The Millennial generation is still retail’s most coveted demographic, but Generation Z isn’t far behind. Both cohorts grew up as “digital natives” with computers, game consoles, and mobile phones. It’s no surprise then that being online is second nature to them, and increasingly, chatbots is where Millennials & Generation Z can be found interacting online with retailers & other businesses. Conventional marketing wisdom has long dictated that you have to meet your customers where they are. Our guest on this episode argues that means chatbots are where marketers should go.
As Co-Founder and CEO of Gupshup, Beerud Sheth knows quite a bit about chatbots. His bot-building platform is already being used by over 30,000 businesses worldwide, and was selected by Amazon as one of two preferred bot builders for their .bot registry services. With bots increasingly automating many tasks & activities currently done manually, Beerud also has interesting insights on how they may disrupt the labor market, another topic he’s well-versed in having previously co-founded eLance which kick-started the gig economy. Beerud shares his vision with us about the opportunities that lie ahead, and the steps businesses must take to maximize their return from automating this growing customer touchpoint.
Guy Nadivi: Welcome, everyone. My name is Guy Nadivi. I’m the host of Intelligent Automation Radio.
Our guest on today’s episode is Beerud Sheth, Co-Founder and CEO of Gupshup, an advanced bot-building platform used by over 30,000 businesses worldwide, and which handles over 4 billion messages per month. Prior to Gupshup, Beerud was one of the co-founders of Elance, which many listeners will remember was a pioneering online platform for freelance workers that really helped to kickstart the global gig economy. Those accomplishments alone give Beerud a unique perspective on how digital transformation will impact the future of work, two topics of immense importance on this podcast. We’ve asked Beerud to join us today and help us peek over the horizon a bit to better understand what that future might look like. Beerud, welcome to Intelligent Automation Radio.
Beerud Sheth: Thanks, Guy. Thanks for having me here.
Guy Nadivi: In 2019, Business Insider estimated the worldwide chatbot market as worth a bit more than two and a half billion dollars, but they forecasted that by 2024, it will approach $10 billion. That’s a compound annual growth rate of over 29% a year. We are, of course, now in the middle of a global pandemic, and so, Beerud, I’m curious, how do you think the COVID-19 crisis will affect this growth rate? Beerud Sheth:Oh, I think it’s certainly going to accelerate it. In many at least technical areas, we’ve seen decades of progress compressed into these three months, with a rapid adoption of, let’s say, work from home and telecommuting and remote meetings and so on. I think the same will apply to the chatbot market as well. It is accelerating the interest in developing bots and implementing them and so on for a variety of reasons. One is people are more conscious of social distancing and not getting into physical contact and so on. Once you get into remote customer support situations, whether it’s a human agent or an automated chatbot providing the responses, it matters less. It’s different when you’re face to face at a desk, but once it’s remote, it’s a lot easier to automate. So customer support is certainly one of those big areas. Also, there’s this massive shift to online commerce from offline. Suddenly retailers and shopping assistants and transactions, ordering pain, a lot of that is moving online, which again is rife for automation through chatbots. That’s happening across many, many different aspects, but as we move more and more into a tech & virtual economy, most of those interactions can be automated, can be enabled to a conversational chatbot. So I see it increasing dramatically.
Guy Nadivi: You mentioned customer support. It turns out that customer service and the IT help desk seem to be the two most popular use cases for chatbots right now. Beerud, what other functional areas of the enterprise do you envision will experience growth in chatbot usage?
