 Live from Las Vegas, Nevada, it's theCUBE. Covering IBM World of Watson 2016. Brought to you by IBM. Now, here are your hosts. John Furrier and Dave Vellante. Okay, welcome back everyone. We are here live at the Mandalay Bay for the IBM World of Watson. This is theCUBE. We go out to the events and extract the signal from the noise. I'm John Furrier with my co-host Dave Vellante. Our next guest is Andre Koenig, co-founder of Open Topic. Startup, couple years old using Watson. Welcome to theCUBE. Excited to be here, thanks. So, this is a great example of the innovation that IBM is promoting, or trying to promote, which is, you know, get people using the technology, not just big companies, but startups. What's your story with IBM? Tell us how you got involved. Absolutely, and we got a little booth right here in the innovation concourse, which is very exciting. We started with IBM at the beginning of last year, a little over a year and a half ago, as an IBM Watson ecosystem partner. We were in the digital marketing business. We were a SaaS platform for marketers, and we started to see a need for cognitive technologies, and IBM Watson, that's only a few blocks away from our office in Manhattan, came to the rescue, and we started to build, and it was amazing. We saw Thomas Friedman yesterday. He talked about the speed of acceleration that's getting too fast for too many people, and it's been the same for us with IBM. Today we're here with our booth, our team of four people speaking at theCUBE, getting drinks with Ginny tonight, the CEO. So very exciting journey. So we're gonna be having those drinks? At the concert, right? Still don't know the demand, but you're invited. Congratulations. Well, let's talk about how you guys are operating now. You venture back, self-funded, customer-funded. How would you guys, how'd you guys get started? Yeah, we're on our third live, so to speak. The first version of Open Topic was founded by Sophie Renault, a French woman who comes out of the agency business and was very successful with L'Agence, a French digital marketing agency, and she wanted to build a content business for specific devices that came into the market, such as the Kindle, and she was very successful with that. She had been successful in her previous life, so didn't take it that much farther. Then my good friend and co-founder, Christian York, also originally from Germany, as myself, I've been here for 15 years, took over Open Topic. I was really able to scale it very nicely, but still in the content space, which was exciting, but not what we dreamt of as entrepreneurs. And he asked me to join about two years ago when we took a little bit of a reset and we saw this need from marketers that had really big pain points. How do you understand customers? How do you engage customers? And how do you drive them reliably and consistently to a result? And that requires technology and cognitive technology, and that's how we got to cognitive digital marketing today. So how are you deploying Watson? Because classic, this is actually a great, good case. Most of us will build our own, but we're now in an API economy. Why build it if you can get it on either buy it and or get it open source? Tom Friedman said at the speed of acceleration, we are building today things within a week that a year ago we couldn't even imagine. And when I say building them, we're not only building them, we're putting them into the hands of customers and they work. The customer use them, the customer get value out of it. And we're able to do that because of software, of Bluemix, disintegration with all the different IBM Watson APIs and understanding the use case around it. I think keeping it simple, not being too ambitious and stacking API over API technology over technology, but really understanding the power of specific applications. Do you wire APIs together so you get the benefit of almost building the product? You're assembling pre-existing technologies. Your innovation is, I think, what Bob Pachiano saw, the inside economy, you guys are doing. That becomes our IP and differentiator. How do you assemble these different APIs? How do you package them? And how do you make it valuable to the customer? Absolutely. So how does it work? Take us through the platform. You say you have basically a focused use case. What is that use case and how does it all work? Yep, absolutely. We work with companies such as Time Inc, The Economist in the Publishing Space, BBBA in Banking, a large Spanish bank, Unilever, so brands. These companies come to us whenever they want to do one of three things. Understand that target audience better. Understand what stories, content images, video campaign data to use to engage and ultimately convert this audience and how to automate and scale this process. The big buzzword in marketing today is personalization or marketing in the moment. We are all so distracted by the many different devices that we have the overload of information. So if you're marketing to a consumer online, you're basically lost already by the point that this customer comes to your website. There's a forest research study from this year. 57% of all users already made a buying decision when they first come to your website, be it an Amazon, Unilever, or whatever else it might be. So you need to be in the moment in front of these customers continuously with the right assets. To be able to do that, you need to understand them at a very behavioral cognitive level. And you need to same for your images, your videos. So specifically, how do you deploy Watson? I want to kind of get into the hood. And what's your development process? The mousey startups are always agile. Cloud's great. Cloud native. It's good. So you're in the cloud. I get that. But with Watson specifically, how do you plug into that? Yep, absolutely. We have a cloud-based user interface that customers subscribe to on a monthly basis, hosted on a soft layer. Our own infrastructure then makes the different API calls for things such as sentiment analysis, personality analysis, concepts insights on the customer side, and the natural language processing and taxonomy classification on the content and story side, as well as image recognition, video recognition, and so forth. It really depends on the customer. It's very modular. Every customer can turn on or turn off whatever functionality they want. So