Flash Coffee & Biofourmis Layoffs, Tech-Enabled Food Thesis & AI Productivity Regulations - E335

· VC and Angels,Southeast Asia,Start-up,Podcast Episodes English


“The person who's not going to take away a job is not AI. It's going to be human with AI. The moral of the story is that you need to, as a human, adopt AI to fight the other person, the human of AI. It seems like a very obvious thing and there's always a job for humans, but in terms of technology waves, you're not fighting as humans of AI is, you're fighting infinite copies of AI, effectively at zero marginal cost with a whole stack of humans who are tailored for that, versus your job even if you're using AI. What I mean by that is you look at technology waves like the horse industry. People had horses for thousands of years. And you had wheel rides, various to help with the horseshoes, folks who rode horses and coaches. Once the car came out, with mass manufacturing, everybody who worked at the shops and manufacturing lines, and the horses disappeared effectively across 20 years. So once the adoption curve kicks off, the whole industry is at risk” - Jeremy Au

“A lot of the ASEAN countries felt like it was a bit early to start regulating. In the EU drafts, there are notions of levels of risk categories, and what is the associated compliance burden depending on what the risk category is. There's a sense that we don't actually know what that will look like yet, and we could just be imposing a really high compliance burden before anybody can do anything. So before anyone can even start trying to build something, they're trying to respond to all those compliance requests, which is my definition of a nightmare. We didn’t specifically talk about regulation at the panel, but we talked a little bit about whether we are techno-optimists or we’re doom and gloom, the robots are taking over. I'm an investor. I have to be a techno-optimist. I believe in human ingenuity, so I believe that people will be able to develop and control this. I think it's really hard to regulate. It's the same thing with the lawyer incident when the lawyer used ChatGPT to provide fake information. That isn't an AI problem. That's a lazy lawyer problem. You can't legislate lazy people away. ” - Shiyan Koh

“There are two parts to tech-enabled services. There’s the technical part of how you run this type of business. Two is, what's the thesis generally between tech-enabled services, especially coffee and consumer goods? There’s something fundamentally true, which is that Southeast Asia could drink more caffeine, sugar, and fat. There are a lot of perspectives, which is that bubble tea, and coffee are all fundamental substances that are tasty, yummy, and addictive, and so I think there is a truth to the matter that as the region gets more well-off, it wants to consume the substances. I'm trying to avoid caffeine and sugar, but the truth is, on a per capita consumption, and this is something that's very common, Southeast Asia is an order of magnitude lower than what you see in Europe, and Europe is the order of magnitude of America right into the consumption. So there's something in terms of the tailwind to say, it's a product, a vitamin, a painkiller. All these things are like painkillers in the sense that you want to consume more and more of it.” - Jeremy Au

Shiyan Koh, Managing Partner of Hustle Fund, and Jeremy Au talk about three main topics:

1. Layoffs and Market Exits: Shiyan and I talked about the recent wave of layoffs: Biofourmis (45 staff in Singapore, and at least 85 staff globally) and Flash Coffee (closed all 11 Singapore outlets). We agreed on the importance of executing layoffs decisively, advocating for a single, well-planned and deep cut rather than a drawn-out process of multiple cuts. We touched upon the necessity of transparent communication, ensuring that the remaining team members are well-informed and supported throughout the transition, and highlighted the importance of agility in uncertain times.

2. Tech-Enabled D2C Consumer: We explored tech-enabled food services and analyzed the industry's financial intricacies, distribution challenges, and competitive landscape. We debated how coffee chains utilize tech not only in processing orders, how they choose locations, and differentiate in markets already saturated with traditional coffee chains. We talked about Luckin Coffee’s unique model, which requires customers to download an app before placing an order, a strategy that streamlines the ordering process and serves customer acquisition and data collection, allowing it to better tailor its offerings and marketing strategies.

3. AI Productivity Transformation: We discussed EU’s aim for AI regulations, similar to its GDPR for data privacy, vs. ASEAN's recently-stated hands-off approach, letting each country decide its own path. We also discussed the difficulty of halting AI development and pointed out the practical issues of enforcing regulations. We were skeptical about JPMorgan CEO Jamie Dimon's forecast of a 3.5 day workweek due to AI. While AI enhances productivity, it also intensifies competition, ultimately transforming entire industries rather than individual jobs. We also talked about AI bots and how they can already emulate the conversational style of specific individuals like Jason Calacanis and brainstormed possibilities like historical fiction podcasts featuring debates between figures like Napoleon and Gandhi.

We also touched upon the NASDAQ performance of the MoneyHero SPAC (Seedly), authenticity and value of human error and spontaneity in content creation, suggesting that there might still be a preference for “artisanal podcasters”, the necessity of continuous learning and adaptation, the ethical dimensions of AI deployment, fostering innovation while ensuring data privacy and the strategies for maintaining balance and humanity in the fast-paced tech world.

