Kopi Time E073: Vulcan AI’s Manik Bhandari on new tech’s application and ethics

Interpret entrepreneurs to achieve that. Our guest is married Bhandari Ceo and founder of Vulcan Ai, an artificial intelligence company that focuses on building safety and productivity solutions for industry. 4.0 will elaborate on all that. Previously. Manic was a senior partner at EY where he set up and ran the analytics business for Asean and delivered a i powered productivity projects across government. Agriculture, transport and logistics sectors. Manic and very welcome to Covid Time. Good

there may not be a ready solution for that. And if you want the smartest brains to work on it, you would typically do a, you know an RFP and a big consulting company will come in and they will custom build a very nice solution for you and it turns out to be pretty expensive, takes a year and it's, it's, I mean and it may still fail, right? So so that kind of is one of the reasons why

an open innovation program works quite well. So what the government has tried to do is they have multiple problem statements which they collect sometimes from government agencies, sometimes from Corporates and then they have an ecosystem of startups as well as companies And they connect the two. So there are problem solvers and there are problems statement owners so they host these challenges every few months. They lived down the problem statements and then as a startup you can take a look at it

and then two things can happen. One is, you may look at the problem statement and say, hey, I've, I've done something like that before and I can take what I have changed it and I think I can solve that problem or the second is you look at it and then you say which is what we did is that this problem seems worth solving right, if we can solve this problem, I think we can make a lot of money because a lot of other companies will have similar problem statements so you go

in with that mindset and what the government does, I am doing what they do is they essentially give a little bit of a prize money to do a POC.

So you get 2000, Yes, so you get a proof of concept, so you get some seed money, so it's like a seed round but without any state, but basically I m d puts up some money, the problem order puts up some money and that that gets you going and then once it's proven like you know it works, the product works, then the company may adopt it or they may actually help fund it further so that you can actually product is the solution and sort of go to market with it.

And I think the reason it's a very good program because what it does is basically channeling the best thought leadership and capability to a problem statement which is very relevant and it's, it's not like just for one person, it may actually help the whole industry, so it's a good utilization of resources, you know, fast prototyping fail fast and basically startups also get the exposure because sometimes startups may not have the business acumen or the network to actually understand what

the real problem statements are from the industry. So this connecting of the industry with some young and sometimes, you know, not so young entrepreneurs to solve problems. It's a it's a pretty good way of doing things.

then they post those problem statements. But yeah, I've not come across as maybe there are other programs, but the I N. D. Open innovation program, I think now it's in its seventh round and I think there are like 100 problem statements which which have been put forward and it is operating at scale and the good thing is they are basically unlocking startups capabilities across the world, so it's not just, you know, Singapore startups can apply like one of the challenges which

we won. We were selected from 18 other startups around the world, right? So it is not just that just because you're a singaporean startup, you get some extra credit. But um, but yeah, it's good

buzzes and it is basically on the prevention side. So I think that was pretty good. I think the technology was being used in cruises and casinos and other sort of use cases, they quickly took that and adapted it. And then I think now thousands of workers are already using it.

do good at the same time. And, and of course, you know, money as we go deeper into the conversation and I'll talk about this more. So let's talk about your company Vulcan. Ai I'm in case you're curious about the genesis of the name and why and how you set it up.

improving yield and or now more recently digitizing of human work. Right? So you're focusing on industry 4.2. And then I was looking at like who was the like God of, you know, manufacturing or metal works and

then welcome was that? And then welcome click. I, I immediately thought, hey, that's also like a spot and his whole point was that, you know, that alien sort of race was about, um, not going with, you know gut or emotion, but every decision they make is very logical and fact based and that is what, that is the cultural change we wanted to drive in industry four point oh so it was a combination of those two ideas and then how that's how I came up with the name balcony.

and banks and that's mostly consumer data. So it was becoming quite mature already but it's the same thing, how do we sell more? How do we manage risk? That is the problem statement essentially. Then when I switched gears to another industry, so the whole agriculture and forestry industry was going through a early stages of some big transformation.

So the drones were becoming more and more pervasive. But the problem with drones was basically when you fly drones on these huge plantations or forests, uh it's hours and hours of flying data that you've collected and you traditionally had people looking at that, I'm holding it and then saying, okay that three there may have an issue. It doesn't look that green. Right? So we did the math for one of the biggest

paper companies in the world. And if they actually processed all the drone imagery manually, it would have taken them 100 people in 100 years to go through it. So it was just not possible. So one of the innovations we did was basically used of ai to automatically process all this imagery which is being collected and then pinpoint exactly which tree had an issue so they could do something about it. Now this project because 12 months, 20 people $3 million.

