 Okay, thanks a lot. Yeah, my name is Andreas. So I think we have more expats today here than locals. Kind of interesting. So as you can read, my title is ReInvent. This is actually the company motto. So in case you don't know why ReInvent, you will find it out in there, but it's basically obviously about something is working already, but you're completely reinventing, dismantling it, starting from scratch, right? Or improving it in order to do something better. So just a quick snapshot of Numeroys, my company. Lots of stuff there that you've probably never heard about. So we're concentrating on the telco industry, Southeast Asia only, doing something which is called customer life cycle management. Everything about getting customers in. So acquisition, keeping a customer, letting them spend more with their mobile and preventing them from walking away. That's what we do. And largely lately as well, going into the direction of data as a service. And for example, there is one application which is the last one with the links up here which was been launched in February last year in Malaysia and Selcom, is a gamify loyalty program. And I come to that because that's gonna be basically the business setup that I'm talking about, but don't be too worried. There's not too much business there. There is still some technology, even though we are not going into the code. Because of the 20 minutes limit, I will do the demo of that too, actually next time when the next talk.js is. Okay, you probably never see me like that because usually I come like this to talk.js. And actually quite frankly, it's a bit older picture. That was the official picture at Selcom where I was working last time before I actually made the jump into something that I always wanted to do in my whole professional career. So before that, I worked for a couple, I always said I worked for every company that was back then number one in their business. That was also Citibank, not there. And then I somehow, Vodafone lost number one position to China Mobile and from there on it just went down. I went to Selcom, which was number two in the market. And now even I'm just my own company, right? Okay, so good actually that Greg had a presentation before. Not only I think it got everybody exercise, but what we're talking about here has a bit to do with games. In the topic that the underlying principle why we are reinventing loyalty or especially telco loyalty in that sense here is we wanna apply a bit more fun to it, right? As you probably know, all the loyalty programs are pretty boring. You collect some points, you get stamps or whatever, right? You get miles when you fly, but you actually never know how much you have. You never know what you're gonna get with it, right? Everything is completely intransparent. And then you know that you go there and you order and then you say like, oh shit, I forgot my car. Can I get the stamp? No, you can't, right? So it's completely intransparent. And this is actually one of my team members, my ex team members I have to say, Rimey. So all the pictures are actually from my team. This is how it works. The first step is we all have already an intrinsic motivation to use the mobile, right? We wanna stay in touch with people. The fear of being left out. It's a communicator. It's organizing our whole life. We play games on it as we saw, right? Basically who can live without the mobile phone? We leave our purse at home. We leave our keys at home. Everything, we don't return. Mobile phone is missing, we go back, okay? That's how it is. Sometimes partnerships are being negatively affected by, I always see that when I'm in a restaurant and I see the two are not communicating directly. They're just maybe chatting with each other. I don't know if you've seen that. So the second step is we're all excited about tracking progress, right? That's how we've been trained, especially here in Singapore. I think everybody's been heavily been trained through school on tracking progress. Your parents are gonna give you direct feedback about the grades that you bring home, right? The third one is that feedback actually motivates us to do more, right? We wanna get good feedback, obviously. Then we're more motivated, bad feedback. Usually pushes people down at a certain kind of right limit is also motivating. The fourth one is we wanna be rewarded when we reach certain kind of milestones, right? Think about a game where there's the final end boss, you beat it and the screen comes and it tells you all about the loot drops that you're gonna get. That's great, right? So you wanna get or you unlock a new level and now you have better weapons or better abilities. So the milestones is very important in that sense as well. And the fifth one is up here that as we go through this as a circle, we continuously go through the same kind of actions, right? Our motivation is actually gonna be heavily increased. That's very important and games are using that on and on and on to really bring you into something which is called flow. By a Czech psychologist, I can't pronounce the name otherwise I wouldn't name it as something like Chik-Chen, whatever, Mihaly, right? Chris Chik says Mihaly. Okay, thank you. Thank you for that, thank you for that. That's definitely that's a bonus. I think you get a badge for that, right? He worked at my mighty media lab, I guess. Yeah, he's, by the way, still living. Interesting life story. He was been, he's Jewish and he was been, unfortunately, by my former ancestors, he was being forced to move out of the country to kill this whole family, so that's a bad side of it. The thing is that actually as we spiral onwards, right? We're growing interest and yeah, higher level of achievement. The things are getting more and more