Beerud Sheth: The way I think about it is, for a business, literally every customer touchpoint is going to get transformed with conversational experiences through chatbots. If you think about it, the entire customer life cycle, businesses, it starts with marketing on the front end, and then sales and transactions in the middle, and then support at the end of that customer life cycle. Across all of those touchpoints, you will see a greater adoption of chatbots. Let’s take the example where a customer is looking for products. If you had a chatbot that can educate, that can explain … I think Sephora had a chatbot where you can look for different colors and different shades of cosmetics long before the buying decision is made. Or you have complex products. If you’re buying insurance, if you’re buying health care services or large complex products, you want to look at review, want to look at products, look at options and so on. It makes it … So that’s the whole front end of the buying cycle, the marketing part of it, or even sending deals and offers and coupons and engaging with that. That’s on the marketing side. Then when you come to sales, the actual transaction, the merchandising, the purchasing, the shopping cart, the checkout and so on. Then even the support long after that, in case there are problems, returns and so on. I think every customer touchpoint really is going to get transformed. Today, if you think about it, most businesses have websites and apps. You really have to think of chatbots as the latest reincarnation of websites and apps. Because consumers, especially millennials, the younger ones, they prefer the messaging … They’re using mobile devices. On the small screen, messaging is used far more heavily than either browsers or apps. So businesses will have to meet consumers on their turf, on their terms. You have to meet customers where they are. If they prefer the chat and conversational experiences, then clearly businesses will have to interact in the same way. Everything from browsing to window shopping to ordering food and items and so on. You can see a version of this future if you look at WeChat in China. Everything there, a lot of it happens through the messaging app where you can interact with, not just businesses, but even every vending machine or every billboard or a restaurant table. You just arrived at a restaurant. You want to order food at that table. You can just do that through conversational experiences. In their case, it’s mini programs. But that’s the idea. Just through the messaging app, it becomes a super app that allows you to do virtually everything. I think businesses will have to transform their digital touchpoints into chatbot.
Guy Nadivi: Gartner believes there are as many as 1,500 chatbot vendors worldwide, and that “The majority of these conversational platform vendors offer very simple platforms using modified open source components to deliver simple question and answer chatbots.” I think it’s safe to assume the market doesn’t need 1,500 chatbot vendors, & there’s an inevitable shakeout coming. Beerud, even though we’re still fairly early in the conversational AI era, what kinds of things do you think will differentiate the survivors from the chatbot vendors who will fall by the wayside?
Beerud Sheth: I’m a little more optimistic about the future for chatbot vendors. There are different elements. Of course, some vendors provide platforms, and the number of platforms will be few. But a lot of other vendors provide services, and there there’s limitless potential. The analogy I like is think of it as website development or app development. There are numerous, perhaps thousands or more, dev development agencies, app development agencies, and so on, or even, let’s say, social media marketing agency and such. Those provide very custom solutions to brands. So if we say that virtually every business and brand will need to build conversational chatbots, and not just one, but multiple ones across different use cases, then there is a need for lots and lots of agencies and chatbot vendors to provide these services. To survive, what the vendor should do is get deeper into the business workflow. If somebody specializes in banking chatbots, and even within that it could be retail banking, or banking in the US versus in Europe versus in Brazil. There may be differences in each of these cases. The workflows can get fairly deep, fairly complex. There will be a need, because all these organizations are going to need to build these chatbots. So I think there is room. The key is, like in any business, you have to figure it out your core competency, your unique differentiation, something that you can offer that somebody else can’t replicate. If you do that, then I think there’s enormous room for growth, but I think we’ll need lots of … If you’re going to have millions of bots, you are going to need thousands of vendors providing some of these services. The caveat is, I think, some of the underlying platforms will be fewer and will get standardized on a smaller set. There, again, it depends on the comprehensive nature of the capabilities, the breadth and depth of the platform, how many things it can provide, what it integrates into and so on. That’s what the platforms have to watch out for.
Guy Nadivi: Interesting point. Beerud, there are some concerns about biases, intentional or otherwise, creeping into the AI that powers things like chatbots. In order to root out bias, do you think that AI algorithms should be audited in the same way financial statements are for publicly traded firms?
