The Retail Podcast by Vue.ai: Leaders in Retail | Passage AI

Julia Dietmar: Hello and welcome to The Retail Podcast by Vue.ai, our podcast series on Leaders in Retail. I’m Julia Dietmar, the chief product officer at Vue.ai and your host for today. Today’s episode focuses on a topic that has been trending for quite a while now, and that’s conversational technology. Most of us have heard about and experienced chat bots, Alexa Google Home and we’re all anxiously waiting for a wider release of Google Duplex. Google’s newest human sounding assistant. Brands and retailers are exploring how conversational technology applies to their businesses and the ways it could positively affect their bottom line, from conversational marketing to conversational commerce and customer support. What are the most effective ways businesses can utilize chat bots and voice interfaces? To discuss this topic with us today, we have Ravi Raj, co-founder and CEO of Passage AI and Natural Language Understanding and Processing Platform that can be used to create deep conversational interfaces for a new Web site or business with over 20 years of experience in product development. Ravi has led teams at Yahoo, Kosmix, Wal-Mart Labs and BloomReach before he founded Passage A.I, Robert. Thank you for joining us today.

Ravi N.Raj: Thanks, Julia. Thanks for having me on the podcast.

Julia Dietmar: Tell us a little bit about yourself. You’ve been working in retail, e-commerce and specifically technology for retail for a while now. Most recently, Wal-Mart Labs and Bloomreach. What prompted you to start Passage A.I?

Ravi N.Raj: Yes, Julia. So my background is in search and machine learning going back to Yahoo! Where I worked in the Search and Marketplace team. And after that I worked at Kosmix as head of product and GM of their websites. Kosmix ended up being a top 20 site on comScore before it was bought by Wal-Mart. So Kosmix is where I met my two co-founders, Mitul Tiwari and Madhu Mathihalli So we met at Kosmix and we’ve known each other for almost 10 years now and we’re still talking to each other. That’s good. Which is a good sign. So we all have deep respect for each other their skills and talents. So one thing we all have in common is the background in AI and machine learning. So we wanted to apply that. You know, we felt like A.I had come of age like five years back. If you said that your car would drive itself to the airport and pick up your mom and bring her home, you would say, I’m crazy, right? But that’s not a reality. So A.I now has real world applications at the same time You have you know, in 2015 and 2016, a number of these messaging platforms opened up so you could create apps or what people call chattbox on these platforms. And people are spending a ton of time on messaging platforms like Facebook, Messenger, Whats App,WeChat and so on. It’s already the number one activity on a phone. So and also you had all these voices, systems like Amazon Alexa devices, Google Home Devices, Microsoft Cortana devices all flying off the shelves at the stores. Right. So all these three things coming together A.I coming of age, messaging platforms opening up and voice systems, getting more and more popular at home and even in the workspace. So what we felt like we needed to do was to leverage our AI background and create a chatbot platform with really high accuracy and state of the art NLP or natural language processing technologies that customers in multiple verticals, including retail, which, you know, all three of us have a background in, Retailer can go in, use our bot building console, build, train and deploy a chat bot on over 20 platforms. So that’s what we set out to do. One of our first customers was skools. They launched their chat bot last year on Facebook Messenger. And it’s the number one retail chat bot on messenger. They then went on to launch on their website, mobile apps. They even have a bot in the store that helps with the returns and things like that. So that’s the platform we offer – a console to build, train and deploy.

Julia Dietmar: Ravi, how do you think conversational commerce has changed the way people shop,.

Ravi N.Raj: Conversational commerce? Basically a way to buy a product. You know, just in a conversational manner, either on the retailer’s website or messaging platforms or on voice assistance like Alexa, Google Home, Cortana and so on, that has really made it frictionless for customers to buy products. You can just speak to a device or chat with brand and get the help you need to find the right product. Also, a virtual agent can basically help by asking a few questions about your taste, your likes, and then recommend products that match in those days. So basically it’s about the friction with such, reduce friction, you’re going to see more and more retailers deploy a conversational interface to their site.

Julia Dietmar: How do you see the future of the A.I powered conversational interfaces? Do you think personalization and engagement with customers will evolve in retail in the next year or so?

Ravi N.Raj: So, Julia, advancements in deep learning technology, which is the underpinning technology framework for a natural language processing system with new algorithms and techniques and advancements coming in deep learning. You’re going to see bots improve in accuracy so they can understand exactly what the customers are saying, just like a live agent would. So over time, you’re going to see this accuracy get better and better as technologies keep improving. Bots will also be smarter so they can understand what the customer’s shopping behavior is, what they like, what they don’t like, their tolerance towards price. Do they prefer a high end brand or they prefer a mid-range brand? These kind of things a bot will start to understand over time. So that will make recommendations better, which will then lead to increased sales for the retailer and also increased customer loyalty towards the brand.

Julia Dietmar: There’s been a recent study talking about how adopting automation can save companies up to a hundred and sixty five billion dollars by the year 2022, whether it’s in the retail, automotive or manufacturing sector. Can you tell us in what way AI is applicable across sectors and industries?

Ravi N.Raj: A.I can help automate repetitive tasks that do not require things like intuition or insight. Anything that can be easily automated using AI, you’re going to see that applications for that across multiple industries. So tasks that can be automated using AI could be anything from a self-driving car, which is one of the more complex ones, or drones to deliver packages with high accuracy or robots to pick products in a warehouse. Right. All these are now being powered by AI to improve efficiency, lower costs and the end customer experience is just so much better when AI is applied. You also have a computer vision technologies that help with visual search. Similar to Vue.ai’s products that’s also being powered by A.I, even things like our drug discovery and cure for diseases like cancer. They’re all being aided and accelerated using AI

Julia Dietmar: So I have a question that I’ve been hearing different opinions and voices on, and that is specifically around shopping through devices like Alexa or Google Home. What we’ve been debating within Vue.ai is that when a customer shops, especially if they’re looking for a product, they may not necessarily care about which retailer or which brand to buy that product from. Right. So I can go and say, Alexa, find me a pair of jeans. So when Alexa, find me a little black dress for an event. How do you reconcile that with the desire of individual brands and retailers to build their own bots and kind of drive customers towards that experience?

Ravi N.Raj: It really depends on on the customer the kind of loyalty they have towards the brand. If there is a particular retailer they really like, they’re very loyal to because they like the products. Or it could be even because they are earning loyalty points which they can redeem, then they have affinity towards that retailer. They will just like in the real world, they’ll go to the retail location and then shop at the aisles. Likewise, on Alexa or Google Home, they’ll just ask Alexa to open that skill, skill for Kohl’s or Levi’s or Nordstrom. And then you would interact with the skill and find the right pair of jeans. On the other hand, If you don’t have loyalty towards the brand, it’s like going to a mall. You know, you’re shopping around multiple stores and finding the best pair of jeans. So then you might say, Alexa, I’m looking for a pair of jeans and there is a bot like comparison shopping bot that aggregates catalogs from multiple retailers and lets you search across multiple retail catalogs. That could be a different experience. So it really comes down to the loyalty. You have towards the brand.

Julia Dietmar: So is it? If you were to advise a brand on whether or not to invest in a chat bot specifically for shopping with their brand, what would you say?

Ravi N.Raj: I would say, you know, chat bots offer a great way to build trust and loyalty with your customers. The main thing is building awareness of the bot. So just launching a bot will not get you the usage and the loyalty that comes with it. So driving awareness possibly in the physical stores saying talk to our bot, it will help you find where the batteries are or even on a website. You know, Kohl’s, for example, has done a really innovative job of building awareness. So when you shop on Kohl’s, on the order confirmation page they offer a way to track the order on Facebook Messenger. So that drives a lot of usage. You know, there is a whole demographic that millennials, they are more or less stopped using email, their own messaging platforms all through the day talking to multiple people. Right. So they don’t find a notification message coming on a messaging platform to be distracting at all. They’d much prefer to get it on that platform. So we’re seeing a lot of usage there. It’s a great way to build awareness. So you start to get notifications. And then once the order has been delivered maybe the retailer can send you deals and offers and then you can continue to have a conversation with the retailer, which then increases your loyalty towards the retailer’s brand.

Julia Dietmar: Speaking of Kohl’s, specifically, how was experience working with them? What improvements have you seen or have they seen with using your bot?

Ravi N.Raj: So it’s been great working with Kohl’s. They were one of the first retailers to launch a chatbot. So back in April of 2017, using our platform, they launched a bot to automate some of the most commonly asked customer service questions. So one of them being tracking an order which we talked about, the other is checking on loyalty points. Kohl’s has these rewards they offer. Yes To you rewards. So for every time you shop at Kohl’s, you get a certain number of points. And when you get up to 100 reward points, it turns into five dollars in Kohl’s cash. So a lot of the questions are people calling in saying what’s my balance or saying I did not get points for a purchase i made, things like that. So we automated all that using a bot. So you just say how many points to have you then log in securely and then you get the points. And in fact, the one benefit of a messaging platform is that you can show things like a barcode, if you have five dollars in cash. We show a barcode that you can take to the store and you can scan it. And you get five dollars off. So there’s no need to install an app or anything. You just interact with the bot and that has access to all the native features of your phone, like the camera, pictures, voice and so on. So that was another interesting use case in terms of, you know, helping customers check on their loyalty points and things like that, finding a store, checking on pharmacy hours, for example, really common use case on most apps or websites. It’s really hard to find hours of a certain department, things like that. Whereas in a conversational medium, you can just say, hey, Kohl’s, is the store open right now? Right. And then the bot could respond back saying the store closed right now. It’ll open again at 10:00 a.m. tomorrow. So things like that are very conversational. They started to focus on. They also gave us a knowledge base of frequently asked questions, things like your return policy, How do you earn points? So they gave us that database of questions and answers. And on their mobile site, They replaced the help center with the bot powered customer service center. And that was getting a lot of usage. And we’ve seen accuracies on that bot reach up to 95 percent as I mentioned before, leading to a friendly good customer satisfaction and then, of course, a lowering of customer service costs for Kohl’s. So it’s been really good. They are even experimenting with helping a bot return items in the physical stores. So you can go talk to the bot and say I want to return this item. You scan the receipt. It says that it’s eligible for a return. It gives you a number for a locker. And then you go to the locker. Place the item there. And then when the refund has been processed, you get a notification saying you received forty five dollars in credit. So things like that. We’re also seeing pretty good engagement from customers on features like that.

Julia Dietmar: That seems like a really good to use case. So I assume that the way you built your bots, you have a decision point at which you may hand it over to a human. So what are you seeing in terms of the metrics like I understand that the efficiency of the customer service has increased, the cost of customer service has have gone down for Kohl’s, but the customer satisfaction piece bot versus human versus first bot then human.

Ravi N.Raj: So, yeah, that’s a really good question because of the bot should not try to do too much. So there are scenarios in which the bot hands off the live agent with pretty much three bot We’ve launched for our customers. So there are three scenarios. One is if we see that the message, the sentiment of that message has turned negative, we also have a complaint classifier. So if we know that the message is actually a complaint saying, you know, I ordered this pair of jeans, but I got a different color or it was torn, whatever it might be. We can actually classify that to a complaint. So if the sentiment turns negative or if it’s a complaint, we hand off to a live agent, we just say, sorry, we’re not able to help you. Let me get you a customer service rep. The second scenario is with the bot does not have an answer to the question. You have to when you build the bot to create different intents in our console, you upload different knowledge based articles. You can create decision trees. But if the customer asks the question that’s not being configured in our console, then we hand off the live agent. That’s the second scenario. The third is in pretty much all our bots. We have an option for the customer to go straight to live agent, for whatever reason. They don’t want to interact with the bot. They can directly connect to a live agent and we’ve integrated with a number of live agent tools, sales force live chat, live Agent Oracle, Zendesk and so on. So these integrations make it really seamless for the bot to hand off to a live agent, when finished can handover control back to the bat.

Julia Dietmar: There’s a lot to be said about conversational bots that are coming to take the human jobs versus making human jobs more efficient and productive. What is your take on that?

Ravi N.Raj: So we don’t believe that boards are going to take away jobs. In fact, they’re going to hand out new jobs in the future by automating repetitive tasks, things like finding the refund policy of the website or getting help with exchanging an item, By automating such task bots can make live agents more productive. So when live agents their time is freed up from answering the same question over and over again, they can actually become more creative. They can add a lot more value for the customer by helping them select the right product, and in turn, they can help their enterprise by driving more revenue.

Julia Dietmar: Ravi what are you seeing in terms of the new patterns in the way retail brands are implementing the use of AI

Ravi N.Raj: We’re starting to see more and more new use cases for the use of the AI You know, most of the retailers are using A.I to automate customer service, using a chat bot similar to what our platform offers. Some of them are using it for conversational commerce, which we talked about earlier, where a virtual agent helps the customer find the right product. There are other applications of the A.I as well, including computer vision, where you can take a picture of a person and then you get recommendations for a panel based on what they are wearing or it could be a celebrity on Instagram, you take a picture of that person and a bot or A.I could be used to find similar apparel either at a specific retailer or across the web. And their applications, like in the warehouse, most functions are now being automated using robots. They’re picking and packing products. That’s leading to lower shipping costs overall. And then there are some customers. Some companies that are experimenting with drones to deliver packages. It probably has to go through regulatory approval process is my guess. But when drones are used to deliver packages, they can dramatically reduce shipping costs. So you’ll start to see that pretty soon we’re even seeing use cases where a retailer is doing surveys over the phone using a bot. So instead of sending a survey which gets very little participation, they’re experimenting with calling customers and just having a conversation with them and getting feedback on how they can improve their service, how they can improve their selection and fulfillment and so on. So lots of new use cases, but the most obvious ones are the ones I mentioned, automated customer service, visual search and applications in the warehouse.

Julia Dietmar: And what about your company passage A.I? what’s in store for passage A.I in 2019 and beyond.

Ravi N.Raj: We have a number of advancements in natural language processing technology or NLP. We’re gonna be launching pretty soon. Our bot’s already perform at an industry leading accuracy of ninety five percent right now. You’re gonna see that get even better over time. We’re also launching new features like Machine Reading Comprehension or MRC, which basically from a passage of text, it tells you the exact answer. You’ve seen Google do this in search. If you see if you’re typing what is Tom Cruise’s height, it will say 5’7″or whatever his height is. So it’ll provide the answer. So likewise, we’ve also launched features that give you the answer as opposed to reading the entire text of the article. So this makes it very messaging friendly. And in fact, going one step further, If the answer is yes to a question. We might just show a thumbs up emoji right. So we’re trying to make our bot’s responses really of messaging friendly and voice friendly. So really precise and concise answers to the user’s question on the bot building site. We’re going to continue to make improvements to the console, make it really easy to use. And for the most common use cases, we’re creating templates. So all you have to do is come in and enter a few information in some fields and then you have a board for a customer service using a knowledge base. You have a part for lead generation by helping the user fill out a form in a conversation manner or even conversational commerce, which we talked about might be a templatize from a bot building perspective where again, you just enter a few fields, maybe some API information and then you have a conversational commerce part. So we’re trying to make it really, really easy and quick to build a bot without sacrificing accuracy or natural language understanding.

Julia Dietmar: That’s very exciting. Well, Ravi, thank you very much for joining us today. It was wonderful to discuss conversational commerce with you. Thank you.

Ravi N.Raj: Thank you, Julia.

Julia Dietmar: With M.I.T. making a billion dollar abet on the AI education and the future of retail being fueled by A.I, there is a lot going on in this space, especially in fashion and luxury. For example, Ali Baba is installing high tech mirrors in women’s restrooms of all places, so that women can try and beauty products in front of virtual mirrors while their wait in line for the bathroom. Earlier this year, Amazon secured a patent for a blended reality mirror which could superimpose virtual clothing on your reflection. So A.I is truly slowly taking over retail. Stay tuned for more episodes of The Retail Podcast by Vue.ai, where we discuss this and other topics on fashion, technology and the future of retail. Until then, goodbye.

You may also want to see more episodes of Leaders in Retail.

The Vue Podcast: Leaders in Retail | Dhruv Toshniwal
The Vue Podcast: Leaders In Retail | Colomba Giacomini
The Vue Podcast | Sauvik Banerjjee & Ashwini Asokan