difficult. Flow is basically about the balance between how difficult is something and how good is your ability. And if you are in that perfect little thing, you forget everything around it and you play for hours until you realize the birds have been singing and it's actually morning. So this is the underlying principle and idea that we use behind it. Some of you might have heard about the topic that is called gamification, but as it's been a bit, has a bad connotation to it, I'm not using it too much, but that's applying game mechanics to something which is not a game in that sense. So to show you how we do that at CellCon, so first of all is an opt-in program, right? Because we wanna really track that you are in there. The second thing is we have a mobile app that's actually just the dummy that we built back then. It looks way more pretty before you say anything. So here you get direct feedback about, okay, there's a leveling system, you reach level, you are at 75% to get to the next level and you have an overview of the rewards that you have. Rewards by the way are the ones that you like the most which is minutes, data, SMS, VES, right? Everything that you use mostly. I hate it when people are giving me a reward and I don't even want it. I don't want a voucher maybe for McDonald's because I'm a Burger King guy, or maybe I'm vegetarian, there's nothing there, right? So then the next thing is you come up, you get the remaining 25%, you unlock a level and you get a new reward which in this case you get free data on the weekend and you have to see that all these things happen basically if you don't look into your mobile phone, you don't check the app, these things happen completely randomly to you. So you do an action, right? You make a call and you finish the call and then you get feedback all of a sudden that you unlocked something and out of the blue you get a reward and that's so much different than you're flying on an airplane, you get out and you don't even realize that now you have a silver status, whatever it's worth, right? Then the next level is okay, you can use that on Google or whatever to serve around the web and as you are using it, you may be unlocking an achievement, a batch that we had earlier on, right? Which can be in this case the social B, right? Because you did certain kind of actions there and you hit actually 500 megabyte of data during the weekend. You didn't even know that exists. For those of you that still know Foursquare, the company by the way still exists amazingly, right? They had all these badges and nobody in the beginning knew these badges. I looked them up on a website and then I did all these kind of stupid stuff like going on to a children's playground three or four times checking in in order to just to get the batch, right? Stupid, same principle here. The batches are actually not known to the people. The next one is you have something called personal best. So you didn't even know you have your weekly personal best achievement and data because you manage 1.2 gig. Normally it's like, oh my God, I only have 1.5 because Sintel slash Sahab slash M1 doesn't give me much more, right? But in this case, actually we're recording it and we're turning it into something positive because we are again giving you something for free. And the last level is then, it's called a challenge. Same principles, all of a sudden something pops up on your screen and says the challenge for today is purchase a daily broadband to make it three days in a row. So you already did that for two days. We are knowing that, we're incentivizing you to do it a bit more. That's how the company makes money and that's how I make money by actually telling telcos to do it this way. So everything was perfect. We launched it, like I said, in February after, I think it took me four years to invent this whole stuff with a large team. We have over one million customer I'm still saying we, but yeah. Cellcom. Last year in September, we won this award for the best loyalty program. So everything was being fine, right? There's no reason to do anything different. It works, it runs. So the question is, so why reinvent that whole thing? And for that, I have to take you now to the technical side and talk a little bit about how it was being implemented in that telco and you will probably say like OMG, right? But you have to also understand that telcos traditionally have a lot of legacy systems. And even opening, the CFO opening is purse in order to invest something there is very, very difficult. So we had something, a loyalty platform which is on a system called Pcare. Pcare is for customer service. So they misuse the legacy system to program basically that layer that is talking to all these channels here. Everything, the logic behind it, the levels that you can reach and all that was being programmed in a system that wasn't even built for that. Big trouble number one, okay. Second one, right? They didn't know back then, okay, where are we gonna store all these data, right? We're talking about a huge amount of data here. So for example, an operator that I work in Vietnam right now, they are generating 11 billion records per month. So that's a bit bigger kind of thing to look at. You can't just do that anymore on your laptop, right? So they choose Oracle to store everything on that, a relational database in the usual setup. We bought a campaign management system, Unica, which in my opinion is, sorry, is one of the best in the world. And I'm not sure, last time a couple of people were at a data science conference, a meetup. Somebody talked about an IBM product called Streams, right? It's a bit similar to Spark on Hadoop. So it's very, very efficient. Programmed in C++ can look at streams of data, high, high amount