Brainstorm AI Singapore 2024: Opportunities In The APAC Market
Juhi MCCLELLAND, General Manager and Managing Partner APAC, IBM Consulting Justin NGUYEN, General Partner, Monk’s Hill Ventures Chan Yip PANG, Executive Director, Vertex Ventures Ren Yeong SNG, Managing Director, Artificial Intelligence Strategy & Solutions, Temasek Moderator: Ellie AUSTIN, FORTUNE
Category
🤖
TechTranscript
00:00Good morning everyone. Thank you very much for joining us. So just to set the scene as I'm sure everyone knows it's been a
00:07pretty turbulent global environment for investing in recent years. But against that backdrop APAC has remained relatively
00:16resilient. I want to start with Justin Chan yet. And when young kids you will work in the funding landscape in some way and
00:24then do he will come to you in a second. Which sectors in which sectors are you seeing the most transformation as a result of
00:31A.I. and Justin. Let's start with you. OK. Well so I'm pretty broad. I think I think I was actually going to touch all
00:38aspects of our life. I don't think there's such thing as going to be an A.I. quote unquote first company. I think we're going to
00:43embed it into into just like mobile. Right. That came before it or the Internet. So I think it's going to be quite broad.
00:50However to get to specifically I think there are immediate in our sort of immediate future I think around around education
00:57anywhere we can have relationships with people. Right. And certainly around anything around content is going to change very
01:02very drastically. I think health care actually is going to change pretty drastically. So roughly those are kind of the three
01:11immediate ones. But I think it's going to touch broad aspects of our lives quite honestly. Can you talk to me a bit about
01:17education and the specifics of what you're seeing and what could be possible. Yeah. Yeah. So so so we have a portfolio company
01:25that actually helps people pronounce English better. So that doesn't teach them vocabulary and grammar. And previous to this
01:31they've been around for a while. Previous to this they were mostly lesson based. But now you have the ability to say OK I'm
01:37going to I'm going to go do a fortune brainstorming session tomorrow do a mock do a mock interview with me or I'm going to go
01:42interview for a job as a as a controller. Do a mock interview with me. And they've incorporated that into into their
01:49product. And actually almost overnight there will be their average order value doubled. Right. And so what happens is what
01:58lesson base you can only get so far of that. Now the scenario is a limitless right. And so it becomes a useful everyday type
02:05product. And so that's one very concrete example that's that's actually out there already with our company and also. But I
02:11think it's it's got to be quite broad. Any you know for years it's been things have been sort of narrowed down to transactions
02:19and now we can have these longer relationships. And that's essentially what we're seeing is products becoming you know
02:24being relatable to us and talking to us like we talk to each other.
02:29Chanya when you're where are you saying my transformation. Yeah perhaps I can start first. So for me I think I've been seeing
02:36more transformation in B2B sales as well as a potentially venture capital as well. So we ourself we are actually using M to
02:45help us identify companies that are interesting to see. So if you look at the lifecycle of a venture capital it's essentially
02:51sourcing picking the right company then eventually building together with the entrepreneur as it's harvesting at the end
02:59through an exit. So the first two bucket in terms of identifying the company as well as doing the analysis on the company
03:05LLM have the potential to actually help us. And so far we have been testing and building our own models. We have been able to
03:12identify quite interesting companies of that. So we teach the model what kind of companies are what kind of traits that we want to
03:18see. Then we ask the LLM to help us to rank from one to 100 score. Then from that score then we will know which one to
03:26prioritize that eventually go and reach out to some of these companies. So so far it has been quite fruitful. So we managed to
03:31find interesting companies that we have missed in the past. But right now with this then in terms of ability to capture more
03:38companies I think it can be quite transformational for us. And is it speeding up the process as well. Is it changing the
03:45timelines of how you work. Yeah definitely. In terms of view sourcing is speeding up because in terms of in the past what we
03:53have done right is that we try to reach out to as many companies as possible. So we wouldn't really have much information
03:59about that company whether it is something that we should be prioritizing or meet first. So with this LLM in terms of the ability
04:07to score some of these companies. So right now we are able to prioritize which one should we actually reach out to first. So I
04:14think that part at least for us so far in the six months that we have been trying it has been quite interesting. We haven't really
04:21invested in one in which it came from this LLM in terms of view sourcing but has maybe one year from now. Maybe we have a new
04:28story to tell. So fascinating. When you're. Yeah I think you know just building on what Justin and Charlie have mentioned I
04:36think we take a step back. Companies that have adopted generally the fastest are those that have moved to the cloud earliest.
