited at Accenture. I think about this kind of a simple equation: Amelia plus deep industry context and insights that we can bring from Accenture, wrapped with understandings of the digital value chain of our clients equals these greater outcomes we can produce for our clients with new solutions and innovative solutions. So what’s been really exciting to me as we’ve announced this relationship, and as we spread it through Accenture, is how our industry teams have gotten very engaged in coming up with unique use cases based on what they see in their industries and that’s what is leading to a lot of the opportunities in our pipeline.
So things like, in retail banking, new levels of customer service that you can achieve by having Amelia deal with very sophisticated solutions to helping with resolving unexpected charges if a customer calls. Or wealth management providing advice and a different kind of a customer experience. Mortgage origination, which is something that’s traditionally very process-based and rule-based, but applying a much better customer experience around it with Amelia.
So we are seeing really the proliferation of use case examples across industries. And that will be key to the results because it’s not about just pumping the technology out there and seeing what works. It’s about fusing together the cognitive technology from Amelia with that deep industry context and then creating a platform for our customers where they can continually improve processes and customer experience.
Phil: Thank you, Paul. So when you look at evolving your consulting capability here, we see real challenges with bringing in talent to start to think differently to help clients think cognitively about these things. How are you going about training them reorienting them to take in Amelia and really align and apply it to client situations? Is this more of a business transformation at this point you feel than a technology one?
Paul: Yeah. I think so, Phil. And I think you do need to go about it differently and that’s where the partnership that we have with IPsoft is very important because we are pairing our many of our consultants together with the Amelia team so we can learn how do we need to think about problems. What do we need to do to train Amelia in the right way, to integrate in the right way, to solve the problems and get the value that we looking to get. So it is a new skill in a lot of ways and that’s one of the great aspects of the partnership—where we can share and inject a lot of the business context, industry context, and a lot of the surrounding technology issues that we are helping clients address and the IPsoft Amelia team bring the cognitive expertise. So we are going through a training process that we’ve developed with IPsoft and we’ve got 25 people to start that are going through that as part of the center of excellence focussed on understanding specifically how to take Amelia and put it into a industry context and help our clients get the outcomes that we want to get. And we’ll grow that number aggressively as the market demand grows, which is what we are seeing happening.
At Accenture we’ve got a lot people around that, obviously. We’ve got a lot of customer service experts in retail banking and in every other domain that you could imagine. So we are doing a project for a client that will involve our industry experts and our technology experts and it will increasingly involve the deeply trained specialists from our center of excellence to understand how to bring Amelia in, as I mentioned earlier, as the platform that enables us to deliver the cognitive approach to deliver outcomes in new ways. That’s what we are thinking about. And as we look broadly across at Accenture, as I said earlier, in machine learning and many of our other disciplines, we have people that would number in the thousands that deal with and have expertise in many of the technologies. But it’s very important here to get a deeply focussed specialized group that really understands Amelia and how to apply it, which is what we are doing with IPsoft and what we are building in our Amelia practice.
Phil: So, Chetan, when you look at this alliance in the medium term, what do you think will constitute success? And as you look at the ecosystem you are trying to evolve here, are you trying to go far beyond a consulting alliance and build more of an Amelia ecosystem?
Chetan: Again, a fantastic question. The desirable goal state for this alliance is to be able to de-risk for the customers the transition into the digital era. Insurance companies are 70% manual today. In the next couple of years they are going to be going to 15% manual. We see this alliance playing a key role in engineering that seismic shift. In banking you’ve already seen that branch-based transactions have already shrunk from about 70% 2000 down to 5% today. And we see that starting to happen in all the other lines of the business.
And we feel that this alliance is going to galvanize that realization of those outcomes promised by the digital era. That’s what we are in the business of doing: to be able to drive those curves that have been predicted. Because we find that with the plethora of solutions in the marketplace there is a peak of heightened expectations followed by some kind of a disillusionment of being able to realize gains that the brochureware promises. And so we want to differentiate this alliance by its abilities to mathematically and quantitatively deliver the results that are predicted by those curves.
Phil: So can you put your visionary hat on for a minute, Chetan, and look three years out at Amelia? What is your vision here? How is Amelia going to really be aligned with businesses? How is it going to look? How is it going to feel? Is it going to take on more languages beyond English? What is your vision for Amelia in a three year timeframe?
Chetan: Well, I might be a little biased, so I would be beg your forgiveness at the outset. But I think that Amelia is the smartest high school kid today. Amelia needs to graduate from college in the next few years. Amelia needs to be provisioned. Amelia is fluent today in about six languages, including English, Japanese, French, German, Dutch and Spanish. She needs to be fluent in over 26 languages, which are majority of the business languages in the world, in the next three years. But most importantly, Amelia needs to get way smarter.