of data. NGIN for those that are not familiar is Next Generation IN Intelligent Network that is basically making sure that you not only can use the services when your mobile phone, but you're also been built for it, okay? And that should be a base station, which is generating data. One of the other problems that we had was T minus 15 minutes. So whatever you generated with your action, it took already 15 minutes until it came in here in total, depending on the load that we had, it was between 20 and 25 minutes between the action, finishing the call, and the SMS that you receive that you got a reward for it. In the meantime, you were saying like, why did I get it? I just made so many actions in between. So it wasn't really clear for what you got it. So that led to three main problems, which is the reason why I thought the reinvent was necessary. Number one is this whole thing was being hard coded. Hard coded in a system that wasn't being used for it, right? It's a customer care system where you're basically tracing complaints and all that. Every single change request costs them now over 100,000 ringgit. So that's roughly 40,000 sing dollar. Easily over every little bit. You wanna change the level because you think it's not high enough or it's too high. It's gonna be another 100,000. And this is just a small little changes. Why? Because everything has been hard coded. The system wasn't been set up to be changed. You know how IT is, right? IT says, well, wait a minute, you gave us technical requirements. We programmed exactly what you wrote in here. It will be said, yeah, but the world is evolving or competitors are actually doing something, prices go down and now we wanna change something. They say like, well, you didn't write it in there, right? In your business requirement. And you didn't read the technical requirements, right? So second thing is, as you probably know, an oracle or a relational database in general doesn't scale very good. In scale, I told you before, right? If you're talking about 11 billion records per month and that's obviously not the biggest telco in the world. They only have 20, 25 million prepaid customers. It scales a bit bad and it costs a shitload of money to scale up, right? And you look into that, I once was invited in Vodafone, Germany to see the server farm of our BI system and the room was being three or four times bigger than this one and this was just the BI, crazy, right? And the third problem and that was where I had the biggest problem with is there was no real time, like I said, about 20 to 25 minutes and that topic of no instant gratification was very, very important. That was let down the whole user experience. The whole idea was we are being used to that. Instant gratification is what you want right now. Just remember a game and then 15 minutes later it tells you you level up. And in the meantime, you play on the old level and you get beaten and you die and you respawn, you die, you respawn. It would be completely pissed off, right? It wouldn't work. So because these three problems were being massive and by the way, they still have these problems and I definitely don't want to go there because they have a lot more problems with their IT. They just did a whole revamping for the last two years. Run that project against the wall. A lot of big companies, including my ex-company, Accenture just tried and failed, right? So they have more than that as a problem. So first of all, I want to address that problem on the right and I choose couch base. So probably some will say, okay, why couch base? Why not MongoDB, for example, right? Because everybody talks, first thing you talk about in no SQL solution, everybody says MongoDB. Why is that? In my opinion, because MongoDB makes the best marketing in the market, right? They got a lot of shitload of money from investors and they're just gonna spend it very, very brilliantly. But in my opinion, couch base, and there is a couple of reports, I can show that probably next time, independent reports. As you scale up, as we're talking maybe about 20,000, 25,000 connections per second, MongoDB goes down. Couch base still flies and goes over 100,000 and it's the only one that can do that, right? So if you're really talking about large scale and if we talk about 20 million, 50 million or 10 million mobile phone users, you can't direct or you can't tell them guys, you can't play like, sorry, Greg's presentation earlier on who says like, sorry, only six people allowed, the rest has to wait. That doesn't work, right? You make a phone call, you make an SMS whenever you want. And when there is an important event like New Year's Eve or Chinese New Year or whatever and everybody wants to do it at the same time, guess what, give some trouble. So how do you scale up? That one scales up quite well. And not only that, I think it's a lot cheaper to scale it up, right? Last time we were at Amazon services, right? Is a perfect example how you can easily scale things up. Second problem was everything down here. So my problem with the latency or you can't even talk about latency anymore when you talk about 25 minutes. It's not milliseconds, right, it's minutes. So the first thing I did here is I use actually talent for data integration. They're amazingly fast. It's like crazy, it's a Java program but still amazingly fast because what I just need is from that huge record, usually we talk about here, a record of 150 to 200 variables that comes in. So there's a lot of stuff in there that is technically needed, which I don't need. I only need a couple of variables, slightly over 10, nothing more. So something's to go in here, do some ETL processing or ELT processing and then fire the REST via an API call into something that we all know, which