04:44Right. So you look at the e-commerce play players. It's much more easy for them to adopt generative AI. Secondly in this space I
04:51also think that the professional services sector has a lot of opportunity as well from legal accounting as well as what he's
04:58mentioned financial research analyst. If you look at the costs in the last 10 years the cost of producing services goods have
05:05declined. But the salaries of software engineers or professionals have gone up. And I think this is the way where that would
05:12change. So as an example we invest in a company called Robin. I did ask. I would enable contract review and translation. We
05:20always say that the best AI model is the business model. So essentially what they have done is replace a service to charging by
05:27per user by output and by outcome. I think that is transformational. We also have another company that one of our funds invested
05:34in this in stealth mode right now is looking to automate workflows within accounting and tax preparation which means the agent
05:42would work like an accountant and still use QuickBook. We're not replacing QuickBook. We're replacing a workflow. So essentially I
05:49think within the professional services space there will be workflows that will be replaced by certain agents or autonomous type
05:56of AI enabled software. And I think to what I mentioned on financial research and analysts I so recently saw a company that use
06:05agentic frameworks to create more like a financial research analyst. It was pretty good when I tried it. So those were the I
06:11think sort of perspectives. Workflows being replaced rather than roles as it were. Exactly. So workflows within professional
06:18services where hallucination is a feature not a risk. Right. The early part of your discovery within your workflow. You actually
06:26want hallucination. Right. And then that that that that is sort of like those are workflows I think will get generally replaced by
06:33potential a new air and a boat. Yeah. First it's what Justin said. Software. Now do you think you have a different perspective. I
06:39imagine to your fellow speakers. Can you talk us through a bit how in your role at IBM you're interacting with AI on a on a daily
06:46basis both internally and also outwardly facing with your clients. Very good. Yeah. First of all I mean I agree with all my
06:54colleagues here. They've made some excellent points. I would say look I would say the CEOs. We've done a study that says 60
07:03percent of CEOs are getting pressure from their investors. My colleagues here and others to do something about AI. Right. And
07:13when the CEO gets pressure 60 to 65 percent of them are being forced to do something. They want to react to the investors.
07:21So when you are in react mode. What did what happened for the last three to four years. We saw a lot of POCs. Right. There was a lot
07:30of proof of concepts a lot of trial and some error and learning from that. So what I am seeing in the last six months is a bit of a
07:42shift there. So there are still POCs going on driving innovation and investments from my colleagues here. But we are seeing
07:50companies especially large enterprises really embrace AI very seriously. And to your point not just changing an application but
08:00that end to end processes. To give you a couple of examples we have completely changed our HR process. Now we have something called
08:08Ask HR and we would you know as we are 300000 employees we have a lot of questions to ask where people company. So the ask HR
08:18process has really accelerated our time to value from an HR perspective our entire procurement process right. Our contracting.
08:25You know you have some really basic things that have been so super automated that the time again to do the boring tasks and leave
08:33the humans to do the value added tasks is how we're seeing a big change internally. Those are just some examples. Externally I
08:41see a lot of adoption happening in code generation. So right. Like the engineers who are generating code now are seeing almost 80
08:52percent productivity because a lot of the code is auto generated and they only need to ensure that the code is good. We're seeing a lot
09:00of adoption in banking and financial services again and risk management and growth. We're seeing growth in health care. I think that
09:09point you made Justin and health care. We're seeing a lot of innovation happening in messenger RNA development. So I would say we're
09:17seeing our CEO says that we're expecting four trillion dollars in productivity to happen in the world from Jenny. I four trillion
09:26gentlemen get your pockets ready. But also we're seeing a lot of growth happening in business model disruptions. I do want to share at
09:36some point if appropriate what are some of the inhibitors that are coming in the way of this scaling and an even faster innovation.
09:45But I want to make sure we have a balanced perspective here. Let's talk about those inhibitors. OK. So maybe outline a couple of them
09:51do you hear. And then I'd love to get your your perspective as well on whether you agree. So maybe name two or three of the most
09:59challenging ones that you're seeing at the moment. I would say I'm seeing three inhibitors. The first one is skills AI skills. And look
10:08this is upon all of us to elevate the entire world. IBM's made a commitment to train 30 million people a million women in a pack 30
10:19million globally on Jenny. I saw skills and AI skills and I'd love alternate perspectives on how do we all raise the watermark here.
10:28Right on skills. One thing I love about AI is it's inclusive. You don't have to be a software developer to embrace AI. Right. Like our
10:36children are writing AI apps. I have a cool daughter. So but but seriously like it allows everyone to participate. So what what is
10:46coming in the way. What is this inhibition that's holding people back and how do we empower them to embrace it and do something with
10:54it. Use the agents that are available to us right in simple ways. So skills is one. I think the second is cost. We do know that 40
11:02percent of IT budgets are shifting to Jenny. What does that mean. Data is going to be everywhere. So how do you balance the
11:11investments you're making in Jenny. I read the value that you're getting and it always has to be this way right. The value has to be
11:18higher than the cost. And honestly the third thing I'm seeing is concerns around privacy security fairness ethically I governance.
11:29So if you put all of that together you have to ensure that there are no biases in AI. Is it accurate. We've all heard about all the
11:37breaches. In fact you and I were flying at the time when the big breach happened with cloud strike. So how do you ensure when your
11:43data is more out there data is moving more to the edge. How can we ensure that that's secure and all the way through. So. So
11:50skills are balancing cost and value. And the third would be privacy ethical governed. I. And you mentioned as well the huge amount
12:01of growth that you'll see. I was expecting to unlock as a result of Jenny. And Justin I'd love to know from you what do you look
12:10for in terms of key indicators of sustainable competitive advantage in AI when you're thinking about investing in various
12:18companies. Yeah. Yeah. It's a great question. It's actually a pretty I think it's a pretty tough time actually to be an investor in
12:26AI right now. Yeah. I was saying earlier I've been around since the early days of the Internet and the applications and the usage
12:32of what the technology you know it's 10 5 10 20 years now. It's unimaginable today actually to us. And so that's kind of nice. We
12:42get to sit here and see it. But to be a little bit more specific right now it's very very difficult to see what is the thin layer on
12:51top of what you know IBM is doing. Google's doing AWS is doing a thin layer on something that's pretty commoditized right on these
12:59large language models. And what is truly valuable. Right. And I think at the end of the day it'll come down to data and and how
13:07you use that data when you when you know when anyone can prompt then what is proprietary to you. And how do you go about
13:16collecting amassing and getting results from that data. Right. I think ultimately it comes that almost a lot of what we see a lot
13:26of the investing that we see out there just thin layers on top. There are just new interfaces for for prompting actually. Right.
13:32And so how do you move beyond that is sort of the immediate first steps. But actually I'm super excited at what's coming next. You
13:40know the early early early days of the Internet and of the Web actually. And I was around then completely unobvious. We knew you
13:48could feel that it was something there that was going to change your life. But but but it was totally unobvious. Obviously Jenny I
13:53has a much much bigger aha moment. You can see it right away that that it's going to be big. But once again you know how it
14:02impacts us right now. Imagination is is is kind of all you can do. And China. What are the most successful CEOs and founders
14:12doing at the moment with regards to data but also across the board to make sure that that position in their companies is one of
14:19the ones that doesn't unlock this huge potential growth. Yeah. So. So I think so far the best example publicly would be
14:26Klarna. Right. So whereby they actually managed to kind of use a I to do customer service but improve it in a way that the
14:36response time to customer actually significantly decrease from 40 minutes to two minutes. So that is what we have seen that is in
14:44public. But at the same time within the vertex portfolio which actually do have this company called Store Hub in Malaysia. So
14:51it's not hard. You can think of them like a equivalent of Square in the US to provide kind of sales terminals to FNB operators. So
14:58for them to do sales they actually need to hire a bunch of sales development rats to do cold calls to reach out to potential
15:06customers that eventually to convert them. So the problem that they face in the past would be it's hard for them to actually understand
15:14the performance of the SDRs until like three to six months later because they aren't able to sit down with each of the SDRs to understand
15:21how are they pitching to the potential customers. But with Gen A.I. and L.M. come into place then right now they are actually able to
15:29track. So by tracking what they say within the call so they can quickly do ratification right. If this particular person is not
15:36selling the right messages they can actually tell and retrain the person in order to help them to increase the conversion. So so far
15:45the in terms of results has been quite positive. So they were able to increase the conversion rate by 30 percent and able to cover
15:53the number of calls from 2 percent to 70 percent. So the other 30 percent they weren't able to do so is because the data clarity is not
16:00there. So that has been I think quite transformational transformational for the company because they are able to reduce the cost of
16:06acquisition down by quite a bit. When you can you talk us through a bit how you go about valuing an A.I. startup.
16:14Yeah. Firstly this is a very difficult question to ask.
16:21If you look at the valuation metrics they really are very very wide ranging. Instead of asking that we ask ourselves where is the
16:27value. Where does the value accrue. Right. So if you look at like what Justin had mentioned earlier it's not easy time for
16:35investors. Now of course as we look at any A.I. company by now you've heard that three very important components talent data
16:42compute. Right. And of course capital capital provision is also very important. But we ask ourselves where does the moat accrue.