So of course for a big company, I mean, which is like a $10, $12 billion dollar company, this is still pretty good on the way. Then we started thinking that hey, if the same thing, if we have to democratize it and make it available for a small farmer somewhere, this is not going to work. Right? So how do

we make that shift? So the reason I started working, I was then how do we take some of the best kind of solutions and ideas which we had conceptualized for big MNCs big companies, billion dollar companies and build it into products which are bite sized, which even the small guy on the street or the small farmer can start using. So so coincidentally and I think

that was a pretty good meeting of the minds. One of my biggest clients which is a big Indonesian family business which have interests in oil palm and paper and forestry. They said that's great. Like you build it for us and then can be monetized it and open it up for others because this should become a utility. This should not, this does not have to be a secret competitive advance. This should be a utility and everybody should benefit from it. Right.

So that's how we started. We started as like a JV between me and another big family business and now we are at a stage where we have built these products and we are opening up for other companies and we are making the product so easy to use that and so

easy to deploy that smaller smaller companies. And the vision is that a small farmer saying Bangladesh somewhere can just open an app and get access to the same technology which just five years back people were paying millions of dollars to access.

a year. And then every day basically there are harvesters who are going to the tree and harvesting the fruit and then the fruit is then sent to the mill where it's converted into oil and there's the same oil which is you know if you eat the Mcdonald's the soft serve cone or french fries, it's the same oil which is being used. Palm oil is pretty prevalent, We may not know it, but it's there. And even if you have some beauty clean products, it's their great

And then the second thing you control is harvesting, you have to harvest is that the right thing? Like it's simple, right? So if you have a garden then you if you're planting tomatoes you will you know pluck it when it's like ripe and very because you want to use it immediately but with fruit it's a bit different. So right so the timing of the harvesting is is really important. So they have 30,000 workers, right?

So they have to then Basically go and walk this 500,000 hectares and look at the trees and then figure out whether the trees are healthy or not. It's hot. There are there is wildlife, there are snakes. So for a regular person who's in the village and who's helping run this plantation life is pretty tough. You may be walking tens of kilometers every day trying to find issues but in the end you may not still find it because you don't know where to look, right? So the big change we did was get the

satellite image, we do the processing. Then we give the location of each tree where there could be an issue. Then this guy goes to that particular location and verifies whether it's correct or not and then takes a picture, sends it back and say yes there is an issue. Either we need to spray some pesticide more, more fertilizer. So it's basically completely changing the way things used to operate earlier where people are just going around randomly walking

and tough life. Whereas now they only go to the places where they actually need to go. So it's a much more targeted and less work. But the impact is a lot more.

which are good meaning you don't need to fertilize. And So so one of our clients spends $300 million dollars A year on fertilizer and we save like 20%. So which is that 16 million to 16 million a year. But at the same time it's great for the environment because you're not putting chemicals into the soil, right? So it's a so on one side you're increasing year but you're also doing it in a more sustainable way.

spectral sort of profile for different things. So what we've done is we've taken like millions of trees and these trees have been labeled by ground proving that whether it's healthy or not healthy and then we sort of we have matched So the ai matches the spectrum profile with what's good, what's bad and I'm simplifying it, but basically that's what it does. But it has evolved to such an extent that not only does it tell whether it's healthy or not, you can even tell the type of disease

which is suffering from. So it could be gannon, normal leaf eating caterpillar. And the beauty of that is once we tell that the person doesn't even have to go and check, You can straightaway go with the right intervention and the insecticide that you have to spray right there and the sooner you fix it, right? So one tree is infected, it's like COVID-19 it can spread. So if I please have it and you don't intervene on time that disease will spread and that's a massive

impact on yield. So with satellite imagery you're getting more recurring every three months. What's wrong? You take the action and you sort of nip it in the bud in a way it's like contact tracing

you only see the visible spectrum. So you see what you see. But the satellite has more spectral band data coming in and in fact a lot of the nutrients and health information is captured in those banks. So the tree may look perfectly fine from outside to a naked eye. But actually there may be something wrong. So it helps still helps with early detection. So it does make sense. I think the only constraint probably is the order size. Let me rewind a little bit. So like six years, seven years back.

