Megaworld Corporation's Artificial Intelligence Innovations for Safe and Secure Townships

 In this episode, Francis Adrian Viernes, CFA, MSF, CCREP, PMDSA, Chief Data Scientist and Head of Data Analytics at Megaworld Corporation, discusses how they are making waves with its innovative use of artificial intelligence (AI) to create safe and secure townships. Francis delves into Megaworld's plans for expanding AI applications beyond accident detection, such as weather forecasting, crowd monitoring, people counting, and air quality monitoring. He also shares his perspective on how Megaworld's digital transformation initiatives will shape the future of urban living in the Philippines.

Megaworld Corporation is a subsidiary of Alliance Global Group, Inc., one of the largest conglomerates in the Philippines. Megaworld is listed on the Philippine Stock Exchange, with a market capitalisation of USD 1.06B (₱59.56B).

Transcript

AIBP Intro: 0:02

The AIBP ASEAN B2B growth podcast is a series of fireside chats with business leaders in Southeast Asia focused on growth in the region. Topics discussed include business strategy, sales and marketing, enterprise technology and innovation.

Vanessa Kwan: 0:28

Hello and welcome to the ASEAN B2B growth podcast, where we sit down with individuals responsible for driving growth within their organizations here in Southeast Asia. My name is Vanessa, and I'll be your host for today, we have a very special guest joining us, Mr. Francis. Adrian Burnes, chief data scientist, head of data analytics, data science at Mega world Corporation. Mega world, founded in 1989 is a real estate company listed on the Philippines Stock Exchange boasting a market cap accident us $1.1 billion without further ado. May I invite mister Francis to give us a brief introduction of yourself, give us an idea of your background and perhaps share some personal hobbies.

Francis Adrian Viernes, Megaworld: 1:10

Okay, so thank you for that, Vanessa, and it's actually very nice to be invited in the podcast and meet you again so well, as you know, in my capacity as the chief data scientist and the head of data analytics, we create here data science models, but not just the fancy data science models. There are a lot of ways to approach data working with a big company like Mega world Corporation, and by extension, the Alliance, Global Group, which is a very large conglomerate in the Philippines, is, well, it's actually very interesting. You know, one thing that I found is the larger the company is, sometimes the simple data science models or even simple analytics sometimes does wonders for the for the for driving growth, and that's because you have the power of big data to support you. Now, let me discuss a bit why that data that I am handling is a bit more interesting. So AGI, as a company and as a conglomerate, we have many subsidiaries that are dealing with different operations. But really the core is all about real estate. And maybe, if I would like to summarize it for you, the best word might not actually be realistic. Maybe it's lifestyle. AGI Alliance, global incorporation in the Philippines, is all about lifestyle. Of course, the leader in there is Lego world Corporation building townships where each of our townships have a very different personality. So if you go here, and I would like to invite you, or anyone here visiting this podcast, you go to the Venice Grand Canal McKinley, it will give you a glimpse of like Italy, if you go to our Eastwood city, it's like our new week Hollywood, where you see the local stars having their names imprinted over the sidewalks, where with stars, very similar to Hollywood and we have, as well, other townships that are being planned. Let's say, for example, like the Newport kind of look like Macau. So if you've been there, you've seen, well, it's really the advocacy of our chairman, where, for our Chairman, there are some Filipinos who might not be able to go regularly or visit some of the places, and maybe mega world as a company and a lifestyle brand would want to bring that to the Philippines. It's really nation building. Other brands are forced. McDonald's, as you know, a very powerful in terms of merchandise. I will also have the megawatt lifestyle and hotels. And on top of the mega world hotels, you also have a different world sibling company under the AGI conglomerate, which is the travelers older of a lot of the five star hotels here, the Mario, the Sheraton, the Hilton, and recently, the hotel the pura. And we also have, emperor, the world's largest and best selling branding, right? And we also have startup builders, agile and picaro, which competes with grab, and actually many more, like infra port. So a lot of those companies are under the AGI CJ. And of course, if you look at that different data, how they talk together is a little bit interesting. So this is why I think this position has a very huge potential.