Supported by Ringkas

Ringkas is a digital mortgage platform aiming to solve the access to financing problem for home seekers in Indonesia and Southeast Asia. Ringkas currently collaborates with all major Banks in Indonesia and the largest Property Developers across more than 15 cities. Ringkas vision is to democratize home ownership and create more than 100 million homeowners. Don't just dream about owning a home. Make it a reality. Explore more at www.ringkas.co.id 

(02:10) Jeremy Au:

Morning! Rise and shine.

(02:12) Shiyan Koh:

Morning, Jeremy. I should say that to you.

(02:15) Jeremy Au:

I know you're all the way in the New York time zone again. So, lots of news this week, but first of all, we want to give a shout-out to Ian Sanchez. He met at the Hustle Fund Organized Investor Camp at Bali and he said that he's a great fan of our podcast episodes so far.

(02:30) Shiyan Koh:

I mean, I told Ian that I labor under the delusion that no one actually listens to us. So I don't feel too self-conscious when we're on the air but it's always really gratifying to hear that people enjoy it and hopefully get something out of it.

(02:42) Jeremy Au:

So lots of tech news across the region. So, first of all, we have Biofourmis conducting a second round of layoffs. So to quote The Business Times, they are laying off 45 staff in Singapore and at least 85 staff globally. This is the second round. So three months ago, they laid off 120 employees where they operated across Singapore, Boston and Bengaluru. It's also vacating its office space in Singapore by December with the remaining employees working from home permanently. So, this is the second time we mentioned Bifourmis after the CEO departed. Any thoughts that you have here, Shiyuan?

(03:11) Shiyan Koh:

I think we did an episode on layoffs earlier in the year, a lot of people make the mistake of not cutting enough in the first round. And then having to do subsequent cuts and that's actually just like a big hit to morale. And so I would encourage folks, if you are considering layoffs to be more aggressive in the first go around and try to avoid having small successive cuts, because that actually just makes people really demoralized and also makes everyone think whether their own job is at risk and slows the recovery to everybody being focused and back to normal. So I think this is a really common pattern.

(03:45) Jeremy Au:

Yeah. And I think Bifourmis is a unicorn. So I think I was really rooting for the company because we want to see more unicorn exits in the region. So I think not being privy to any information here, this is kind of like a flag that a lot of decisions are being made difficult ones that have to be made. And so, I don't know why it's going to happen, but I think there'll be more news and really hope fingers crossed that, they're able to figure this out.

(04:05) Shiyan Koh:

Yeah, and I think that sort of tees us up for Flash Coffee's exit from Singapore and I think for some, that may have been a surprise given that they had recently announced a raise, a pretty substantial raise in the spring, but as we know, a lot of these raises get announced well after the fact and they had done a pretty aggressive expansion in Singapore with a number of outlets. And I think there are reports of staff not being paid or not being paid completely over the last couple of months. And it should do other creditors. So that's a big one. And I think Jeremy, you'd had some skepticism about these tech-enabled commerce models previously. So I guess I have to say you're right.

(04:39) Jeremy Au:

Yeah, I've discussed it in previous episodes is less about skepticism, more about question marks about what is needed for it to be successful. So maybe, doubling back a bit. Yeah, I think it's right to say that Flash Coffee, they raised like what, a 50M series B earlier in the year and to kind of like, at least underscore it, they closed his entire Singapore country operations. So all of his outlets, over a dozen outlets, I think the last round is 11, but there were more before that. And so they owe 14M so far in debt that needs to be covered to creditors and staff. So let's see how that pans out. I think there's a certain kind of wind down process of country operations, but it means it's still going a move going forward, a going concern. In other words, it's still an operational company in its other markets. So I think that's the conversation.

I think there's two parts of it, that's the technical part of like, how do you run this type of business? That's one, but two is what's the thesis generally between tech enabled services, especially coffee and consumer goods. I think that there is something fundamentally true, which is that Southeast Asia could drink more caffeine and sugar and fat. What I mean by that is that I think there's a lot of perspective, which is that bubble tea, coffee, like these are all fundamentally substances that are tasty, yummy and addictive, and so I think there is a truth to the matter that as Southeast Asia gets more well off, that it wants to consume the substances for myself. I'm trying to avoid caffeine. I'm trying to avoid sugar. I'm trying to avoid fat because I'm trying to be healthier. But the truth is, on a per capita consumption, and this is something that's very common is, Southeast Asia is an order of magnitude lower than what you see in Europe. And Europe is the order of magnitude of America right into the consumption. So I think there's something in terms of the tailwind to say, it's like, it's a product, a vitamin, a painkiller, all these things are like painkillers in the sense that you want to consume more and more of it.