if I'm a publisher, what do I do? I sign up for your SaaS platform and then what? How does your SaaS platform get to my data? Yep, a publisher. We see two use cases. One of them is inspiring editorial. It's becoming tougher and tougher for journalists and editors to write new stories, really great, interesting news stories. And what they come to us for is data. Give us data on specific trends, people, celebrities, topics, business trends, so that they can find new angles, new perspectives, new data points that make the story interesting. So they would get an audience inside Stashport and we then have a service division where we help them brainstorm through that data and level by level by level start to dig deeper. Where does that data come from? That data comes from over 140,000 online sources that we've aggregated ourselves over these three lives that we've had over the last five years. So you're getting the data source from a fewer database. Does Watson have anything that they're adding for data? Or is it? Absolutely. So we have our own 144,000 sources. We also call the Watson News API to get additional news and online content. Many times, especially in the case of publishing, the publisher will add their own content because they tell us, we have all this amazing content. We produce content all day long. So we could be a customer. Absolutely, please. How do we understand this content? What do we do with it next? So we add that to it. So it's a superset of one of your clients. It's a superset of their data, right? OK, when you're using that as a proxy, now what kind of segmentation do you do? Do they plug in a persona that they're trying to target? Absolutely. You can start with a theoretical buyer persona. We've integrated with a lot of personalization platforms, Bluekai from Oracle, Lytex that gives you personal information. That's code you wrote. So what did you guys develop? So I'm trying to separate up because I think the audience would love this because stars love to go faster, as Thomas Friedman said. So you have Watson. That's API-based. What are you guys actually coding on your end? We are coding the infrastructure that allows you to call these different APIs for the specific use case to answer your question. You get an audience dashboard, for example, in the case of that publisher that wants to inspire editorial. They need visualization. They need insightful data. They need to make it actionable. They need to share it with their teams. They need to push it into social media on landing pages and analytics widgets and so forth. So that's what we provide on that. You write the visualization software on your end, the UI. Absolutely. You do the dashboarding, basically, and the APIs are. So Watson's the API interface. That's it. There's no visualization from them. So I might come to you and say, OK, I want to target, let's say, developers. OK, and then you would allow me to presume me it's self-service. But no, you said you've got a services organization as well. So you might hold my hand a little bit, teach me how to fish, how to target those developers, and then understand what the fish are eating, if you will. Absolutely, the teaching how to fish is an important element of it, especially with cognitive technology that is still so new. Back to personalization, we recently, working with a large bank, took them from five target audiences to almost 500 target audiences. So now they can up and cross sell the credit cards, mortgages, loans to 100 times as defined and specific target audiences, all in an automated process. We did that through data that they had, product data, customer data. We also did that through brainstorming. How do we approach this by a regional segmentation, a product segmentation, and so forth? So it's the two combined, data and that teaching how to fish. And the engagement model is engaging with some offer, or is it could be anything, could be some kind of interaction or chat or what is it? Absolutely, in that case, they have a next best product recommendation, which we mapped to our asset recommendation and the 500 target audiences. So now you're going to get a highly specific offer that fits your persona. And do you surveil my data as a client? Can I say, OK, I want to make sure that my data is represented in your 40,000 corpus? Not yet. We tend to stay away from that. We work a lot with health care providers and the Financial Services Bank so that gets a regulatory and compliance point of view difficult. It's really more about once you come to us, a social media channel, a website, a newsletter, understanding you, tracking you at a behavioral level, clustering you in what other people that are like you at the smallest level possible. We want small groups, small target groups, and then serving that up. But we don't track you interventionally. So you have an arms length from my corpus of data. You have a large data set, and then you can subset that, target that, however I program or I define. Absolutely, that could be a social media campaign. It could be a newsletter campaign. It could be dynamic websites. So working with the economist, their challenge is how do we drive subscribers to the economist.com and print also for those of us that still read that. So they want to serve up my personal website that economist.com as soon as you come visit. And in that case, we know who you are because you're probably locked in already. But it becomes a dynamic serving up of the best possible content just for you. And the whatever, machine learning, cognitive, all that stuff comes from the Watson API. The machine learning is something that we wrote ourselves. So once you click on the subscribe to the economist button, the subscribe to the newsletter button, the share on Twitter button, we track that information and we're able to make the customer understanding and the stories understanding smarter and you more successful next time around. Talk about your technological lab that you have microservices, you say eight microservices. Now when we talk about microservices, we always talk about like Docker and microservices. Is that the same kind of thing? Is it APIs or they truly microservices? Can you explain what that is and how that relates to, because this seems really cool because now you've got microservices being applied to digital marketing. Yeah, we're very proud of our technology lab and just incorporated a legal entity in Poland where it's based. Five years ago when Sophie started the business, she hired freelancers in Poland as people and I start up typically do. They now all have come in-house. We have a team there of developers within this technology lab entity and they've built these restful APIs as microservices. So we basically take a lot of the learnings that we get from Watson, a lot of the data that we get from our customers and we try to customize it in these APIs, microservices. So they're recipes or best practices packaged up as a service? Absolutely, as a scraper, a content scraper. How do you parse websites? How do you identify language, author? We have our own sentiment analysis, these types of things and the reason we do that is because we always want to use the next and newest available Watson API. We can use all of them, that would be too expensive for us. So we try to learn, build ourself and then move on to the next. So you got connectors that BlueKa you mentioned. Did you guys write those? No. That comes with that? That came from the client, yes. From Watson? From our client. Okay, got it. So they have their own connectors? Yes. You just kind of build it into their API? Yeah, since we are API based and they have API end points that becomes a very simple integration which is another strength of ours. Awesome. So what's it like being a startup working with Watson? Most startups like to go fast and loose. You guys have a little bit of history, three years. You know, kind of navigating out to here without a technology lab. You got some cutting edge cloud stuff. What's the impression working with IBM? It's amazing. I was speaking to an IBM or yesterday and she told me that IBM has 83,000 different products. And if you just think about that, that's, I can't imagine that. And standing at our little booth over there, we just have people after people coming to us from all over the globe at IBM, all different business lines, all different product lines. I said, we heard about you through a tweet, through a case study. We're on the IBM website since two weeks ago, a couple video interviews and they come. This is really cool stuff. I need that for my utility customer. I need that for IBM connections. For example, the messaging platform that we recently integrated with, or IBM Workspaces that was announced yesterday. So for a startup, it's amazing because IBM is this huge playground without really any boundaries. But it's also up to you to go out there, be professional about it, be proactive about it, and deliver on what you say. If you don't deliver, be it on your customer promises or your partner promises, then you're not going to do well. So a couple years ago, when you talked to a developer about Watson APIs, they said, yeah, some, you know, what do you think, what do you want? They said, I want more of them, I want them to be more mature, a little easier to work with. Fast forward to today. Did that improve, what's on the to-do list for IBM? Well, let me start with the business person because that's the biggest change that I've seen over the last two years. IBM is cool. We get calls all day long from senior executives, not only within marketing, but in general. We saw you do something with Watson and it seems to work. Please help us figure out how we can make Watson work because we want Watson. We might not really understand what it is exactly, but we want Watson because we believe only cognitive can solve our problem, whatever it might be. And that we didn't see two years or even a year ago. There's this huge demand and pull for Watson for information and IBM is the cool kid on the block for that. Yeah, and that message resonates because people can see the automation and AI, as Jeanne Romney talks about, is augmented intelligence, which means this extension of their existing stuff. Absolutely, and in my industry, a marketer shouldn't spend their time analyzing social media clicks and retweets or your Google analytics or figure out which CAD video is the next best video. That's a waste of your time. You need to be out there and thinking about your marketing strategy, your branding, engage with people. That's where your value is. So we really think about it that way, giving you that information and those assets. Andre, final question. First of all, thanks for coming on. Andre Coney, co-founder of Open Topic. Your thoughts and advice to marketers that they're watching in terms of trying to grok the idea of the cloud, understanding cloud technology in cognition, cognitive computing, real time. It seems like a complicated thing and I want to go digital but what's your advice on how they get started with IBM or you guys? And we talk to these folks all day. They get pitched by startups like Open Topic and others all day long, especially in New York where we are. My advice is simple. Start with a real problem that you have. What is a pain point within your marketing stack, within your campaigns, within your branding initiatives that you want to address? And start small. Start with a pilot. Don't try to plan for six months, for 12 months. Don't make any big investments. Work with a startup such as ours that is willing to be flexible that will cut you a little bit of slack. Don't lock you in on a long, long contract and commitment integration. Or sell you a bloated software package that you have to build an ROI. You have to use it for three years. No one's using it. I mean that's some of the hangover people are feeling right now. Absolutely. And we spoke about that in the beginning. Today we can build stuff with Watson and Bluemix in a week. So if a vendor doesn't deliver results to you within a week or two or three, they're probably not doing a good job. If there's a two, three months integration period, stay away from it. Andre, thank you. We'll see you at the concert tonight. Have a great time. Thanks for coming on theCUBE. Appreciate it. I appreciate it. Thank you. Hobnobbin with the CEO of IBM. IBM is cool, he says so. Of course we agree. We think they're cool as well. Obviously great strategy. Keynote was phenomenal. That would be a more analysis on the keynote at the wrap up segment. So check that out. Day two wrap up here. We'll be back with more live coverage here at the Mandalay Bay after this short break. I'm John Furrier with Dave Vellante. Be right back.