Vanessa Kwan: 4:56

Can you share a little bit more with us about how you are actually. Driving growth and innovation for mega work, with regards to the current role that you are sitting on as the chief data scientist and head of data analytics.

Francis Adrian Viernes, Megaworld: 5:09

Well, I think any data scientist has that opportunity to actually provide the growth that the the company needs, just by analyzing, for example, what strategies in the past have worked for them, which brought in more sales and which one haven't been as successful. So really, at its simplicity or at score, this is how we as data scientists drive value and and, you know what? For a big company like Corporation, and for, hopefully, a lot of conglomerates here in a cm region, you have the benefit of having a huge data set that you can analyze and you can further Submit. The benefit of that. Interestingly, for someone for mega world, who, by the way, is celebrating its 35th year, by this year, is that you'll be able to see what we call in data science, the domain shift. But the more well, maybe the simpler term is like the generational shift in terms of the behavior and how you would now want to point your company to adapt to the trends. Now the more flashy ones, which may not be as applicable to other data well, other conglomerates and other companies as we do, invent things for data science models or ai ai products, let's say, for example, the accident detection, which happily won the award from AIB. And also, we're developing some type of geospatial project as well, is we want to develop products that would, well, of course, enhance the safety of our locators, because the product is a bit different. When I say that the product's a bit different, not a lot of suppliers will rush to produce or to mass produce the AI projects that we need. Let's say, for example, unlike you know, for example, we have McDonald's right. Hunger is a basic, is a basic for food is a basic need that McDonald's can mass produce, which is, for example, burger price. And everybody needs that. So there is some point to scaling or producing a lot, but for someone who's product is the township and is which is an integrated development. And let me explain what that is. Integrated Development is like creation of a mini city where components are present. That's why it's called integrated. Integrated, meaning there's hotel, there's residential, there's malls, and they all harmoniously, work together. And because of that, that kind of product sometimes have less suppliers. So for example, compared to well, the bigger or the more the vanilla type developer, which produce a lot of buildings. You may have ai products that serve you on a building level, but because your product is integrated, the harmonious integration of all these components, that product may not necessarily be attainable, and we sometimes want to develop that so that we can enhance the safety, but we will also be moving forward with the other objectives, which is, of course, to increase operational efficiency. Then, of course, maybe in the future, increase the profits.

Vanessa Kwan: 8:58

Understand, earlier you spoke about the accident detection software that won the AIBP ASEAN enterprise innovation awards for the Philippines last year. You know, it represents cutting edge computer vision model trained using AI mainly to identify and analyze vehicular coalitions on the road. Perhaps you can share with us what inspired the creation of this software. You know, how do you see mega work utilizing data science and AI to address the unique challenges and opportunities faced within these townships in the Philippines?

Francis Adrian Viernes, Megaworld: 9:33

the opportunity, or the idea of developing the accident that is also came from analyzing that there are not major absence, but minor absence and bicycle absences in the township. And you know this is what separates us from other developers, because you are an integrated development. Integrated developer, the integrated developments have function. Realities of a mini city. And a mini city has what it has, its own fire, fire trucks, policemen for orders. So what happened is that sometimes, with the during the pandemic, a lot of people have started the spikes, the new word, creating your cities? For example, a lot of our cities here in Asia region would probably be like that. You weren't really thinking of the youth, the huge amount of bikes, right? You maybe are just so the bike lanes, they're kind of nascent or recent for a lot of it right? So that being the case, there is new there are new developments. There are new yourself, seemingly the high enough amount of accidents also particularly, maybe because that a lot of people started to also just learn. So maybe that's also one of the reason I'm just thinking right now. But so, so as an integrated developer whose job is to keep everyone safe inside your own country, we needed to find a faster way to respond, because sometimes some of those who may actually fall, let's say, for example, maybe an old person you might want to respond respond faster, because you increase the chance of making that injury less severe with faster response. And as you know, with accidents, sometimes the difference, but like difference between life and death can just mean a couple of seconds, right? If you often heard this in a lot of your movies, for example, or being advertised, or anecdotes where, if only the ambulance have been here a few seconds before we would have administered Emergency, emergency response, and he would have lived or she would have lived, right, something like that. So that, being the case, this inspired us to actually create one. Why create? Because we have search for suppliers, and no suppliers are able to provide this. Because, again, the need or the demand is very low. The leader, the demand is very low because the creator of integrated township for developments are actually few. So because there's no supply for that, why would I create a Demand just solely for that kind of product, if there's only one in the Philippines, right? So that that dilemma allowed us to take, maybe we can invent it right there's, there's a lot of the technology has gotten a little bit better, and we have very talented, even young professionals right now who are more about the work, implementing their their ideas. And maybe if we pull all of this, we'll find that solution