(06:20) Jeremy Au:

So these are fundamentally decent. I think if you're selling like pizza, it's very different from selling coffee with sugar and milk. Like just the hit, the consumability, the pick up and go nature of it. You don't need to sit down space makes those per outlet economics way more powerful, much more efficient, but also I think it's actually very few foods. I can't think of a person eating chicken rice every day or having two chicken rice a day, or three chicken rice a day, but people say that about coffee all the time. It's like, oh, the maximum I can do is three coffees a day with my espresso, or with my macchiato, and I think that is a fundamental tailwind in terms of the substance. So, I can go more into what I think about the business model as well if you want.

(06:57) Shiyan Koh:

Well, first of all, I just want you to know that there's an Instagram account called GUI Peng and it is a guy who eats chicken rice everyday, done it for three years. So some people can but yeah, I mean, are there goods that as people get wealthier, they consume more of? Is caffeine one of them? Okay. I buy that. Is technology really a part of that story? Should these things be given technology multiples or raising venture capital like sums is equity, the right vehicle to do some of this stuff. That's like an open question for me. I mean, in old history, like Starbucks was venture-backed. It sort of created a category of a global chain of this third space as they call it. And so, it has been done, but I guess my question is in this day and age what is it that you're bringing to the table that's so different, that is it a segment, Hey, you're offering at a price that's a Starbucks coffee at this point costs like eight bucks. Who's gonna spend eight times three a day. That seems crazy. And so you're coming in like, hey, this is a two dollar coffee. It's not go, it's like espresso drink. It's Atah's coffee, but still affordable. What's the angle that makes, and then how does somebody get addicted to your thing over all the other things that they're gonna walk past on their way to their office, or leaving their office on their way back to lunch or to, to the MRT or whatever it is. I don't totally get that, but I'm not a consumer person and I'll freely admit that. So I'm kind of curious if you have a different view.

(08:17) Jeremy Au:

Yeah, I totally agree with you. And I think that I have some experience with CPG in the past, and what my point of view is that first of all, I think it depends on the market. Is the market really saturated or is the market is starting to experience the joys ? So, the joke, you can see that for bubble tea, right? So bubble tea started in East Asia first, it became very saturated and obviously you started to see Singapore grow and then now it became very hot. And then, you had this boom and then I pruned a lot. It's downsized, but it's a daily habit right now for many folks. And so, the thing about coffee, of course, is that Singapore GDP per capita is about 100,000 USD, equivalent to the US and higher than the Uk. So the truth is coffee is actually a very saturated market in Singapore. So I think any new challenger, tech-enabled or not, will always have had a tough time competing in that saturated market.

But I think the story is a little bit different for the Philippines. So we have Pick Up Coffee, then we have Flash Coffee and other Kopi Kenangan in Indonesia. So I think their story is a little bit different because their story is saying that we have a rising middle class and historically they drank tea and then they have had only experience with bad coffee. And so this is the first opportunity for them to have a better quality coffee, not the best coffee. They're not trying to go for the luxury or premium, and not even actually honestly going for mass premium, but maybe even mass coffee and then they don't have the opportunity to buy, but now they do. And more importantly, they also want to sew up that distribution player. So the joke that Starbucks used to have, which was like, is this the onion headline where it's like a Starbucks location opened up inside another Starbucks? So the joke was that Starbucks was really aggressive at store expansion. I think there's some truth to it because the truth of the matter is that coffee, it's a distribution play, right? So sales times marketing. Obviously people think about it, but people really underappreciate distribution for physical goods. And the truth for the matter is that you're not going to walk 20 minutes more to get your coffee. That's the awkward reality.

Yes, somebody on the podcast now probably screams, of course I'll walk 20 minutes for my better coffee but the truth is most people who buy it are buying it as a convenience, and that's the mass. And so I think to some level, having that saturation, the right points allows you, and this is again the bubble tea argument. Have the sighting arguments that have the footfall, that you're able to saturate the spot and have perfect coverage because you have both cover not just offline footfall now, but now that you have food delivery, you can also cover online group orders as well. So that's supposed to work together, stitch together. I think the argument is that there's a land grab, which is that if you buy and manage to get all these footfall areas, then you squeeze out potential at new entrants because you've drained the area of the coffee consumer. So I think that's the argument and thesis and there's some level of truth to it because in a wall of locations, Starbucks has a huge team just for figuring out footfall in the right locations anyway.

So there's some truth to it, but like you said to some extent it's replicable because it depends on your competitors as well. So again, I think we want to divide that fall in a row, which is like saturated market, advanced market, but this unsaturated category creation. So I think the former doesn't really make sense in general for a new entrant. And then later, I think there's some pieces for it, but how durable and how enduring that competitive advantage is?

(11:08) Shiyan Koh:

Tough business.