Um satellite imagery was really expensive and the big governments of the world would use it for defense and intelligence. Right? So now with the satellite space booming and it's pretty congested out there, the cost of satellite imagery is going down. So they have reduced the order size. But even now if you want to get satellite in misery you have to get at least 100 square kilometers. All right, so it would not make sense to get that kind of satellite imagery for the small farmer.

But co operatives and governments are sort of getting together source sourcing it at scale and then distributing the insides to the farmers because it does help the economy and the GDP

robots but they can't get everywhere. So my I mean I think the point of view I have is that yes, I think productivity gains will happen, but humans are pretty essential and there will be a lot of things, especially physical work, which which may not be completely automated. So how do we make it work and more enriched, safer and better for for blue collar workers. Right, So like during covid 19, your go to Starbucks, but

and you can work from there. But then along when you go there, they could also be a cleaner who's cleaning the road, right? So life is pretty tough for them. So I consciously decided that we should be using ai not to just replace humans, but make lives better for the more sort of blue collar workers whose life is pretty tough. So so we developed a solution which basically taps into variables and there is a i in it which sort of gives nudges to them if they are sort of not taking care of their

they're working. But more importantly, we're also giving visibility to the enterprise level to the employer to basically point out areas or teams which may be overworked and may be at risk of workplace accidents because sometimes people, You know, you may have 50 people working at a construction site And there may be somebody who has a preexisting heart

condition and you may not even know that. And you've given him the same amount of work as somebody else and then you but you don't want people to just die on your watch. Right? So this gives them that. Okay, fine, I have the same 50 people but this person I can, you can reallocate resources based on how they react to physical stress or exertion because what we can now show employers is that same amount of exertion. Physical exertion, which is captured based on physical movement and heart rate.

What is the impact on fatigue? What is the impact on their heart rate and you can rebalance the workload. So this solution is, you know, targeted at the construction sector, maritime sector facilities management. We are now starting to use it for aircraft cleaning because the other thing we do is just like in your foot bed basically using your hand motion. it can detect whether you're

exercising or not. We have work activity detection meaning if you're doing some work it can quantify that work actively. So like in agriculture when you're spraying fertilizer you spray like that. It's a very unique and motion. We use AI which is embedded in the variable to convert it into fertilization then if the plantation owner wants to see okay which part of my plantation was fertilized? Which one was not because most of it is still

manual application. We can see we can see that that and we can sort of have a geospatial view of that. So the idea is to make the person's job safer make him more in charge of his own health and and fitness but at the same time also given aggregated view to the employers. So they know where the risk and the hotspots are in their workforce so they can manage it better

as an employee perk. So the reception has been quite good. I think on the corporate side there is some resistance, I won't say to the idea, but some resistance to like what the business cases, right? Because like if you're a company which provides cleaners say to um BFC then they will say I am footing the bill but it's not like DBS is paying me for this, right? So because I am also a service provider to somebody.

So the angle we are taking in and this is with the government and the the union is that we need to educate the service buyers. The end clients to basically tell them that you know, safety is important. You don't want a cleaner to slip and break his hip while he's working at, you know, mbf see tower one, right? They say yeah of course we don't want that.

Then you say, okay then you're cleaning company. You tell them that you know, they need to equip these solutions in their workforce and to be honest it's like 5 $10 a month per person is hardly anything. So if the solution is there and if people wanted people will definitely take it. Right? So that was one the second one

which I don't think is the right attitude. But we do come across some people, I was talking to a person who manages like 100 cleaners and his point, I think he was a bit too transparent was that if I give this watch, then we may discover people who have heart trade issues and then we may have to find replacements for them. Which is true.

it can be impacted just because I'm not healthy. Like for you and me it doesn't make a difference, right? So if I have a hypertension or something and you know I will go to the doctor but I can still continue to my work. But if you're like a small breadwinner and somebody gives you a watch all good. But the watch tells you that actually you're a very high risk of getting a heart attack while you're working and hence we have to find somebody else. It's pretty scary,

the roads regularly. So given that there is not enough people in Singapore were not able to inspect the roads as regularly as you want and you would have noticed over the years the number of potholes has has has gone up and the portals don't get fixed as soon and it's not just about a comfort, comfortable ride, right? It's a safety risk because people sometimes may swerve out to avoid the Porthole and accident. Plus we have delivery riders.