Vanessa Kwan: 13:07

Can you share some specific insights or like matrix from the deployment of this particular software in South wood cities that led to policy improvements, safer roads? You know what? What were some of the results that were achieved in the initial pilot.

Francis Adrian Viernes, Megaworld: 13:22

so for the accident detection software, it's able to detect when two objects are about to collide. It uses, or it predicts, the trajectory. For example, one object is going 90 degree and one object is going at 45 degree, using the speed as well, okay, these two will collide more often than not. We have a lot of false positive because, of course, if you driven in in an emerging economy, you know that they're very expert in avoiding those collisions. So it would look like in the camera that they would collide, but they would actually stop. But sometimes, if it does happen, then we are notified by an alarm. So now in the Southwood city, there were around 400 incidents of that, and we were developing here that it's all minor, it's all just bicycle accidents, so nothing really classified as a major traffic accident, right? So using this, we were able to pinpoint or locate on that CCTV. Then we were able to see what spot it actually that accident actually occurs. Of course, we couple that with the vanilla dashboard where the locations happen. So we have a heat map. That being the case, we were able to now review the scheme there and devise a new traffic flow, eliminating almost 90% of that accident act, if I'm not mistaken, around 92% so the accidents are so much new after that, because, of course, it's not. Just the detection. At the end of the day, accident detection is just accident detection. It's just the text. It's about our response. But it wouldn't have happened if the detection were able to record and notify what time and what location it is happening. And I guess for a lot of our going back, and certainly back to your first question. That is, where data scientist actually does drive or data scientists do drive the growth or produce values. It's really where you can, if you gather the data, and you can pinpoint the pain point actually, or that which you want to resolve, it's easier to tackle, because without data, what a company would be doing is to blindly, blindly think of or maybe use intuition, which, by the way, may always may may actually be correct. So some of our experienced leaders have intuitions that are so powerful that they are more powerful than the data, and that's that happens, but, but sometimes, because of the wide variety, maybe, for example, for megawatt, where you have a lot of subsidiaries, maybe it will be very helpful to really just pinpoint and act fast, because it's no secret that a larger company acts made, finds it more difficult to respond faster because of its size. But the data science, if you can pinpoint the data or the insights produced, like for example, for the accident detection, makes it a little bit faster to develop that solution.

Vanessa Kwan: 16:43

understand, you know, in such instances, when you are rolling out certain projects, certain initiatives that are driven from the data science team, how do you usually get the buy in from the management team? Is it something that you would have to present a proof of concept, convince them? And then in such instances, how would the company then decide, you know, what kind of resources are they going to prioritize and advocate? For example, in this instance, the accident detection software,

Francis Adrian Viernes, Megaworld: 17:10

okay? Well, there are many ways to actually get your proposal across, but one that I found very, quite, very powerful is actually to present what you said, the proof of concept or a prototype. I guess it's that not that uncommon. If you watch series like Shark tanks, they kind of need to see what they're investing into, and that's just being a good investor or a good management for example. So I often come to them and show a proof of concept. And then, of course, now, with the this, with the resource, okay, with the resource, it's actually now up to you, for investors, if they like the prototype, okay, how much do you need? Of course, they might have some some benchmark. And of course, doing some studies or comparing with how much it would cost by outside or maybe, for example, for the accident, how much it would cost to have to ensure all those 400 accidents. Maybe that would be a basis of a cost, right? But it's really just one one way, because, for example, for a project like accident detection, where it's really about safety, sometimes just using the word safety should be enough for for the for the management. Yes, we want safety, even though it's expensive. Of course, if that's too expensive, we're going to have to cut it down. But the considerations are a little bit different. Say, for example, we are going to propose a data science project, or a proposal where an existing competitor, or maybe a commercialized product exists, then maybe the main consideration would really just be the total cost, because it already exists. Maybe also not the total cost. I think there are a lot of Conservations, but I'm destroying it all here so that many people can plan accordingly. Not just the total cost. Will be the quality of response, actually. So this is something that not a lot of people are going to be talking about, but when you when you have this data products, it's very dependent on the data it learned upon. So if the data it learned upon is actually quite different from yours, then it's possible that the accuracy is so much less. For example, a lot of maybe the providers that are coming from America have used data that are coming from America, and therefore, when you use it in deployed here in Asia, where the demographics the landscape are different, it might not be as accurate. So let me share with you something. For example. About the age and gender detection, you also develop something like that. Oh, by the way, that can be a prototype for you, if you ever visit here. But having to develop that. And we all know that the facial structure of the West, the Caucasians, are a little bit different. And therefore, when we try the model that have been created using those, then the predictions for us happily would be that we're too young, like I'm 16 years old, because probably if we compare our facial structure with theirs, it's a little bit more but we tend to look younger, and I think that's that's it's a bit too based on the model. So that being the case, those are consideration cost of its commercialized quality as well, and sometimes even the privacy concerns. That's also not being talked about. So this is why. And Vanessa, thank you for this one. Let me drive this to the company. There is a point to creating your own data science lab. It doesn't need to be large, but creating your own data science lab that create data products for you makes a lot of sense, because of the last two, sometimes the quality of data, and the number two for privacy concern, because you don't want your data. Because, okay, so for number one, for the quality concern, how do we resolve the quality concern? Then you have to give data to the suppliers that they will retrain the model based on your data. If you're not comfortable with that, because you know they can resell that, because once the model has learned, see, here we're going very technical right now, but a lot of our intellectual data, not the product that the data produced. So for example, if a machine learning model has been trained on that data, that model is just as valuable, but that's not what is being protected by data privacy. It's just the sensitive data. Now, because you have a model that has been predicted on this sensitive data, if another supplier uses that with the other company, they can say, but I didn't use, I didn't so I didn't sell your data, right? It's just a model that I knew how to create right now. So that being the case, if you don't want that, then the data science lab, maybe just two to three or four persons will be able to do something about that, just for example. Just to give you an example for the CCTV, the placement of your CCTV is a little bit different. It's it's depends on how you would put it right. But now that being the case, that also affects the quality of accident detection, how your the angle of your CCTV training, all of that are a little bit important so that so, so those are the strong considerations on whether to create or to buy. Cost definitely is, is, is consideration, especially for companies that have that require proof of concept. But you know what? It's not that difficult right now to propose because of the massive investments being showered upon that. So if you can do it a little bit lower than that, your job is a little bit easier.

Vanessa Kwan: 23:37

Thank you very much for sharing. Francis, you know the accident detection software is just one of the many technological innovations developed by mega wool. You know, you mentioned earlier about the TAT lab as well. Can you share a little bit more about that with our audience? How? You know, it supports the development deployment of machine learning models and AIS for you know, like you mentioned, also townships safety and security.