(11:09) Jeremy Au:

Tough business. Actually, I thought that Luckin Coffee had an interesting play because they require you to have an app in order to buy a coffee. So it is like a mandatory step for them. So it's quite interesting because from my perspective is, excluding its prior scandals and misconduct but they've cleaned up shop supposedly since then. But I think the play of requiring customers to sign up is interesting because obviously they've made a trade off internally. They had to say Hey, we are not going to have a lot of customers come in because they're gonna bounce out because of this privacy concern. So people are just not gonna buy. But I think it's kind of interesting because at least now they have a supposedly 100% coverage of their customer and I think in business you always want to control either your suppliers or you want to control your customers. And Luckin Coffee is interesting where I think the argument that they're thinking is that it's better to bounce out all your sporadic people who are weirded out by having to install an app to buy Luckin coffee to saying like, Hey, I'm willing to focus on my core consumers who are really insistent really incentivize, but have them come in through the app and then we give them promotions. So really activate and keep that core consumer. So I thought it was just a very interesting, different approach, which is quite contrary to most everybody.

(12:12) Shiyan Koh:

I haven't studied it deeply. Did they have the app from the beginning or was this something that they implemented later on?

(12:17) Jeremy Au:

I think they've always been very app-oriented because they've always wanted customer information. They wanted targeting, they wanted promotions. And I think that was also a factor in their previous misconduct because when Muddy Waters did that short sale report I think a lot of those codes, promo codes and incentive traffic was being driven through the app.

(12:34) Shiyan Koh:

Good to know.

(12:35) Jeremy Au:

Yeah. I think there's lots of interest still, and generally it just makes sense that as people get richer in Southeast Asia, they want to buy new categories of goods. It's just that, like you said, how enduring that technology advantage and as a result, what kind of capital to raise with those expectations is like the question mark.

(12:50) Shiyan Koh:

Do you have those friends who are like, I'm tired of being an investment banker or a lawyer and I want to open a cafe and you're like, no, please don't open a cafe. It's a very difficult business.

(13:00) Jeremy Au:

There's all kinds of food, verticals. it's not that bad. I think that if you're willing to build a business, it's like I said, labor of love, I think you can have a decent salary and it's just an outcomes perspective. You know,

(13:12) Shiyan Koh:

Jeremy, if you tell me you're going to open a cafe, I'm going to tell you please don't do it.

(13:16) Jeremy Au:

Okay. What's interesting that has happened in I think Southeast Asia is what I call the emergence of food brands. So they, somebody remember those groups that either where the second or third generation takes over the Hawker brand and makes it into a cafe concept.

(13:28) Shiyan Koh:

Yeah. But you have scale. If you have scale, but like, you know when you're just gonna go open one artisanal shop just like the level of effort relative to the return, I think it's just out of work.

(13:38) Jeremy Au:

So you got to tell your friends to be like, Hey, set up a search fund and let's go buy a old grandma, you know, tako itiao, or some sort of pick up and go brand. I like the Popiah one, by the way. I always thought to myself that there should be a Popiah brand pick up and go. It's like salad, but in a wrap. That's like such an easy play in my head. I mean, it's like an Asian burrito, and then now you just modernize it. You add, like, all kinds of unholy condiments.

(14:01) Shiyan Koh:

Why do you need to modernize Popiah? Popiah is already very delicious. Don't, don't mess with a classic.

(14:07) Jeremy Au:

No, they offer the classics for the purists will still sneer at it and go to the Hawker Centre. You know, I'm targeting the bottom of those food courts, you know, those, not food courts.

(14:16) Shiyan Koh:

I'm gonna tell my wife about that. She should add Popiah to the to the roll menu.

(14:20) Jeremy Au:

I mean, that's the same. It's like, what? There's radish, turnip, or whatever you want to have, right?

(14:26) Shiyan Koh:

It's a good question. I'm gonna ask her how come they don't have a bò piàn on the menu.

(14:29) Jeremy Au:

I look forward to the seasonal addition to salad stop. I mean, you can imagine it working. It's the same ingredients.

(14:36) Shiyan Koh:

Yeah, I told her she needed to add a winged bean salad. That would be a really Asian salad, but I think the material cost is really high.

(14:42) Jeremy Au:

And that's the tricky part of all the salads is that the seasonality of these ingredient costs is actually underappreciated.

(14:48) Shiyan Koh:

Okay. Back to technology, Jeremy, no more talking about food. This is not an area. It's not an area we're experts in.

(14:55) Jeremy Au:

Wait, I got to ask you, what's your point of view at least on, I wouldn't say like tech-enabled coffee, but at least tech-enabled food services as well.

(15:03) Shiyan Koh:

I think they're tough because I think you just have to think about where the technology really matters. Is it like top of funnel? Like I'm going to be able to acquire customers more effectively and keep track of them and target them. Is it in OpEx because I'm going to automate like the humans out of it but my sense is that a lot of the robots and stuff are not actually good enough to automate the whole process. So there's still humans involved. And then the CapEx is really high, and most people don't have the CapEx upfront. So then, you see these robotic startups with models where they're doing, Oh, you can treat the robot as a rental.