is finally why I'm talking at this conference here, right, in this meetup, which is Node.js. Okay, so again for me, I'm not a front-end programmer. Actually, I started programming when I started in university, but in the meantime I made a career on actually not programming at all. So again, why the choice of Node.js? Because as the whole process, I wanna be as fast as possible. And it doesn't matter actually, there is no bonus points being one in here. If I get a response out, you drop your call and in two or three or five seconds or 10 seconds. So when you cut it from 10 seconds to eight seconds, I'm not Google, right? I'm not obsessed by a millisecond or 10 milliseconds being faster because these guys just scale and then the billions are coming in. But where it's important and where Node.js needs to be highly performant is when we're again talking about 100,000 users that are maybe concurrently been working there, then it becomes a problem, right? That body can easily do that. But what kind of programming language can handle that many requests coming in, right? We're talking about API requests that are going in and also going out into the campaign management solution, which by the way is the easy setup because the whole provisioning, so making sure that you get something, for example, in some of the cases, you get a free weekend or you get maybe 50 minutes for free, something like that. It's all being handled by the campaign management system which is in place anyway and the rest is just outside of my problem, right? That can handle it. The massive of data is just been flowing in here and if you look into this, this is maybe one out of 5,000, one out of 10,000 where there is an action being triggered, right? I'm not sure. I actually never did the calculation over there. So that leaves one problem over here which is that problem of using a legacy system, misusing a system for something else that there is. Sorry, just two comments on here. So one is actually less than a millisecond right now. So I will demonstrate that next time on this machine actually. So I will see, right now I have the problem that I wrote another program that is pushing the data from this computer to my Node.js program and then pushes it into couch base and back and forth, right? The problem is that this program runs to a limit. It's not a problem of couch base, not a problem of Node.js. The problem is the program that is basically fetching, throwing the data there. So I will change that and I experiment already with talent and talent is so much faster. So I guess talent will somewhat bring it down or bring actually my notebook down, right? The second thing is with this we can do instant gratification which is something that especially the youth and this program has been designed especially with the youth in mind which is in most Asian countries is the largest population group. They are just being born with instant gratification, right? None of them has any kind of patience, yeah. So largely we take that part out and I use the software that I used for a couple of years that's why. So like I said, I'm not a front end programmer. You're probably gonna say why you do that why you're not using anything on front end. I didn't know anything about that. Not sure if it's ideal. It's called the SOJO. It's quite good to build basically a Windows environment. You can develop in parallel on Mac, on Windows and for Linux. It's just out of the box. They now do actually Apple iOS as well. And the thing that I developed over here is that this platform becomes highly parameterized which is the big, big difference. So you just go in there and you change, for example, a level like I said earlier on which is gonna be a big change request in the old model. All this thing does is it's just gonna talk to couch base and changes a value in here and enables that, right? And immediately everything has been changed. It doesn't change anything here because it queries against this value. It's all it does. Okay, and the change requests are gonna be a lot less over there. And the rest is just the platform has been there to test things, to analyze things, to look at reports because for example, your CFO wants to know if the program has been successful. The people want to know how much you actually spend on your rewards. Are you getting money with it? How many people are inside the loyalty program? How many people are actually been gone out? Why have they gone out and all that? So all this kind of logic is in here. The reporting and analysis system is gonna be in here. It wasn't even in the old system. There is no report, nothing at all, right? Because like I said, it's not been built. They don't have a BI being set on top of it. That's the biggest problem. Okay, just my announcement. So if you're looking like that or not, if you have a computer like that, I actually had one Atari. So I'm looking for somebody who can help on the whole programming side. Because that's like I said, I can't handle both. I'm doing the whole consulting of it, which brings the money in. I need somebody where I can put that money in, which is a programmer who can build the MVP further up. And if you wanna contact me, just keep in mind numero, you can look up on LinkedIn, write on email address, mobile number, gonna sit over there if you wanna approach me later on. If you're interested, just open to that. And thanks for the opportunity. I was basically going to talk to you for a year now and I thought it's finally some time that I give back something to the community. You guys been awesome, I learned a lot here. And see you next time then on the, hopefully the live demo works out. Okay, thanks a lot. Any questions? Sorry, I didn't wanna cut down on questions, but okay, thanks.