Francis Adrian Viernes, Megaworld: 24:00

Okay, so right now, our next project is actually the creation of the rooms where the blue, it's a natural extension of CCTV. So when we have CCTVs, CCTVs are heavy static, right? We don't move. And because of that, you're demanding a city, even though that city has a lot of CCTVs, the average coverage is around 50% meaning there is 50% of the townships we do not surveil. In fact, that's why, when I was watching all this, all this series, for example, and I'm a very big fan of by the way, I failed to discuss my my personal ho being it's actually through crime, through crime and watching all these years Asian series, actually, and well, for true crime, even this mystery novels, for example, a lot of the well, murder, sometimes outside, happens on dark alley. Is, for example, or tarpass, but the in between of the streets, for example, because all of those aren't offered by CCTV rooms when they roam around the township, immediately compliments or enhances your coverage area to the areas that the CCD with are not really capturing, and that, by itself, is one way for us to further the safety agenda, right? So we're developing it with smart capabilities, and hopefully it will be able to count to monitor and hopefully we don't expect, we don't want another pandemic to happen, but if a pandemic does happen, you can also monitor two objects being closed to one another for social distancing, all of that we can deploy once we have this moving, moving CCTV 88 room,

Vanessa Kwan: 25:51

understand, understand. And you know, in addition, again, to the accident detection software. And you know, I remember in recent conversations, you were talking about utilizing similar models for weather forecasting, you know you spoke earlier about crowd monitoring, people counting, and also, in some instances, air quality monitoring. Can you share some about your future plans to develop such initiatives?

Francis Adrian Viernes, Megaworld: 26:15

Yeah, so for weather monitoring, we actually do have it now, and that's a service we give our locator. So by the way, there's one thing you need to know about that for weather and air quality, they're a little bit localized. So like how I've always been saying this to just learn this. So for example, when we get a forecast for, let's say, Metro Manila, you get a forecast for a particular place. Let's say you're going out of the country. Go to Singapore. You go to Malaysia. Oh yeah, this is actually true. I think when you invited us to the Malaysian AIB conference, so you were there, you were asking the driver about the weather forecast. And so if you go with Google and search it, it will give you a general forecast, meaning the forecast for most of the location in Malaysia, for example, at Kuala Lumpur. But what would happen is that sometimes, as you can often, as we've often seen probably, is that there are times that one place is actually encountering rain, but you walk a few steps or a few meters apart, but away from that place it's not raining. That's because weather is a bit localized. So you know what? So we're utilizing this actually, it's a it's a supplier, provider, because they already, they already are able to provide it with a much cheaper cost. It uses the image of the cloud to predict if that image is actually right. Now, very interesting. Why I'm talking to you. I'm looking at mocking me here, and there's a portion here that's a little bit darker, like scary dark, but looking at a few more buildings there, it's a little bit newer, so that, in this case, uses those image to predict the probability of rain. And with that, your forecast for weather for that particular location is a little bit more accurate right now. Why is this important? Well, this is important because sometimes with those weather disruption comes disruption of business operations. They've had a lot of dbos where they have multiple offices so advance or knowing in advance what would be what happened to your localized place would therefore make you plan in advance whether you're going to divert your operations to your other satellite offices and and therefore we would experience a greater, more, more continuous business a year. Business Continuity is better for air quality. Same goes because, for example, we wanted we're all about safety. Now we're all about health. If, for example, in the future, this is a plan that we're doing right now we haven't implemented, but we have plans to do it as part of the component for the health module of the township. If, for example, the health quality severely drops at a particular place. Maybe what we can do is to stop incoming details as a priority is the township. So stop incoming vehicles, close it off. And maybe first, maybe say that is for for air quality purpose, right? So all of that are in the pipeline, and we're very excited about it.