So it's paying salary except you're paying rental for a robot that does something, maybe not the whole thing. Coffee, though, is one that you would think they can automate most of it, coffee orders have a formula. You're operating a machine, it is a machine, so that's one thing that seems like, hey, you can, and you see them in airports and some hotels and stuff where they're like making espresso drinks and it can be open 24/7 and there is something there. So I think you just have to be really clear in your mind about where the technology is actually acting operating leverage. And it's not clear to me that you could overcome high real estate costs. Do you get so much leverage out of the technology and is that the best use of your dollar? And you definitely shouldn't get tech multiples for it. So I think it's hard but, there's people who made a lot of money investing and building franchises, right? So like scaling out brands, scaling out process and things like that. I think those are like, it can be interesting private equity types of investments.

(16:26) Shiyan Koh:

I don't know anything about them. I sort of live in this sort of strange little tiny world of software and the internet.

(16:32) Jeremy Au:

No, I think what you reminded me about is that for sure, the people who make the most money from this is the landlords from the bubble tea as well as the coffee shops, like they make the most money, because the rentals and every time you see a gain, I think there's a big problem for a lot of these stores is I think the lease. So, you always in that quandary because you always want to have a long girlies because you want to step them up, for example, during the pandemic, I think a lot of folks, landlords were desperate for folks. So I think a lot of folks try to buy longer leases but obviously, it puts you in a situation where if you're stuck, then because you picked the wrong location, then you're stuck there for a long time, versus I think if you have a shorter lease, then you're the other way around, which is the moment the market gets better, the landlord increases rent by 20%, every year for the next five years, because they know you're doing well. And you're like the interesting position so I think that's an interesting dynamic and problem for a lot of distribution folks. On that note I know that you want to talk a little bit about the MoneyHero SPAC on NASDAQ.

(17:27) Shiyan Koh:

Well, I think it's another Singapore based company with a regional business that's managed to make it out into the public markets. The SPAC hasn't performed as well as one might've hoped, it's traded down immediately after it. But it's still interesting to see the numbers come out. It's a scale business. They do sort of 60, 70 million of annual revenue. Still loss making, but managed to cut that loss in half over the last year. And they own brands that I think many listeners will be familiar with, you know, SingSaver, Seedly in Singapore, and then MoneyHero, in Malaysia and Hong Kong. And they used to be the old sort of like, CompareAsiaGroup that's just kind of been rebranded. So, I think SPACs have been super out of favor they tend to get dumped on retail which is not great but I think for the company at least, it's trading, it's liquid. I think the challenge from here on out is just making sure you bring this thing to profitability, you grow the analyst coverage because you don't want to be in micro-cap purgatory and kind of continue to grow into that, but at least you are now like liquid and publicly-traded.

(18:19) Jeremy Au:

Yeah we put some numbers here is, it's the deal was valuation at 300 mil. As of today, October 22nd, it's currently about 60 million in terms of valuation. And they had an exit proceeds of about 100 million from the SPAC to keep growing. And for the year of 2022, it had an unaudited net loss after tax of 49.4 million. So these are some numbers. Personally, I've always enjoyed reading Seedly. So I think on a previous brief episode, we had Kenneth Lou who was the founder of Seedly, who sold to MoneyHero Group and is currently the founder of Mito Health, which is longevity startup. So I thought it was just interesting that it's actually a resource that I actually check almost every week, to be honest. Interestingly.

There's a lot of personal finance stuff that, I guess, there's no NerdWallet, by the way, I know it's something has a flashback to the fact that you are a big, not wallet executive, but there's a lot of PC for information in Southeast Asia. I mean, it's just a small market, so I guess I understand it. So it's just trying to understand. I remember reading like before I had a kid, I was like, what are the projected expenses of being a parent in Singapore? And I'm just like, I read it.

(19:25) Shiyan Koh:

And endless! As many as you want. Something you could spend endless amounts of money on.

(19:29) Jeremy Au:

Yeah. So the funny part is they actually had, I think the scenarios, which was like the base scenario, then there's a cheap scenario and there's an aggressive scenario, so exactly like what you said, took the time to be like, okay, this is what the lowest you can probably take. And there's the most aggressive version probably of this. And I thought it was a wonderful set of articles. Yeah, I enjoyed it. I guess you don't, I guess you're totally

(19:48) Shiyan Koh:

No! No, No. I mean, I worked in personal finance for many years, and so I spent a lot of time reading and thinking about these things. But I have to look up CPF things too, because, I'm not as familiar with the CPF system. And so when I came back, I had to look it all up. So I'm with you. So I think like any, going back into like tech and software and things like that. It's great that they got out. They've got work ahead, but management claims there's a path to profitability and, all you listeners, keep getting mortgages and car and credit cards and help them get there. I do think the one thing they pioneered, would you see this market, which you don't see in The U. S. market is the gifts, so, they make money through you would transact on the recommendation, and then you open the bank account or get the credit card or whatever, and the issuer will pay them.