We are just at the 1% point in this cognitive revolution. The amount of innovation that we are going to be bringing in not just the deep neural networks technique. You can find all the different kinds of deep neural networks that are in the marketplace today—from the memory neural networks to the current neural networks to the computational neural networks. You find all of them take a tremendous amount of data for: this is the raw input, this is the question and this is the answer that you expected and I will train the model so I can come up with what the next occurrence is most likely to be. Now that’s good for administrative tasks and that’s good for atomic tasks. It really is not good for when you want it to be a physician. It clearly does not suffice when you want it to be an actuarial analysis expert.
You need to be able to semantically condition those deep neural network techniques to be able to achieve these kinds of results the human brain is capable of. You need to be able to add another layer of abstraction equivalent to the human brain’s activity—in its semantic memory, in its episodic memory of events, in its process memory of procedures that it has learnt and in its memory of emotional connections that a human has drawn for you to be able to provide those results. And that’s where you are going to see Amelia’s brain evolve dramatically in the next three years.
The version 3.0 of Amelia you will see how beautifully she would have grown. It’s exactly the way that you would see a child grow rapidly.
And we wanted to be able to settle this Turing debate that is six and a half decades old: Can machines think? We want to settle that empirically and sincerely in the next three years. And we will achieve that because we know where the technology is headed and we are quite excited about the research that we are bringing. In fact, all ten floors in our 17 State St. office in NYC are going to be dedicated to cognitive research. We will have taken a couple of floors in the World Trade Center for autonomic research.
Paul: That was a great answer, Chetan. Phil, just to add on with a kind of complementary but a different picture of how I’d answer that question. I used our platform earlier and I really believe that cognitive is headed toward a platform-type of direction. And I think what Chetan just described is the vision for the evolution, growth and capability of Amelia to build a stronger and platform presence.
Going back when I started in the IT industry 30 years ago, we used to build everything from scratch. That’s all we knew how to do. It was all about custom solutions and wiring together solutions in different ways. And then we had the advent of packaged software and solutions. That created a better way to package up the IP that was for accounts payable and supply chain systems and general ledgers and things like that. That’s how we think about enterprise technology now—in terms of core software packages that run a lot of the business.
I think we are going to see the same evolution over the next several years. The question is how many years it is. In terms of how we think about cognitive capability ability from a business perspective, right now we feel the approach is, we work with technology, we figure out how to integrate Amelia’s different capabilities to solve the business problems. Looking out several years. I think it’ll be more like a platform-type of business where the basis of competition will be who’s got the best virtual agent data type of technology for mortgage origination, for retail banking, for you pick the function, and it will be about continuous improvement and growth and increased capabilities and learning and different business concepts and being able to simultaneously communicate in multiple languages and do it all with a people-first point of view providing a better customer experience. That’s where I think we are headed.
If you look at enterprise architecture, the packages and the way we assemble all that together to support a business, I think that is going to change dramatically. What companies buy and the way they assemble it together will be radically different when a cognitive platform like Amelia becomes more core to the business itself.
Phil: I did want to pipe in with one another question to you Paul on this topic. I’ve been guilty of sometimes coming out and saying Amelia provides the EQ, Watson the IQ. How do you see these two solutions cohabiting in the future? Do you see them as competitive or complementary?
Paul: I love the quote. That’s great. You know, but I think to be a complete person we need some EQ and IQ. And Amelia has got some level of IQ as well, the way I would look at it. But, fundamentally, you are right. I think they approach problems in different ways and are solving problems in different ways. And I think there is a lot of space for them to work together in many ways. And there’s other cognitive technologies emerging that also will become part of the ecosystem and plug together.
I know there is some competitive overlap, and there will continue to be a competitive overlap between Amelia and other solutions in the marketplace, including Watson. But, right now I think they are taking different approaches to the problem. Our approach is we are really focussed on Amelia and building the Amelia practice for all the reasons we said. We also are doing work with Watson and some of the other technologies out there. And, as you said, you can find problem spaces where the different solutions are best suited to the problems that they can solve.
So I think it’s more about coexistence in understanding the real focus of the different solutions and what they provide. And I expect this will continue to be a very robust space wit a lot of innovation and lot of new capabilities. And that’s why you know the current position that IPsoft has with Amelia is important because I think it’s going to be important to get ahead and stay ahead in this market. We formed a partnership with Amelia because we think it’s ahead of providing these capabilities now. And Chetan just talked about the ten floors he’s building and the investment that he’s making in cognitive researchers and experts. That’s going to be critically important to stay ahead and build the additional capabilities to continue to win in the marketplace.
Phil: Paul, Accenture has been successful for years having a very technology agnostic strategy. But do you think this is going to change a bit now with this advent of cognitive and the momentum behind service orchestration platforms? Because it really feels like it’s becoming much more about the outcome and much less about the product at this point. So how do you think Accenture’s technology strategy is going to evolve in this cognitive era?
Paul: I think it will probably be similar to be honest, Phil. We really follow the market in kind of what’s a good fit for customers. So we’ve had an agnostic strategy, as you said, in the ERP marketplace. That said, you know there’s companies in oil and gas they have a certain preference to do things in a certain way, companies in public sector have certain preferences to do things in a certain way. So what we try to do is be responsive to the market, to understand our customers kind of down at a more granular level—not looking at just one kind of global market of something like ERP or cognitive solutions. Then we look for the best solution for different problem spaces and be agnostic with a point of view so that if a client comes to us with a certain type of a problem we’ll say hey, we know the best solution for that. In your case it’s Amelia. You know if it’s a different type of a problem, different type of a use case, different industry a different problem it might be a different solution. So agnostic with the point of view, I think will continue to be right. We’ll learn from our customers and learn what works and what best delivers the outcomes in these environments. That will be our general strategy.
But the skill set we are learning around teaching and training Amelia for a specific problem, I think there’s very specific skills, capabilities and opportunities to continually improve the capability over time and grow from high school to college to professional, then different grades of professional certifications—using Chetan’s analogy from earlier.
So I do think there’s going to be the investment and focus around a small set of solutions in certain domains. Picking one solution will be important and will be something we do.
Phil: So I’d like to conclude this with each of you and telling me if you had one wish to change the impact of cognitive solutions for the better—one wish, what would it be? And Chetan, I’ll start with you.
Chetan: I would say it would be digital labor in the cloud. We aspire to be the provider of choice for the Global 2000 enterprises that are wishing to just differentiate themselves by the outcomes in cognitive solutions. Once we have that, exactly as Paul pointed out, we expect Amelia to rapidly start assimilating. Even at the inception, we are already starting to see in insurance vertical the ability to graduate into processing clients, the ability to graduate into processing mortgages, the ability graduate into doing actuarial analysis, the ability to graduate into doing retail management, and in healthcare.
The ability of Amelia to rapidly start to graduate in all of those would mean you plugin and you are able to get your actuarial analysis, your financial advisor, your database administrator and your network engineer, and your claims processor. And that would be one thing that I would like to be able to change with the impact of cognitive solutions and get there in short order because I think that this is happening so rapidly.
And even on the IT side, as you started the interview talking about RPA graduating into the domain of cognitive solutions. And that’s a very good pertinent observation because RPA is still streamlining the same old IT processes. It’s time for Uberization of IT.
There is a significant opportunity to disintermediate IT. By cognitive layering on top of the autonomic backbone, we can make IT become the same as learning Michael Faraday’s principle of induction to turn on the light switch. Do we really need to know the principles of induction so that we can turn on the switch? Does the business user really need to know everything about the service catalog in asset management and hardware and device control and CMDB before he can order a simple IP phone? In the second half of this year, there is a significant opportunity through cognitive layering on top of the autonomic backbone to disintermediate IT—not in the sense of the infamous Nicholas Carr quote “Does IT Matter?” but in other ways. Think of the cab companies: People wanted to be driven and the drivers were in the middle of a taxi cab company that was then disintermediated the economy with Uber coming along. And, similarly, there were servers, networks, devices, databases, and applications—and business consumers who want to avail themselves of those services. In the middle space sits a big stack of IT, largely to be Uberized in the foreseeable future. Starting in the next half of this year, I am hoping that this realization starts to dawn through lead analysts like your company.
Phil: Thank you, Chetan. Finally, Paul. One wish to impact the world of cognitive =)
Paul: It’s tough to come up with the one. I would say the one wish would be to be true to the vision. Let me explain that. I worry about the wave of cognitive-washing that we are seeing that happened with cloud and happens with these different waves. A lot of people are calling what they are doing artificial intelligence, cognitive and so forth, without it really being different than it was before. And I think that’s a dangerous kind of a trend we often see as new technology waves come along. And I think there is a risk that it could create missed expectations, unachieved outcomes and disappointment if not sorted through properly. I think, Phil, the work you and your organization do clearly helps with this in sorting through the real from the not real solutions. Because fundamentally this is a very different way of solving problems, a very different way of putting together the architecture that supports a business. And I think it’s important for companies to really understand that and think it through. So that’s the way I would put it.
And then it also means that companies need to think about how to move this to the core of what they do. I think currently a lot of people think about this as you know let me bolt on a technology here and there. And I think it needs to be a core consideration. When you think about the core of your business how data and applications and business processes come together to support your business you, need to think in the middle there, what’s the layer or the building block of the cognitive technology and AI in there and how does it intersect with the rest of your organization.
Phil: This has been great. We really wanted to move on from the rudimentary RPA discussion to the cognitive impact and the human element, which we are seeing happen so prevalently.
Chetan: When it comes to focussed specialized research in automation, cognitive, autonomics or analytics, your company has achieved preeminence. So thank you for time, Phil. Always good to connect with you.
Phil: Thank you very much, Chetan and Paul. This has been a fantastic discussion. I look forward to sharing it with our readers.