Vanessa Kwan: 29:52

Got it, and in the broader context of smart city development, how do you see mega votes, digital transformation initiatives shaping the. Bucha of townships, urban living in the Philippines

Francis Adrian Viernes, Megaworld: 30:03

I see that as the leader, maybe not just in the Philippines, but maybe in the whole of the ASEAN region. And here's why. The concept, why I really like the laboratory term here in the data science terminology. Yeah, in science and a lot of governments have done this. A lot of government have implemented what they call Special Economic Zone. So the the similar to that concept, for example, if you can learn what works in a small scale, then you can try to emulate the successes on a much bigger scale. Townships are smaller scale city, right? So that being the case, it has the potential if you can have an experiment on what makes a small city or a township works well or be harmonized. Those results can actually be scaled up, because if you start with a larger city, the cost is going to be as high. It's going to take too long because you don't know where to start. The mistakes will be more expensive because it's a larger scale. You're going to spend a lot of money, but try doing that to a smaller city first. So we have cities here that are seven, around less than 20 hectares, for example. Maybe that can be a good example for that. And the successes that you can do there you can emulate with the mix and whatever successes you have there, you can also carry out what works. You can leave, leave out. So That, by itself, makes us potentially one of the largest well contributors in this in this space

Vanessa Kwan: 31:59

understand. So to conclude, start small and then scale big, once you get it right. And you know, what's the next big project you have in the works? You know, we've heard a lot about generative AI chat, GPT, as a data scientist yourself, this must be an area that you're looking at as well.

Francis Adrian Viernes, Megaworld: 32:17

Of course, of course, in in what I'm doing right now is actually trying to consolidate all of this one and see how we can leverage a smaller scale of the generative AI product that, for example, for our executive that you want to ask will give you the answer, right? But it's a, it's, it's something that we are also having in the pipeline, maybe once I'm done with it before, as an entry into our the next round of AIBP How are you?

Vanessa Kwan: 32:54

Yeah, definitely looking forward to more projects, more initiatives that you are leading for mega bolt that aside, what are your views of future business growth for your industry, for mega world and outside the industry, what are the areas within ASEAN that you see most growth over the next three to five years?

Francis Adrian Viernes, Megaworld: 33:14

Okay, well, definitely, let me be a little bit more bias, but also going to be more a little, going to back it up with more objective results, I think, for mega world where we're reaping the benefits of data science approach, I do think that we're going to have the growth is going to be exciting. You don't see that a lot with companies that are already of a massive scale. That mega world is but but with a newer, innovative approach to decision making, like for example, right now, how we are integrating our data sources, I think that the growth prospects are going to be nice. It's actually going to be true for most companies. Remember, we're all coming from from pandemic. So if that's our base case, everything is going to be going to be increases and and that's good for everyone, because everyone will benefits from it. But in terms of the countries that I'm more excited for, the CN region, actually, I'm going to have to also say the Philippines. So there are a lot of forecasts, if you can generally just look at the forecasts are being made by the World Bank or the International Monetary Fund, for example, the IMF, ADB, and I think even there's another, well, not sure if that's Nielsen, but there's another popular well survey, survey group. That means that the group forecast for activities are a little bit higher than the Sen six and a. And well maybe when you think about it, it's also because you've also lagged for quite some time. And it's not unfeasible to think that way, right? So for example, because all of your other neighboring countries are much into already considered developing faster. So us developing a little bit faster this year in the coming years, are not surprising. But more than that, just to share with you some of the news that we've been tracking, because we also do that for mega war, and we also look at the developments the transportation systems have a lot of the transportation projects here are, are quite massive, an example from my last reading. May or may not be, not be the final count, but 51 projects are scheduled at either subway. So I know for a fact for Malaysia, they have already a very developed ones, and for for Singapore, especially Singapore, a very developed public transport. But it's just interesting to see that those plans are being highlighted right now, and I do think that's the first that you need to, well pay attention to in order for the company that you pass

Vanessa Kwan: 36:25

Understand. Thank you very much for friend for sharing Francis, and thank you very much for joining us on the ASEAN B to B growth podcast. Philippines is definitely one of the markets that we are all very excited about. Um, you've got a good demographics and a good GDP growth. Um, so for sure, we will be keeping an eye out for mega world what you guys are working on, as well as the other enterprises in the Philippines. Thank you again for joining us. Thank you Vanessa, and see you soon. Thank you, Francis. See you.

AIBP Intro: 36:55

We hope you've enjoyed the episode. For more information about business growth in the ASEAN region, please visit our website, www.IoTbusiness-platform.com