In the early days, there wasn't really much of an affiliate market. And so they actually offered some of the rewards themselves, where they would give the gift, like, you can get the AirPods, or you can get 100 back as a way to convince consumers to use them.

(20:39) Jeremy Au:


(20:40) Shiyan Koh:

And they grew that, and then were able to develop affiliate relationships with issuers. So, I thought that was like a cool part of their story.

(20:45) Jeremy Au:

Yeah it does remind me of somebody who is waiting on their bicycle. So I think there was a promo and they'll get a folding bicycle. And then they bought a financial product for, not for the folding bicycle, but it was a big part of it. And I think that was quite interesting.

(20:58) Shiyan Koh:

Is that someone you?

(20:59) Jeremy Au:


(21:00) Shiyan Koh:

Who is this someone?

(21:00) Jeremy Au:

I already have my own folding bicycle. I bought it in cash. And if I had known I should have signed up for a credit card instead.

(21:07) Shiyan Koh:

Okay. Just the way you said it, you're like, I know someone, I know a guy.

(21:11) Jeremy Au:

No, it's not me, but it's someone we both know. Right. Okay.

(21:14) Shiyan Koh:

Oh, okay. Discuss this offline.

(21:16) Jeremy Au:

I would discuss this offline. Yeah. I mean, it's not a bad decision, like you need a financial product that you're like, you might as well just get it. That has reminded me, I did sign up for the Ed Sheeran concert tickets coming to Singapore. So I had to download the KrisFlyer app, and then, sign up and do one activity, but it was very reasonable. So don't get me wrong. I'm a sucker for all these promos, just got to find the right one for my wife and I'd have a date night.

(21:37) Shiyan Koh:

Nice. Nice.

(21:39) Jeremy Au:

I think lastly, we organized the Harvard Business School and Stanford AI event panel. So you were there with Jefferson, right? So love for you to share a little bit more about what you thought.

(21:49) Shiyan Koh:

Yeah, thanks for inviting me to do the panel and then ditching me, claiming you're reservists.

(21:54) Jeremy Au:

I organized it and then I peaced out because I look, I had the army reservists, I couldn't come out for real. I really wanted to be there because Jefferson is that advanced intelligence group as well, told me by not pay later, a lot of great approaches and products

(22:06) Shiyan Koh:

Yeah I think it was a super fun event, it was pretty well attended, I think we had a hundred-plus people show up and, I think we did some sort of covering of what is AI? What are we talking about here? Why are people excited about it? Haven't we been talking about AI for decades? Why is now suddenly this surge of interest? And, where should people be spending their time and money, essentially? And then I think there were questions from the audience that were like, what are the implications of this for what my kids should study? Should they still study computer science if AI can write code? There were lawyers who were afraid. They were, we heard about lawyers using chat GPT who then hallucinated fake case law. What regulations can we impose to stop this from happening? And it was, I think a really lively conversation. And I think there's also a conversation around like. Given the US-China trade war and the different monoliths being built, like what's the difference between the US models or the Western LLMs and the Chinese LLMs and how do we think those will evolve? So it was a pretty wide-ranging conversation. It was super fun. And yeah, thanks to the HBS and GSB clubs for putting that together.

(23:06) Jeremy Au:

Yay. Thank you. Yeah. I mean, also a big shout out to Joanna for putting it together as well. She was a big driver of that as well. So I think what's interesting is that from my perspective as well, there's also been South, Southeast Asian news that dovetails with that, right? So I think there's a big announcement that ASEAN basically, I think there's a big news where the EU basically said, Hey, we want to be the leader on AI regulations just like we were for data privacy, for GDPR. And so we want to use our large block of countries to set national standards on AI. And then ASEAN basically said nope. So I was in basically wrote a report saying we want to be more hands off and we're going to let every country do it their own way. We're just interesting. I think interesting part about it. And if I sit down, the fundamental pieces of it is that privacy is about, who has access to the data. So is it individuals, the government, or corporations.

And I think GDPR broadly currently says like, Hey, we want to focus more on individuals, but allow governments to access it. And then when it comes to AI, it's just such a very different context because it's generating content of copyright. So that's actually an interesting differential where some countries have more to lose because they have more copyright materials, for example, being scanned and used. There are some countries that's where it's more beneficial because they're benefiting from the generated outputs. So I think it's an interesting dynamic where it's not as easily to get consensus, I think, at least in this current stage.