So one of the things we did was we basically took our Ai and embedded into like a small sized device this edge devices now being put in busses and other vehicles as you're just driving around the AI is automatically detecting portals, tagging it with the location and then sending it back. So instead of having somebody inspected once a year. So we have like memorized the road inspection process. So that's another pretty cool project which we're now scaling to the rest of Singapore

So a simple rule which basically says, hey, it's nighttime and you've been away from your bed for more than one hour. Send an alert to your family or to somebody else because you should be in your bed. So like simple tweaks like that with technology and you can solve big problems.

it was more around that. How do you optimize the, like the line of credit, like if they want to gamble and how much can you give and how do you manage risk? And but what was missing from that objective function was that are we making this person a, you know habitable gambler and you know, and then that and can you really afford it? Right. So, so now do you do that or do you do

the profit optimization? So there is this trade off of course, in pure profit optimization, you also don't want that person to go completely um over the top so that, you know, he goes bankrupt and doesn't come back. So there's always advance but there are these kind of questions which are coming up more and more because in the traditional world it's a

human and the human uses some judgment. Part of it is emotional part of it is, you know, profit seeking and you try to balance and you do it but different people may react to it in different ways and the way I work has been basically trained on decisions which have been made in the past. So unless you also tell whether this decision was good about the way I will learn from whatever decisions have been made in the past. So there are some obviously some parameters which we can set

to improve the ethical implementation we have. So with that in mind basically the Singapore Government and the Singapore Computer society and it was announced in the World Economic Forum, it's the first sort of body of knowledge of its kind where um there are, you know, case studies in terms of what could go wrong, but they also specify the things that we have to keep in mind to make sure that we're doing the right thing

and it has basically four different dimensions. So one is around governance and the governance is basically, you know, in the ai value chain there are different players, it could be the business owner, there's the data guy, there's the Ai guy and then the person who executes it, so who governs that? How do you formulate the right problem statement? And it should also become part of the board agenda because things can really go wrong if you don't implement

it correctly. So it talks a little bit about governance around that. Then the second important point is around human centric city, which is basically saying hey we still need to put humans at the center of it and when we design something, it has to be designed with the humans best interests in mind. So then there are sort of steps on how to do that. The third pieces around operations management and operations management is more around over a period of time, circumstances change,

environment changed data changes and the model made drift. So when you put an Ai model into production, how do you make sure it is not biased, it is fair and there should be like statistical measures put in place to manage the operations of that. And lastly it talks about stakeholder communication, so like how do you communicate to um every user in the summer? The ecosystem. Right. So basically if you are in the gambling example, if you're refusing to give credit widely doing

that and how do you explain it? Because people may say, hey, I didn't get it, but my friend got it, so it doesn't have all the answers, but at least it gives you a toolkit to start thinking, start sensitizing yourself to ethics and some frameworks to be more smart about ethical implementation.

benefiting from it, but it has some degree of global reach because you know, it affects all of us. Facebook is not just for some rich countries,

it's everywhere. Uh and and similarly I want to talk a bit about your personal experience as an entrepreneur in general and also in the context of, of being an entrepreneur in Singapore now, you know, it's a small nation, it is wealthy, it is well educated, has world class infrastructure notwithstanding the potholes that you were talking about, but the markets here are small by global standards and so is the talent for, So how should a local entrepreneur

deal with this scaling issue as far as Singapore is concerned?

So that's one plus there is a lot of support, I would say both financial support as well as um you know, just guidance and and getting local entrepreneurs connected with local companies to do some schoolwork, prior new solutions, but also globalized. I mean the way they have architected the whole value chain is pretty amazing, right? So, very early stage you have these innovation programs where people give problem statements and you do that. So we also benefited

from that. Right? So some of our products are coming out of that, so you do the problem when you do the proof of concept, you have the prototype, there is some seed funding, you do that. Then the next step is basically you have, so I M B A will help with that then, as it starts to get some traction, there is enterprise Singapore which focuses on local smes right? So as a local startup

will qualify as a local sme initially. Right? So then they have programs where they would even give you 80% of the cost of doing a pilot with any other company in Singapore. So now you sort of have a few pilot clients

and the product market fit is established. The next step is basically you want to globalize now the Singapore government also has sort of funding and support available to open offices overseas and again, do those pilots now, once you've done that, then obviously then the, the VCS and the institutional investors and kicking and they also helped make those connections. So the infrastructure is there, the funding is there? I

think the talent we're slowly getting there. So it's it's actually a pretty good place to, you know, have a startup and do things, but but you're right in terms of scale, you may not have that scale. Right? So, so it's a great place to start a startup, but you have to have the mindset that I need to make it global to really make it big. So even when you start in conceptualizing you,

you always have to put the lens on that. Is this is this problem statement, something which can apply to other markets in hardware globalized. So I think that, so it's just a good springboard, I could, I would say a test bed to do things, but you're definitely, but the capital is here, but you have to think global, if you have to be successful,

But I think in terms of tech, I think our branding is, is not there yet, so other markets may not immediately think that just because you're a Singapore startup, you must be good in tech, that is not the connection

they will make. But having said that, what I realized is that if you're playing in the industries which Singapore is going for like you can be a tech company, which is saying the real estate or construction sector, you will be taken more seriously because Singapore is known for that, if you're in Fintech and yep you will be taken more seriously because Singapore is a financial help, right? And we have the, you know, one of the world's biggest ports and maritime industry is big, right?