16:50Where is the value. And I'll submit to you that the moat is not really in the data. It's in the customers because the generic
16:56economy of the future the data that is being created has really been put infrastructure. What is important is that you need the
17:02channel that you engage the consumers in order to be able to generate data new data that's fresh is able to build value in the
17:10moat that you want to accrue. Very conceptual. Let me give you some examples. Right. We always say that we look for companies
17:16that product forward not product centric not data centric but product forward. What does that mean. What does that mean. Let me
17:22unpack that for you. So get a copilot as you pointed out. It was not looking at all the code that's in GitHub. It started with
17:30what the developers need. That's product centric. But product forward is to know that because we know LLM cost is declining by a
17:37Factor of 99.5 percent the last one year. You have to imagine what is the future going to be five years later and build your
17:42product for that. So GitHub I believe is in such a position because it knows what its customers want. It has a talent to forecast
17:52forward as a compute to be able to test that. Right. The other example we have is a company we built in Singapore called
17:58Minion AI. They launched you what's like what why you you. It's a rewards loyalty coalition on the surface. It just looks like a
18:04coalition. But the back end we've built an AI system that basically tries to launch different products up to different slices of
18:10consumers or where it matters to you. So for example we launch a shopping list to sub to a slice of consumers where it matters for
18:17them. And we want to make it more automated. Of course the chip was launched as well. But many people do that. But essentially you
18:22want to create an architect a system where you go broad. I call it a team not for Tomasic but a team. You go broad. You innovate
18:30on product and you go deep in where you think your advantages would be where your competitive position would be. And that's I
18:37think a bit what you he mentioned organizationally. We need to look at how to enable organizers to do that. Go broad to innovate
18:44enable the innovation and go deep into areas where we actually double down on that. That's so interesting. And yeah of course.
18:51First of all I love that my my financial colleagues are talking about large language models. How cool is that. Right. So I'm an
18:58engineer by training. Okay. Qualifier. But I was trying to avoid being techie. So thank you for opening that up for me. But I do
19:07think to your point I wanted to add that you know look the large language models are excellent and we're seeing efficiencies happen
19:15from them. But there are a lot of smaller language models that are also to your point on data. Right. You could get a better ROI
19:23from a small language model as well. So you know let's let's be let's look at both. The other thing I wanted to mention is that
19:33there is you know I think you were making that point on outcome based value. So earlier a lot of our pricing is done by cost plus
19:43or customer acquisition cost. I mean again different manifestations of cost plus. But I think that would also be sort of not product
19:52forward or not visionary because a lot of the value is still to be unleashed. And you know to your point we don't even know where
19:59the value will come from yet. So this whole outcome based business models and pricing are also another area that I just wanted to
20:07sort of double down on your points that let's look at small models. Let's look at different pricing models and value creation
20:15models. And then the last thing I'd say is there is a lot of opportunity to innovate. And you know really a commitment to open an
20:24open source and open ways of doing things is really where we as a world and an economy will thrive. Right. So this is not where hey you
20:33know this is my model. I'm going to hold on to it. This is like you know let's if a developer in Asia writes code a developer in the
20:41US can look at it overnight while they sleep and bring value to it. So it's really creating like a global pool of people and and tools
20:50and access through like a lot of this open innovation to all the companies. My colleagues mentioned I do think that we are going to see a
20:57substantial improvement in the world based on this open model. Just quickly how do you do that when you've got the two massive
21:04superpowers these two power blocks kind of really running the show as it were. How do you foster that global openness. I think technology
21:13and AI is going to be democratizing. I really believe that. And there is no way when you're like IBM build this granite model and now
21:21we've opened it up to the world. Right. And it doesn't stop at a startup. It doesn't stop somebody sitting in Guangzhou China. It
21:32doesn't stop a developer in Poland. So I do think that through technology and its democratization and the skills and people creating
21:41value we are going to see a much more collaborative world building this kind of you know driving the right outcomes for the world.
21:50And also you know if you see the banking innovation as I mentioned as just an example because this is happening more in the open way
22:00it's it's not holding things back. People want to learn from each other because the value that's going to unfold is so enormous and
22:08so unknown. At this moment what I am and advise a lot of CEOs and C-suite even CFOs I'm seeing people craving best practices from
22:18around the world versus sort of a protectionism mindset because you know with so much money and investment going in this space even IBM
22:26launched a half a billion dollar fund again to support open. I'm not as large as probably my colleagues here. But again I do think that
22:33and it's not constrained. It's not like it can only be in the U.S. and it can only be this. It's basically aligned with our values of
22:40sustainability open innovation. We've only got a few more minutes so I'm going to finish with some sort of quick fire questions.
22:48Chanya let's start with you. When we talk about investing in APAC we're talking about a lot of very diverse countries that if we're looking
22:55ahead to the next year there one or two markets within APAC that you think we should keep an eye on for particular growth in AI. Yeah we were
23:03just discussing with Justin that right now in terms of Asia-Pacific especially Southeast Asia because both of us actually cover the
23:10Southeast Asian market a little bit more right now. It feels like if you look back at the mobile days whereby mobile started in 2008.
23:20Uber started in 2009 and then Grab started in 2012. So that means we are in the phase whereby a lot of the developments are actually
23:28happening in the Western world. And we in Asia is trying to really understand what exactly is the implication right now. And perhaps maybe
23:36in within one to two years time we will see companies that are indigenous to Southeast Asia just like Greg. But having a twist to it
23:44whereby it makes them different compared to the Western counterpart. So I think that it's important to learn this particular trend
23:52right now. For us as we see we are always meeting with companies to see what's the latest innovation. And what we would like to see is
23:59companies that actually have differentiation when they go out and compete. Because at the end of the day in order to succeed you need to
24:06have some advantages. Otherwise you run into price war which is something that we do not want to see a popular companies going to
24:14Justin. Other than Singapore which which countries in the region do you think are showing that differentiation. Yeah. So so I'm from
24:23the our Vietnam office. If you can't tell my last name when I'm Vietnamese it's probably 40 percent of the population. So I'll make a
24:29plug. I'll make a quick plug for for Vietnam actually. So as we know I think yesterday there was a panelist that was up here that was
24:36talking about I think 60 70 percent of the investment in A.I. is actually in the US. And the good news for Vietnam is even though
24:44we're six of the population here in Southeast Asia about 100 million of us in the country half of all the students in the region that
24:51are in the US are from Vietnam. Right. And because of failure payday and sort of all these things a lot of them will end up
24:58working there for a while and end up coming back. Right. And so and so we have this this this this reverse brain drain in a lot of
25:05cases because folks like me who were who went you know earlier and spent most of my life there are also coming back. And so we have
25:12this massive influx of folks coming back built on top of a schooling system that is very very stem and math centric. And as
25:21you know at the underlying the underlying work in A.I. is essentially a math problem. Right. And so I think I think Vietnam is
25:29going to be able to punch above its weight when it comes to more of the technical aspects of A.I. Now of course the great you know
25:37these great great big companies are are democratizing access to the underlying data and models and whatnot. So I think good useful
25:45applications will come out of out of all everywhere in the region. But Vietnam in particular I think has a bit of an edge when it
25:52comes to enough people in Singapore clearly. But from a population perspective enough people to do this. And do you hear a quick
25:58question for you. I know that you're also the co-chair of IBM's Women's Council and I wonder what trends you're seeing and how
26:05generative A.I. is impacting the careers of women in the region. OK. I I do think that.
26:15And I hate saying it but I'll say it. I do think women are getting left behind in this in this journey. The reason is a lot of the
26:25women are not because of education but a lot of women are doing tasks that are more repetitive. They just end up in those kind of
26:34roles. And and it really really breaks my heart and really hurts me to see that women are you know there's a potential to leave them
26:44behind. So we have embarked on this journey in a pack to ensure that a million women and girls are educated on technology and
26:54Jenny. So there is no inhibition. Jenny is not that hard. Right. To use that is not that hard. So we're really putting a lot of
27:03effort in ensuring that women have roles where they are part of you know that the high value roles versus the repetitive tasks because
27:12they will get automated and go away go and go away. So this is a journey I think we and I'll ask for everybody here as a society.
27:21Twenty three percent of the workforce for for all of us in this space are women. And really it keeps me up at night. How do you ensure
27:32that they are not left behind. We have a lot of efforts going on. We're working with the Philippines government. We're working with the
27:37Penang Ministry of Education. We're working with the Indonesian Ministry of Education. We're working of course with Singapore. I think
27:44we're time is like we had launched an initiative as well. So all across this will have to be grassroots. But we've got to enable our
27:51women to to keep up because otherwise it'll be such a travesty to lose such a big talent pool in this. So you know when when I win.
28:02Unfortunately we're out of time because that's such an important rallying cry to end on. And I wish we could all explore that more. But
28:08do you. He just in China. Thank you so much for joining us today. Please find them and continue the conversation if you see them out
28:16and about a coffee later. Thank you so much.