(24:21) Shiyan Koh:

Well, also, if you read the comments, I think a lot of the ASEAN countries felt like it was a bit early to start regulating. So, like, in the EU drafts, there's notions of levels of risk categories, and what is the associated compliance burden depending on what the risk category is. And I think there's a sense that we don't actually know what that will look like yet, and we could just be imposing a really high compliance burden before anybody can do anything. And so before anyone can even start trying to build something, they're trying to respond to all those compliance requests which is like my definition of a nightmare. So I mean, we talked about not regulation specifically, but at the panel a little bit around are we techno optimists or are we like doom and gloom? The robots are taking over. I mean, I'm an investor. I have to be a techno optimist.

(25:04) Jeremy Au:

Is this saying that the pleas of future robotic overlords?

(25:07) Shiyan Koh:

Exactly. Let this be on the record. Dear future robot overlords, I'm pro. think, I mean, I believe in human ingenuity, so I believe that people, we'll be able to develop and control this. And I also believe maybe the flip side of that. I think it's really hard to regulate. Do you remember early in the AI craze where people were saying, let's sign this petition to stop development?

(25:29) Jeremy Au:

I think yeah, It kind of died down.

(25:30) Shiyan Koh:

Yeah. Well, exactly. How are you going to stop them? And it's the same thing with the people asking about regulation on some of this stuff, like the lawyer thing. The lawyer used GPT to provide fake information. And I was like, this isn't an AI problem. That's a lazy lawyer problem. You can't legislate lazy people away. That's not possible.

(25:47) Jeremy Au:

That's true. You can't give, or legislate away the tool that happens to work for both lazy people and hardworking people.

(25:52) Shiyan Koh:

So, I think there's just like some of the practicalities of these things that, it's like, you can make a legislation, but if the law can't be enforced, then what's the point?

(25:59) Jeremy Au:

Talking about lazy people, you shared an article about how Jamie Dimon claims that AI will allow a world where everybody has three and a half workdays, 3. 5 workdays instead of five workdays per week.

(26:11) Shiyan Koh:

I mean, I think it's nonsense. I think you just hire fewer people and make them work five days a week. I

(26:16) Jeremy Au:

Or six days a week. You know, you pay the five days a week.

(26:19) Shiyan Koh:

I mean.

(26:19) Jeremy Au:

But you know, effectively they're working six or seven.

(26:22) Shiyan Koh:

Yeah, I mean, I don't know, but maybe people are happy being paid for three and a half days a week and working three and a half days a week. Maybe there is a state where that is, but I think in general people would rather be paid for five days a week and work five days a week. I don't

(26:32) Jeremy Au:

I think, I think what's true is that the emergence of hybrid and remote work from home has allowed for the creation of jobs that are more part time because you pack it more, you travel less. I think that's true. I think AI as a trend to say that it's going to create a shorter work week feels like the opposite because it seems to me that from my perspective is that productivity enhancing technologies does enhance productivity, but also increases relative competition. So I think interesting dynamic I was reading was like, apparently farmers like had a 20 hour work week on average of, I was reading this interesting study. Obviously this is during times of, plenty or average time, obviously during a famine. It's terrible. But there's a lot of holidays, a lot of break, but then as with the rise of each productivity technology and what becomes more and more industrialized, more standardized.

And so, you see the emergence of factories, and then you see so forth. So there's an interesting, historical claim. Obviously, it's not saying that's true for AI. AI could break the formula, obviously. I'll quote the study in the transcript later. But I thought it was an interesting analysis to be like productivity technologies, increased productivity on absolute basis, but creates competition across producers.

(27:34) Shiyan Koh:

Yeah, people who can use it more effectively.

(27:35) Jeremy Au:


(27:36) Shiyan Koh:

Basically put pressure on those who don't have it or aren't using it as effectively.

(27:39) Jeremy Au:

And I always hear this phrase people say is like, the person who's not going to take away a job is not AI. It's going to be human with AI. And I laugh at that because it seems like the moral of the story is that you need to, as a human, adopt AI to fight the other person, the human of AI, so it seems like a very obvious thing and there's always a job for humans, but I laugh because in terms of technology waves, actually. You know, you're not fighting as humans of AI is, you're fighting an infinite copies of AI effectively at zero marginal cost. With a whole stack of humans who are tailored for that versus your job. Even if you're using AI and what I mean by that is like, you look at technology waves like the horse industry, right? I mean, people had horses for thousands of years. And you had wheel rides, you had various to help with the horseshoes, you had folks who rode horses and coaches and so forth. And then once the car came out, with mass manufacturing, everybody working the shops and manufacturing lines. You know, horses disappeared effectively across 20 years, so, I think once the adoption curve kicks off, the whole industry is at risk, not just like, you just can't adopt AI on an individual basis to compete against a whole corporation that reinvents what's


(28:43) Shiyan Koh:

I think that's right.