Beerud Sheth: I think it’s a little early to institute audits. It’s important to think about it certainly. There’s no doubt. Well, let’s start at the beginning. Yes. The AI is trained based on existing documents, which are created by humans. Therefore human biases may be inherited by these AI models which just learned from these documents. The academic community is working on fixing these biases, on quantifying and measuring these biases. There’s been a lot of progress on that already. The awareness keeps increasing as well, given the social injustice movements going on. I think there’s certainly a greater awareness in the community around that. So firstly, the community is self-motivated to fix these long before the software gets into production. Nonetheless, there may be biases, ones that we may not be aware of or conscious of today. I think that’s been one of the human failings, that a lot of unconscious bias creeps in, things that we think okay today may not be okay years from now and so on. Certainly AI software is improving. The community is working on fixing these biases. Then as we go further, we need to make sure that businesses deploying AI use it in ethical ways. We need to hold them accountable if they don’t. So certainly there is room for some oversight, some audit, some policing of AI, not just of the AI models as well as the applications. Basically what’s inside the model and what they are applied for, both of those need to be … There needs to be some governance mechanisms around this. There’s no question. It’s just that it’s still a little early in the development of these things. I think we should let innovation flourish, but in parallel, there are a lot of conversations going on about how to use it, what to use it for. There are valid concerns around the surveillance state, around privacy, in addition to the biases and equal treatment and so on. So yeah, I think both those will have to progress in parallel, even as innovation gallops on. We need to move quickly to figure out how to monitor it. I think it will. I’m optimistic. I think the AI community is already familiar, both with its capabilities, as well as shortcomings, are already working on fixing many of the known issues. I think it’ll evolve over time, but yeah, I think we should be thinking about governance mechanisms.
Guy Nadivi: Your company, Gupshup, is one of two preferred bot builders that Amazon chose for their .bot registry services. Can you please tell us about that relationship and how you’re enabling the bot building market?
Beerud Sheth: Sure. Amazon happens to own the .bot top level domain. It’s like .com but .bot. I think it’s a wonderful domain name for a whole variety of bots that want to create a unified identity, that can use it for marketing purposes and so on. Those of us that were around in the late 90s remember how critical the .com domain names were in the development of the early Internet, and being able to create a memorable marketing identity to drive the adoption of websites at that time. In a similar way, as more and more bots are proliferating, each bot has a unique or a different identity on the appropriate channel. On Facebook messenger, it could be some name. On WhatsApp, it’s a phone number. On SMS or RCS, it could be different kinds of identities. It makes it very challenging. If a business wants to tell a user to come find its bot, there’s no single central location where it can direct it, because there are different … the identity’s fragmented. That’s where the .bot domain name comes in. It could provide that single central identity, which could then redirect the user to the appropriate bot on the appropriate channel. It can make it easier to identify bots, to market your own bots, to drive traffic and so on. So I think that’s the … In addition to identity, there could be additional services that could be provided right there around perhaps authentication, verification, and other services as well. That’s a shared vision that Amazon has and we have too. We want to make it easier for bot developers to get discovered, to drive traffic, market themselves, because I think it’s good for the overall chatbot ecosystem. We’ve been engaged in conversations with Amazon for a long time. I think ultimately when they went live with it, they selected Gupshup as one of the key platforms. What it means is really any bot developer on Gupshup can very quickly and easily, I mean, literally check the box and get a .bot domain name. It could be shopping.bot or travel.bot and so on. They can easily get those domain names. It’s seamlessly integrated. That makes it easier. So we’re excited about the partnership. I think it’s good for the ecosystem, and it will drive greater adoption and standardization in the bot ecosystem.
Guy Nadivi: Beerud, before your foray into conversational AI and chatbots, you helped launch the gig economy by founding Elance, which eventually merged with oDesk to become Upwork, which today is the largest freelancer marketplace in the world. Do you think that one day we’ll see a similar marketplace for task-specific bots that organizations and even individuals can hire on a per hour or per project basis?
Beerud Sheth: That’s certainly conceivable isn’t it? In fact, for all you know it may already be happening and it’s just not visible. It is entirely possible that some freelancers might already be deploying bots to automate their tasks and activities. That may very well be happening. At that point, what does it even mean to have a per hour billing, because bots operate on a different timescale? Per project, of course, makes sense. So yeah, I think that’s exactly what’s happening. Once you automate tasks that were previously done solely by humans, it blurs this boundary between a human delivered task versus a chatbot delivered task. Or it could be a hybrid task, part human, part bot. Over time, that distinction may disappear altogether, as in the customer may not know, or may not even care, so long as the deliverable is done. Upwork often describes itself as a talent cloud, which is you get access to, just like you have the computing cloud and the storage cloud, I think of it as a talent cloud, where with one click, you can get access to loads of talent. It’s just that talent gets redefined as not just humans, but also automated AI programs delivering these things. So I think we’re in for some very interesting times as this evolves. I think it just opens up a whole lot of possibilities. It’ll be very interesting to see how it develops.
Guy Nadivi: Beerud, for the CIOs, CTOs and other IT executives listening in, what is the one big must have piece of advice you’d like them to take away from our discussion with regards to implementing conversational AI at their enterprise?
Beerud Sheth: Well, the one piece of advice would be to get started on this journey. It’s a long journey for multiple reasons. AI itself is developing. AI is not always accurate, but the accuracy rates are improving very dramatically. It behaves differently in different domains. There’s a lot of trial and error and testing that has to be done. But it is critical that executives get started on this journey, and take small incremental steps towards creating conversational experiences, enabling chatbots, leveraging AI to enable this. Because the cost of not doing something will far outstrip the cost of doing something. If you do something, and if it fails or if there’s some mishaps, I think the learning experiences, that will help you get there faster. We’ve seen these sort of transitions before as well. In the mid 90s, there was a time where a lot of marketing executives would say, “Well, why do I need a website? I already have a phone number. I have this offline retail presence. Nobody’s visiting us on our website.” Very quickly, it switched over to you were fired if you didn’t have a website, if you didn’t have a digital strategy and so on. Then it was the same thing with mobile experiences about a decade ago, with people needing to create apps and such. I think we are, yet again, at a similar point, where we are going to go from, “Why do I need a chatbot or conversational interfaces?” to it’s going to become the critical, the primary way in which consumers interact with brands. Anyway, I think the opportunity is huge. Yes, there are certain challenges in getting it right, but it’s critical to get started on the journey and get started now.
Guy Nadivi: Great perspective. All right. Looks like that’s all the time we have for on this episode of Intelligent Automation Radio. Beerud, it is a real honor to speak with someone whose previous work with Elance positively impacted so many people around the world, and whose current project with Gupshup has the potential for an even greater impact. Thank you for joining us today. It’s been great having you on the show.
Beerud Sheth: Guy, thanks for having me.
Guy Nadivi: Beerud Sheth, co-founder and CEO of Gupshup, an advanced bot-building platform, and previously one of the co-founders of Elance, known better today as Upwork. Thank you for listening, everyone and remember, don’t hesitate, automate.
Co-Founder and CEO of Gupshup
Beerud Sheth is the cofounder and CEO of Gupshup, the world’s leading platform for cloud messaging and conversational experiences. It is used by over 30K+ developers and handles over 4.5 billion messages per month. He previously founded and led Elance (now Upwork, a publicly listed company), the pioneer of online freelancing and the gig economy. Prior to founding Elance, he worked in the financial services industry — modeling, structuring, and trading fixed income securities and derivatives at Merrill Lynch and Citicorp Securities. His graduate research, at the MIT Media Lab, involved developing autonomous learning agents for personalized news filtering. Beerud earned an M.S. in Computer Science from MIT & a B.Tech. in Computer Science from IIT Bombay, where he was awarded the Institute Silver Medal.
Beerud can be reached at:
Beerud’s blog: https://blog.gupshup.io/