So if you do something tech in maritime again, you will be taken very seriously. Same thing with the airlines, you have Singapore airlines, changi airport, So if you, so if you are in a tech startup, which is focusing on these industries, which Singapore is famous for, then I think you have a much better chance of being sort of appreciated and taken seriously than if you like for us

for sure. Like when I talk about agriculture, people will say, hey, what I mean, Singapore doesn't even have agriculture, what are you, have you? Right? So they will not take place. But of course we counter that because we have years of experience and even though the team is here, our experiences outside Singapore, right? We don't have a client in Singapore, but everybody's outside.

So I think that's the challenge. The other thing I've, I think, which is very, some is we have talked to some disease and we talked about like, you know, the ease of getting funded in Singapore and, and so this was some Malaysia based VC and their point was, and I'm not sure whether it's right or wrong or even politically correct, but what he was saying that don't take money from sort of Singapore because

you may end up getting fat and lazy, right? So basically there's a lot of easy money out there and he's saying over the last 10 years, he's seeing so many startups, it's so easy to make money that there, it is not the scene where, you know, there is a startup guy and you know, he's worrying about his next paycheck and he just has to get it right.

So that kind of aggression, desperation is missing from the Singapore startups and his point of view is that, you know, Singapore startups have it easy, there's a lot of government support, right?

which also leads to a massive boom bust cycle. So I think, I don't think that isn't necessarily specific to public support for entrepreneurism, private sector also overreaches and over

invest and invest in ridiculous things many times as well. Um, but no, I, you know, from the very beginning of this conversation, I've been struck by the proactive role the government of Singapore plays indulging entrepreneurs in going in the right direction, as long as, you know, just producing value for the society, I see nothing wrong with that. Um you briefly touched upon skills and labor. So let's get into that. We've been reading a lot about it lately.

And as a, you know, a company that is growing, that you run manic, I'm sure you recruit regularly. So what are you seeing in terms of the homegrown supply of labor and maybe going even once to further, what can be done to get singaporeans ready for this tech heavy future that is inevitably out there.

sustainable. Right? And I think there is this huge divergence which is happening between the tech haves and the tech have not, right? Because at the same time, I, you know, I take grab rights and then I'm talking to some of these drivers and you know, some of them have been displaced.

Um, and they had, you know, regular jobs before covid and they got displaced and, and smart guys, you know, but the, it's so much easier to basically for the short term, like if you have a two year blip something like a covered to basically just become a grab driver, do delivery and you know, you can sort of, you can support your family. Right? So, but that displaced group and I think in the earlier

days we had semiconductor and other industries. Right? I think that to us offers a much bigger pool of people and opportunity and that can change in six months. Right? So if you address the talent from like let's start teaching quoting in schools and then let's have, you know, six new programs saying n tur in us for computer science, it will take time. It will take you know, four or 56 years to see that come on board. But if you take some of these other people who

are professionals, they're very serious. They are very hard working and you know between 30 to like 50 years of age. And if we can tap into that and get them some sort of basic um tech skills that unlocks a huge opportunity for us. Right? So, so I think that if we have, we have to focus on that. But I think it needs

code platforms. So you may not even need to cold and there are other aspects of that you can get into, it could be designed, it could be testing, it could be other things where you may not necessarily have a code, so if we tell people that, you know, you can be a non techie but still do a good tech role, I think that may unlock a lot of the sort of talent which we have, but traditionally we've, we've not been able to tap into

part of the whole tech side. Guys know manic looking at the video, it looks like you're in your office, so you're not working from home.

just come in and your white board and because we have physical infrastructure, right? So like we have the watch and we have sensors and all that. So it's become more like a lab. So you basically come here, you try out different things and it's a good people want to come here. Right? So I think when the government allowed people to come back, they were so happy everybody was, everybody was back. Yeah,

googling Devious Research Library. Have a great day.