 So good morning, thank you for making it to the final day. I'm hoping this will be interesting. I don't have any race sinks. But as they were saying, some of what I'm doing is very different than a lot of the keynote speakers that have been out here. I work for an ad agency for those that don't know what Ogilvy is. We're very large. We have 450 offices and 163 cities. We have 800 developers worldwide. I'm from the New York office. And what I primarily focus on is what is considered creative technology and a combination of machine learning. So what we do spans everything from mobile apps to websites to event installations. Some of what I'm going to show you today is some of the projects that I've done probably in the last six months or so. We work for a huge variety of clients. And ultimately, what our goal is is to make brands matter. So several of these examples are going to be specifically about how we try to amplify brands to get the words out, make people aware of what's going on. We're constantly trying to stay competitive, just like our clients are. A lot of the tools that we build are specifically around getting the competitive advantage as quickly as possible. Like most of you, we have a lot of challenges. My primary one is time. I never have enough. Also rarely have enough money. So we're looking for efficiency in every way that we can. And Cloud Foundry is kind of key to that for me and my team. So I'm going to show you some examples here. This first one is something that we built for IBM from Mobile World Congress. There were actually several pieces to this installation. We had to build four apps in under four weeks. The main ones are, I'll show you some photos here. We did a giant sculpture that was based on the work of Antonio Gaudi. We used Watson kind of upfront to kind of pull out some of Gaudi's characteristics and then gave that to an architect to build the sculpture. On the event, the sculpture is actually reading social data real time and then moving in response to that. With that, there's a dashboard, which doesn't look particularly exciting. But it was kind of key because we had people on site at the event that were monitoring social activity and then changing it on the fly so that the sculpture was reflecting things that people were talking about on site. So this particular display is a real time view of what's feeding the sculpture. And it moves. Not that you can see that here. But it would give the admins the ability to kind of preview whether things were moving enough for it to be visually interesting out on the floor. And here's another one of the views. It's kind of basic, but the idea is that the data is being pulled every two minutes. And we wrote custom API endpoints that is being powered by this particular application and it's being read by the winches in the sculpture as well as a kiosk that is explaining to users what's going on. So this is a shot of that kiosk. As an end user, you could come up and dive into any one of these particular rings, kind of see the tweets that are feeding it. And then the next major installation that we did, some of you may have seen this or not, it's a cognitive dress. We originally created this for the Met Gala last year. But for Mobile World Congress, we ended up having to rebuild the entire application behind this thing, ironically. Mobile World Congress has horrible Wi-Fi. So the original dress had a controller that operated purely on Wi-Fi, the person's wearing it. But because of the event, we ended up having to change the microcontroller to Arduino so that we could run it hardwired. And in doing so, the application that we originally wrote, the firmware was no longer supported on the Arduino so we had to completely rewrite it. And in collaboration with all of that, the clients wanted us to add a whole new piece where we built an iPad app where people could come up, put in their Twitter handle, and then change the dress on the fly, on site. So here's kind of a quick view of all the different pieces that were involved. Most of these applications were hosted in Bluemix. I've tried to keep this pretty simple so that you can see it in the few seconds that's on the screen. Essentially, we were pushing all of our API endpoints into Cloud Int and then updating it on the fly. So both the sculpture processing app and the dress were pointed to Cloud Int and just reading it, anytime anything would change, they would update in either change color or move in response. Here's a kind of final shot of what the event space looked like. And then my next case study is, it's basically a CMS that we built. The name is called Phonic. I did not name it. Every time I see it, it reminds me of Hooked on Phonics. But this particular example is interesting for me personally because of really the background behind why I built it. Right before I started this particular project, we had a site where we were in the middle of a build, we had three weeks, we thought, oh, things are great, we're rolling into the last week. And then two days before launch, the clients and creative team said, oh, that was cute. The entire content structure has changed, the design's changed, you have to rebuild everything and you have two days, date's not moving. So my team essentially spent the next 49 hours awake, working like crazy, trying to rebuild JSON data files by hand, not something I advise. And then the following weekend, I was like, that's it, never again. You gotta find a way to make this more efficient. And so over the course of that weekend, essentially Phonic was born. And it's a CMS that allows us to change all of the content types, all of the templates, everything on the fly instantaneously. It is completely separate from the display. So we have a separate front end application from the CMS that allows us to create templates, to whatever the client specs are. So essentially, here's one of the screens of it. The CMS interface itself is pretty straightforward and simple because it's designed to allow us to create custom content models for every site. The data is completely separated from the display, as I mentioned. And essentially what we've done is we've used this platform over time to massively speed up our development. By taking content management off the developer's hands and into the people that are actually managing the content. So we can just purely focus on building the site and doing what we need to adjust. It's set up, we've kind of evolved it over the last few years so that we can handle all kinds of content tagging. We can change skins on the site based on campaign, what banner they've come from, what audience they are. We've tried to make it as flexible as possible. It has made it so that instead of taking us, I don't know, about a week and a half to build a site, we can do it in about four hours. This is kind of the high level view of it. There's a bunch of different applications that go into it. We throw them in and out as needed for a particular client site. We've set up the site app so that we have the flexibility of either running it dynamically so that it's pulling data real time and kind of rendering it on the fly. Or we can generate a static site depending on what the clients need. There are several pieces here. Most of them are in Bluemix, some are in Cloud, some are in software just for efficiency's sake. We use the poll and survey DB off and on. It's not for all projects, but it's good when we have something like a poll or we're doing benchmark surveys or things like that. Yeah, so we've ended up rolling this particular application out across 138 instances in 15 countries for about 28 different clients. We're still using it, we're looking at making it open source when I have some spare time. So hopefully other people can leverage it as well. This last one is actually one of my love children. It's still in active development. Our company has ended up creating a new behavioral science group as one of the divisions in New York. And we're starting to create a bunch of strategic apps that help support their initiatives. So a lot of them are around audience analysis, trying to identify corporate personalities and consumer group personalities so that we can try to tell them how they should be talking to each other to make our marketing more efficient. And so this beautiful little chart is kind of an example of how we're trying to break people up, right? And it's really about dividing people into what matters to them and what they'll respond to, what are their values, what are their risks, right? Because if you say the wrong thing to someone, you'll turn them off and they will probably never come back to your brand. So we map people like this and corporations. So like here's an example of what a corporate personality might look at, right? Not that you need to read this, but here's an example of what an in-chart might look at look like, right? So a company overall might have a particular persona that it's projecting via internet, you know, their PR and other messaging and the individual brands within that may have their own personality. What we're trying to do is figure out, okay, you're doing this particular campaign or sponsorship, what brand message are you trying to send out and is it effective in aligning with that particular group? So in this case, I've nullified who the brand is for sanity's sake, but they were trying to reach golf fans at a particular event and their message was way off. They were not in the same place and they kept wondering, well, why is my message not working with these people? It's the same message I've used everywhere, it's successful in other medias and we're like, meh, not so much in this one. So using this information, we then pivoted back to them and said, hey, change your messaging so that you're saying this and you get a lot more traction. And making that change meant that they got about 33% more engagement, which is good considering they were about two originally. This suite of projects, right, this is very high level, I don't have all the interconnections because it would look like a giant spider web, but we're kind of doing a long-term roadmap where we're trying to build various pieces and tools that are all talking to each other, all leveraging different pieces of machine learning. So in this case, our original core app is in Bluemix. We're using a bunch of their services and we're looking at building new apps in Google to try to take advantage of some of their machine learning predictive models. We're kind of, as you see, using a lot of different services and looking at trying to roll this out probably in the next six months or so, if we can. This is in the early phases, it's still evolving. One of the things that I'm a huge fan of in Cloud Foundry is that I can do experiments and see what works across each platform and figure out if it works before I stick the rest of the team on it. So, thank you. That's my tip. Thank you. Thank you.