(28:44) Jeremy Au:

I mean, it comes to the podcasting as well. I was like recently, I was like mind blown because there's so many companies now, which are trying to create. I was mindblown recently because, on the podcast side, you and I are humans, two humans. And for every,

(28:54) Shiyan Koh:

Are you? Are you real? Is it really you, Jeremy?

(28:57) Jeremy Au:

I'm already having my three-and-a-half-day work week in Bali. And this is my AI replica talking to you. It's so low value at any rate. But I think that the crux of it was like, people are like, okay, Jeremy, you just had to adopt AI to compete against all these new, AI-driven podcasts. This is not a human of AI, but I'm meeting startups now that are like basically saying, we want to create a million podcasts. And every day it generates like one or two or three episodes at zero marginal cost-effectively. And it can be auto-updated and it gets auto-produced, like they close the whole loop cycle. And I'm just like, obviously one million is a catchy phrase for that. But in my head, I was just like a single human using AI is way less effective than somebody, a corporation just using AI to generate a million versions of that, right? So I thought it was interesting just to have that.

(29:37) Shiyan Koh:

Can you do one? Can you beta-test it? Can you generate an episode that's supposed to be us but isn't us?

(29:41) Jeremy Au:

Yeah, I can. You know what? I'll do that for next time. Yeah. So, they already have that. So you look at this, we can start off Jason Calacanis. They already have a AI bots. So basically like they take a script and then they use a chain based AI based on what she and would say or what Jeremy was said. So you could quiz the bot.

(29:58) Shiyan Koh:

I'd be curious. Yeah, I'd be curious.

(30:00) Jeremy Au:

Right. WWJD, right? What would Jason Calacanis do, right?

(30:03) Shiyan Koh:

I would be very curious.

(30:05) Jeremy Au:

Yeah. So it's an interesting dynamic. I don't know what the answer is. Here, but yeah, I don't know how to answer this. Let's just WhatsApp each other all the time. And is this converted all into a podcast script, right?

(30:16) Shiyan Koh:

Yeah, that would be interesting. If you could take an async conversation, like basically take our process today and you could feed it into a model and it would just produce a podcast, that'd be kind of crazy.

(30:26) Jeremy Au:

Yeah, it'd be crazy, but we'll still be losing against other version. There's just an infinite personas, like it's only a matter of time. Would you listen to a podcast? There's a debate between, I'm just talking about Napoleon and Gandhi. Like, I can already imagine that, right? It's just like a whole narrative sketch of, I don't know. I'm just brainstorming here, like the time context is Napoleon. They're doing invasion of France. Gandhi's an advisor. You could create a whole narrative voice with the right voices, the right personas in terms of the text. Obviously it would be very flawed from somebody who knew Napoleon and Gandhi in person, but nobody today is alive enough to know who they are in person anymore. So, you and I would be like, it sounds like Napoleon. It sounds like Gandhi. And that would be the kind of podcast is now possible. That wasn't possible.

(31:04) Shiyan Koh:

Yeah, that's like a genre of podcast like, historical fiction. But, I don't know, I think, maybe it's crazy, maybe I still think people like to know that there is a real person behind something. And maybe we're just gonna go, we'll be like, artisanal podcasters.

(31:17) Jeremy Au:

Yeah, it's like, it's not that.

(31:20) Shiyan Koh:

It's handcrafted.

(31:21) Jeremy Au:

Handcrafted hallucinations. I want Jeremy just to lie and make mistakes and not know his numbers instead of a AI hallucination. Is this a human mistake? so much more umami than the AI hallucination, right?

(31:32) Shiyan Koh:

Maybe, but also, like, you could, you can actually interact with us in real life.

(31:36) Jeremy Au:


(31:36) Shiyan Koh:

You can run into us at a coffee shop that is not Flash Coffee. You can run into us in Bali or in Singapore and New York and have a conversation with us.

(31:45) Jeremy Au:

Oh, that's so sweet. Yeah.

(31:47) Shiyan Koh:

Or maybe people won't care and they'll be like, no, I would just rather talk to the AI version of you because it's always available.

(31:53) Jeremy Au:

Yeah, the AI version of Jeremy doesn't wear shorts. It's professionally dressed all the time. On that note, thank you so much, Shiyan. It was so much fun, as always. I'd love to summarize the three big conversation topics. The first, I think we did we talked about the layoffs and Biofourmis as well as Flash Coffee. So I thought it was just interesting to talk about that and reference our past patterns and points of view on layoffs and how to do them in one cut rather than two, if possible.

The second, of course, was that deep dive into tech enabled food services about the margins, the bouquets, the barricades, and I think some of the concerns that we have around distribution technology and competition.

Lastly, I thought it was fun to discuss AI a little bit and talk about some of the questions that are still open for both of us. I thought it was interesting to discuss regulation, but also how to think through the economics of what like you said, robots, for example, for salads would be, but also what robot production for generative AI will be as well. you, Shiyan.

(32:44) Shiyan Koh: