December 5th, 2018, 3:00 pm PT / Category: Interviews
How leading brands are using AI middleware across voice and text channels.
Ravi N. Raj is the CEO and Co-founder of Passage.AI. In this role, he leads a team of AI and deep learning engineers that have created the industry-leading AI platform that businesses can use to create a conversational interface for their websites.
Before co-founding Passage.AI, he led product for @WalmartLabs, Walmart’s hub for innovation around social, mobile and retail. Raj came to Walmart through the acquisition of Kosmix, where he served as the VP and GM of Kosmix’s sites, which included Kosmix.com, Tweetbeat.com, and RightHealth.com, then the second largest health site on the Web.
Raj has also played leadership roles at Yahoo, AltaVista, and Elance (now Upwork) and hold patents in search, social media, and retail.
Podcast
Transcript
Priscila: Hi everyone and welcome to another episode of DojoLive connecting tech experts like you, today I’m going to introduce you to my co-host we have Carlos Ponce live in Mexico City.
Carlos: Hi everyone.
Priscila: And we also have Tullio Siragusa who is live in LA right Tullio.
Tullio: Yes, good afternoon, raining in Los Angeles.
Priscila: And my name is Priscila Solis I’m hosting live from the city of Chihuahua and today we have a very special guest we have Ravi N. Raj he is the CEO and co-founder of passage AI in this role he leads a team of AI and deep learning engineers that have created the industry leading AI platform that businesses can use to create a conversational interface for their websites, so hi Ravi please before we get into the topic would you like to introduce yourself and talk a little bit about passage AI.
Ravi: Yeah absolutely, firstly thanks for having me on the show, I’m the CEO and co-founder of passage Ai as you describes we’re a conversation AI company, customers use our platform to build chat bots or conversational interfaces that can be deployed on their website, mobile apps, messaging platforms like Facebook, Messenger and WeChat or on Voice devices like Amazon Alexa, Google Home, and so on, my background prior to passage AI, I was VP of product at Walmart and I came to Walmart through the acquisition of a company called cosmics, and I was the VP and GM at cosmics and we where acquired by Walmart in 2011 and I worked at Walmart for 3 and ½ years, prior to Walmart and cosmics I was at Yahoo in search and marketplace for about a year actually, I was running product for Yahoo travel back in the day, so that’s my background and travel asie from being a big travel enthusiast I joke that between my vacations ‘i’m planning then next vacation so I’m really passionate about travel.
Priscila: That’s great, very nice to meet you Ravi, so today’s topic, the topic that you chose is AI powered customer experiences, can you talk a little bit about this topic and why you chose it?
Ravi: Yes, as you know AI is changing the world you know, from self-driving cars to drones that deliver packages to even thing like recommendation systems things like you know, cure for diseases and drug discovery like AI is really changing everything we do everything we know about, so you know, so that is really something that’s transforming the world, specifically in the area of travel and hospitality we’re seeing a number of applications of AI that is gonna change and improve the experience of travelers worldwide, you know, it started with recommendation systems that was pioneered by, you know, companies like Amazon and Netflix also in travel right, now you go to a hotel page and it has on the right rail typically other similar hotels right that’s all powered by machine learning which is AI and you know in some form so in that sense AI has been around for a while but now you have much more, you know, practical applications much more real-world applications in addition to the recommendation systems I mentioned and we’re seeing a fairly strong demand not just for our platform but for AI in general.
Tullio: Great, I don’t know what that is in the background [INAUDIBLE]
The orient- express is coming, ok, so I’m very curious about what gave birth to this idea right, I mean one of the co-founders like you guys are sitting around and like man I really we need this what gave birth to this idea, how did this come about.
Ravi: So actually three of us co-founders, myself, my co-founders Madhu Mathihalli and Mitul Tiwari, we all used to work at cosmics so we’ve know each other for a number of years, we still talk to each other which is a good thing, so you know, we have a lot of respect and you know, admiration for each other’s talents, I bring product and business, my co-founder Mitul brings you know, AI and machine learning and my other co-founder Madhu brings also machine learning and you know, full stack engineering to the table, so we had what we felt were all three components you needed to run a tech company so this was back in 2016, so the same year or coincidentally a number of messaging platforms started opening up so you could create the equivalent of apps for these platforms, Facebook Messenger, opened up in April 2016, you know, Skype opened up, Telegram opened up, WeChat has been around for quite a number of years they where the ones who pioneered chatbots, and so messaging platforms opened up, this the same year we felt that AI had come of age you know, AI has I mentioned has been around for a long time a lot of it has been sort of in the academic world, you know, and sort of in science-fiction world, right, is you know, with robots and artificial intelligence but we didn’t see a robot come home and do your dishes right, so there where no real practical applications of AI but we felt in 2016 you know, applications as I mentioned, you know, self-driving cars drones, the deliver packages and all that they where starting to become not just science fiction but a reality so that was the second trim that happened, the third was you know, the prior year, Amazon launched a product called Amazon Echo, and that was I am not mistaken, the best-selling gadget you know in 2015 it was flying of the shelves and soon after Microsoft launched Cortana, Google had Google assistant and so on right, so all these devices in your home, you know, call it IOT, called it smart devices, you know, they all had voice capabilities they could understand what you said, they could respond back to questions like you kno, what’s the weather like today, you know, what’s the stock price of this particular company and things like that right, so and at that time it was a little bit of a novelty when you had you know, talking to your speaker but now it’s become pretty mainstream right, Samsung for example has Bixby, which is their voice AI assistant it’s in pretty much every gadget, every device in your home so you could literally be talking to your washing machine or your range right, it’s actually now reality and do some really useful things you could turn off the stove by just talking to it, very soon so that was a third kind of trend that happened you know, voice being everywhere certainly cars you know, voice is the future, seven of the largest automakers in the world you know, they’re our customers, they are looking at many different applications to bring inside the car, the most obvious thing being that if you driving you shouldn’t be texting, you should be you know, using your phone much more intuitive safer and legal mechanism is voice so if you could just talk to you car and say you know, where’s the nearest Walmart for example, and the car could tell you, you know, turn by turn it to give you directions to the nearby Walmart, so those three things coming together again, you know, AI being mainstream and having real-world applications you know, messaging platforms openin up and there being you know over three billion users using messaging platforms and then these voice assistants and devices and smart speakers you know, in your living room and in your home now being able to understand natural language.
Tullio: Ok, so it’s pretty cool actually when you think about it there’s been some recent studies where everybody’s sort of living online and no one’s talking to each other so at least you can talk to the machines in your home, now hold joking aside, there’s a lot of players right, I mean foresters call these big waves of the top ten AI waves, one of them is decision engine it and also [INAUDIBLE] what differentiates passage AI what’s unique about you that will stand out, why should customers care about what you’re doing.
Ravi: That’s a great question, so we are in the conversational AI space for natural language understanding and processing is really important and it all starts with accuracy, we have what we consider is the leading accuracy in the industry in terms of understanding natural language and responding in a relevant manner, so our BOTS once they are full trained they perform at 95% of higher accuracy, again that’s the leading metric you know, in our space, the second thing is that we understand over 100 languages, we’ve built language models for all major languages, as you know you know, the world is flat, every company you know, has a global presence in fact many of the companies we talked to, here in the US are really interested in launching you know, worldwide in you know, languages like, Mandarin, Chinese, Japanese, German, Spanish, Italian, you know, Portuguese, Brazilian Portuguese, so all these variations and different languages they want to be able to launch their bot right, so that is a huge differentiator for us that is something we emphasize with our customers we prove with our accuracy in these languages so that is something to our knowledge very few companies do in terms of understanding natural language in multiple languages, the third is we offer all three thing you need to build a chat board in one platform, bought building, bough training, bought deployment, so you build the bot in our console in a UI driven manner, there’s no coding required, it’s all drag-and-drop and UI driven you build it, then you can train it using our console, you can upload training data like your chat logs, your customer service logs, you can upload them, and you can actually see the accuracy we report the accuracy in our console, so as you provide more and more training data, you can start to see the accuracy start to improve, and get to the level I mentioned, the third is you know, once it’s built and trained, we can deploy it over 25 platforms, all the ones I mentioned, messaging platforms, voice assistants, and also your website it’s a few lines of JavaScript you add to your website, and you can turn your website into a conversational interface, so those are the main differentiators in terms of again the accuracy multi-language native support and simplicity of building about like a one-stop solution or building it with bought building, bought training and bought deployment.
Tullio: That’s amazing, so I don’t have to be a developer to essentially deploy one of your bots, and when you’re uploading the data is it using lingua franca in terms of how you train them or already, or is it gonna have to get code that.
Ravi: No, not at all, actually that is a huge differentiator for us, as you correctly pointed out you do not have to be developer to build a train and launch aboard, it’s all UI driven when we work with a customers, let’s say a travel site, we tend to form a partnership with it, this is it’s not a customer vendor relationship it’s more of a partnership , they bring their knowledge of their business and their industry we bring our expertise in AI and machine learning so together we form a partnership and they understand their domain they use our console our capabilities to build it and they don’t have to write a single line of code.
Tullio: Right, speaking of travel industry, you know as a company here in Nearsoft, interestingly enough half of our clients after our business is all travel tech related, in the past 12 years we just amassed a bunch of travel tech companies it just kind of happen with clients taking us with them wherever they move, so how does this apply in for example for an OTA how does this add value, can you give us some examples of where that’s worked really well and some key metrics that you’re been able to improve by them deploying this.
Ravi: Yeah, absolutely, for OTA specifically it makes it much more frictionless if you have a conversational interface right,especially one that’s personalize if you take the example of the business traveler, they’re not going to spend a whole lot of time on planning visit right, it’s more functional they have to get to a certain place they have to stay in a certain you know, hotel and maybe they need to rent a car or whatever it might be right, for them it’s much more frictionless if they where sitting in their living rooms and talking to Alexa, and telling Alexa you know, I need to fly to Seattle tomorrow come back on Friday and I want to rent a car and I want to stay in downtown, then Alexa could give you some recommendations here are some flight that leave in the morning and maybe the system knows that you know, you prefer JetBlue so because you have frequent flyer miles on JetBlue so the board can respond back saying JetBlue has these flights tomorrow you know, one leaving at 8:00 a.m. other one leaving at 10:00 a.m, and then you say ok, book me the flight that leaves at 8:00 a.m and then, ok return flights, there you get some option you pick one, and then likewise with hotels, and then renting ac ar, it’s just mucho more frictionless you’re you know, you might have been watching TV the whole time and you’re booked your business travel right, and it’s not just business travel, it’s even consumers, you know, obviously take as many vacations and trips as business travelers to, so [INAUDIBLE] planning a trip, so they might be more interested in deals and offers, a messaging platforms is a great way to get deals and offers, especially again one that’s personalized for you, so you could tell the apart you I want to travel to Cancun during this spring break, I’m going to go with my wife and two kids you know, there are 16 and 18 for example, and you know, can you recommend the right deals and places to stay, and the the bot could be giving you deals and offers and suggestions you know here some really good places to stay some really good activities for your kids to do, and then you could be having a conversation with a bot you don’t have to book it right there, and then later when you find the right deal or the right activity, you can say: that’s a great activity why don’t you book that, and even if you don’t but if you’re not comfortable with using a bot to book it, you found the deal, or you found the, you know, the place you want to stay, and you could go online and book that, the, you know, that dealer or hotel, even for consumer travelers it’s you know, it’s much more you know, frictionless intuitive and easy to use when they interact with a bot to plan their trips, so yeah, definitely we’re starting to see real good traction with OTS in terms of embracing conversational interfaces and bringing it to their customers.
Tullio: Ok, I have one more question but before we do that I think we need to open this up to the audience, Carlos, you want to [INAUDIBLE]
Carlos: Yeah, absolutely, yeah, actually there’s, let me tell you something, we recently we have had a number, a good number of AI related interviews and we having an increasing number of AI enthusiast internally at Nearsoft, and one of them is actually, he’s a newcomer to the company Ruben Araiza, he’s the front end guy for especially DojoLive, now Ruben is asking me this question on Slack, this question is for you, for of course our guest, for you Ravi, and he’s asking this: Why should software companies prepare for human level AI technology now rather than decades down the line when it’s closer. I think that’s an interesting question.
Ravi: That’s a great question, yes AI is not at the level of humans you know, AI clearly does not have intuition, someday it might humans have intuition humans have creativity, you know, humans have emotions, things like that right, AI right now can do a lot of things that can be automated things that are repetitive in nature you know, driving a car for example you know, that lends itself to you know, being automated using AI, you know, because the you know, there is a clear sort of methodology to it so the reason why companies should plan for AI now is because already AI is powering a number of cuntions, so you know, they should leverage AI to automate functions lower cost improve productivity and make you know,humans more creative, because the time is freed up from you know, repetitive tasks in terms of why they should plan now for AI being as good as humans, firstly I don’t know AI will ever be as good as humans, some of the skill I mentioned like intuition and creativity, I don’t know if AI will get you there, but if you plan now if it happens sooner than later, let’s say it happens you know, two years from now given how fast technology is advancing given how quickly you know, techniques are emerging for deep learning an AI it could well happen sooner than two, three years, not everything at a human does but a lot of the things could be automated again to improve humans and their lives, this is not to replace humans, it is just like the industrial revolution, it made the world a better place by automating thing that you know, factory workers where doing manually right, it ended up making their lives much better, so likewise this is kind of the next you know, Industrial Revolution if you will the AI revolution, it’s gonna free up for you know, the time of you known live agents humans employees so they can add a lot more value to their enterprises and to you know, to themselves.
Priscila: And now that you mentioned this, are there humans behind the scenes with your platform for the lower accuracy areas?
Ravi: So one thing we do in our platforms is when the not does not understand what the user is saying, the not seamlessly hands-off to live agent, and we have done integrations with a number of live agent platform, Salesforce live agent, Oracle live person, you know, Zendesk and so on right, so when the not does not understand it hands off to a live agent, or the customer can say you know, let me talk to a live agent, if they’re not comfortable talking to a not we make it easy for them to just talk to a live agent, if they want to, you know, be on hold for 15 minutes or 10 hours to talk to a live agent when they become available you know, we completely want them to have the choice, so those are ways the bot, you know, interacts and you know, works with humans or live agents to provide the right service, the best customer experience for end users.
Tullio: Ok, so I have to ask this question because it’s to me it’s interesting a lot of these chat pods, which is ultimately about enhancing the customer experience right, would you consider that this could also be the next evolution in the user experience right, there, let’s go back 20 years , I used to call an 800 number to make reservation talk to someone, we didn’t have online, we didn’t have apps, that is it was a very personable one on one kind of thing, and good or bad, whatever it is, it doesn’t matter that’s how it work then we moved to self-service, we figured out how to make employees of the company by providing online capabilities where the service is always consistent cause you’re providing it to yourself and now we’re introducing the concept of the hybrid right, which is basically a semi-human kind of interface where it’s voice so it’s going back to what we where doing, but there’s the user experience and the benefit of still being a social service experience online, is it ultimately I mean, is the user interface ultimately just gonna disappear and be replaced by voice, is that possibility is that what’s gonna happen and what is your thought on that as a company.
Ravi: Yeah, so someone told me that you know, voice is the most intuitive and frictionless interaction paradigm, so exactly like you said before the internet you had to go to a travel agent you know, and then maybe several times, you know, tell the agent where you plan to go, what you know, plan to do things like that and you get recommendations and then you know, you make a decision, you know, book your flights and so on then, the internet came along, you don’t have to go to an agent you could go to multiple sites, check out different options make your booking, get your boarding pass on you know, by email which you could print, so much more frictionless then you know, mobile came along, which made it even more frictionless, you could be at the dentist you know, suddenly planning your next week or your business travel , just right there on your phone which is always with you, again the reduction in friction, you know you don’t have to go find it with [INAUDIBLE] or a laptop and plan your trip, now you know, you have voice assistance everywhere as I mentioned, literally it’s been not there yet you could be talking to your washing machine and planning a trip while you’re doing while you’re washing your clothes right, so it’s just much more you know, like reduce friction much more intuitive as I mentioned, you know, voice is the most intuitive easy to use interaction model you know, on a website you to find where a particular features, let’s say you know, you want to book a three rooms in a hotel, you go enter a dates, you know, and then find the hotel and then see if it’s the right price and so on right, much more intuitive to say you know, I want to stay at the Wynn Hotel in Vegas from you know, friday through monday and watch that show right, so it’s just more intuitive because that’s how human being converse but now there’s technology behind it that allows that interaction model which is voice or char, you know, to be able to you know, to use that to service your needs.
Tullio: Amazing, so we’re almost coming up on time, it always happens like that when again it’s really interesting, I’d like to shift gear a little bit and talk a little bit about sort of your culture as a company.
RCarlosavi: Sorry, and we have another question actually.
Tullio: Perfect.
Carlos: And actually it’s seems like we’re getting our guys enthused about the whole AI I think, I guess and ok, so here’s the question and the question is coming from actually from another Ruben and who happens to be the first Ruben’s song, wee look at the two Rubens working at the company, how odd is that, ok, well the younger Ruben, Ruben Araiza, he’s asking: what is the best way to get data ripe for AI interpretation to best optimize the travel experience?
Ravi: So in terms of getting the best data to train the model, it’s really important for the data to be accurate to be of high quality because the data is not accurate you’re gonna get you know, bad algorithms, bad AI, bad recommendations, so it’s really important to spend that extra effort to make sure the data you use to train the machine learning models, the AI models are of high quality they are labeled and accurate, and then you use that to train the models, so I would invest the time a lot if it’s in our space which is conversational do you have prior logs of customer calling in, and getting responses do you have prior logs of people searching for things and you know finding what they wanted, you can use all that to train the conversational AI models and create the best chatbot experience.
Tullio: Speaking of that Ravi, are you using like existing stacks, like the Berkeley stack or you’re created your own algorithm stack for this platform?
Ravi: So there are all these frameworks that have been open sourced, google tensorflow is one, Stanford NLP is another, so these things have tremendous momentum behind them, you know, as mentioned they’re open source so a lot of, you know, machine learning engineers from across the world they’ve taken these frameworks and they’ve enhanced them, they’ve pre trained them and you know, things of that nature right, for different use cases, so you know, rebuilding those frameworks would not make sense especially for a start-up so we use things like Google Tensorflow the values in building models, using those frameworks and more importantly to train those models with public data and customers own data and creating the right experience, so we go really deep into use cases, we try to understand the customers pain point and we provide a solution that’s driven by AI are using the models we build on top of these AI frameworks.
Tullio: That’s pretty cool, we actually have the chief envalist of Google as a guest back in the spring and in an offline conversation discovered Google has like an island, an island in Singapore a bunch of AI researcher, so it’s pretty amazing the investment they’re making in that space, I’m not surprised you’re using that, we’re almost, pretty much upon time I have a question about your culture right, because I would have to assume this kind of technology of Christ tremendous curiosity right, so tell us a little bit you know, what are some of the guiding principles you have as a company, how are you attracting those kind of people that are adding value to share a little bit of that if you could.
Ravi: So there’s one thing about the culture I’d like to emphasize, is that the customer is always right, so we have to, at the end of the day make our customer happy and satisfied and delighted so that is if there’s no one thing again I could point out is how we want to build and grow this company, if the customer is satisfied generally good things happen, you know, if you’re a consumer company you make your consumers happy, by building really great products that solve their you know, problems, if we’re an enterprise software company like us you make your enterprise customers happy by making their end customers happy, so that is probably the main point of view for us from a cultural standpoint.
Ravi: You can never go wrong by being customer-centric, I appreciate the time you’ve given us do I have time for one final question Carlos, I see that we’re running out of time.
Carlos: Well, yeah, I mean I guess we can stretch a little bit so yeah, go, go right ahead.
Tullio: One question we always ask our guest towards the end especially co-founders and founders, I mean what’s this journey been like for you, any words of wisdom and anyone who’s sitting working for a big company like Walmart or other thinking like I got an idea, you know, any words or ways to going to share with them on what they should do?
Ravi: I think it is, you know, persistence is really important I mean if you have an idea you know, it shouldn’t be driven by what technology can do, but what you know what problem you’re trying to solve for the customer, and how can you solve it using technology so that is one thing that I myself have heard you know, learnt the hard way, not to be technology driven but by being driven by solving real-world problems, solving the pain points of your customers, os if there’s one thing you know, I’d like to say in terms of you know, running a successful company going back to the point of view of being customer centric is are you solving right problems you know, in the right manner using the right technology.
Tullio: Great words of advice, Pris back to you.
Priscila: Thank you Ravi for your time, we really enjoyed this conversation, thank you for introducing us to Passage Ai and its culture as well, thank you everyone as well for tuning in and we’ll see you in another episode of DojoLive connecting tech experts like you.
Carlos: Actually we have a double whammy next week, we have two another to a Monday, we have one and on Wednesday we have another one, we still don’t have the topics yet but with the only thing we know is that we it’s we’re gonna be speaking with Matthew Bellows the founder of Yesware and KP Naiduo senior vice president of engineering and CTO at Amava he’s a former sun microsystems engineer so that should be interesting as well, again when we have the topics they’ll be up there on the website and DojoLive website, which is 192.168.100.8/dojo, and then well thank you for allowing me to do this little, little announcement for next week, and that’s all I have Pris, so thank you Tullio, thank you Priscila, and of course our guest, Thank you Ravi for having being with us here today on DojoLive, thank you everyone and see you next time.
