Transcript
Speaker: All right, James, we're back after a week's break.
Speaker: Yes, we are.
Speaker: And what's your excuse, Viggy?
Speaker: I think all our readers were emailing me personally.
Speaker: I don't know how.
Speaker: They want to know, Viggy, where's the pod?
Speaker: I know.
Speaker: I should have thought of an excuse before we showed up here.
Speaker: I purposely didn't bring it up.
Speaker: So we're back with some fresher perspectives and hotter takes.
Speaker: Oh, yeah.
Speaker: Especially hotter takes.
Speaker: That's what people demanded.
Speaker: That's true.
Speaker: Well, now Reddit stock predictions.
Speaker: Did you see what was happening?
Speaker: Oh, my goodness.
Speaker: Oh, my goodness.
Speaker: It's up to 57 now?
Speaker: Oh, man.
Speaker: Okay.
Speaker: So, interestingly, it did go up like crazy.
Speaker: I think it went up a lot.
Speaker: Yeah, tell me more.
Speaker: Tell me more.
Speaker: Yeah.
Speaker: I think it spiked a lot on like day one, day two.
Speaker: And then there's been more of a correction happening in the last couple of days.
Speaker: So, it's back now to $48.
Speaker: My biggest complaint is I usually when I want on a stock or something, I Google it, like my phone's default search engine.
Speaker: I just type blah, blah, blah stock.
Speaker: And they still don't, Google still doesn't realize that.
Speaker: Okay, sorry.
Speaker: But you said that it has, it popped and then it came back down.
Speaker: I see that.
Speaker: Okay.
Speaker: It came out a little bit of a drop and then it goes up to something crazy.
Speaker: Like I know 64 or something.
Speaker: It's back down to 48 right now, I think.
Speaker: Yeah.
Speaker: That's a wild ride right there.
Speaker: I'm waiting for it to be at 25 before I buy sell-out.
Speaker: I mean, it's still above their target.
Speaker: It was 31 to 34.
Speaker: I think they went out at 34.
Speaker: I look forward to covering their earnings releases.
Speaker: They're so wrong.
Speaker: That's true.
Speaker: Yeah.
Speaker: Maybe that's a good time for us to pulse check.
Speaker: maliciously cover their earnings.
Speaker: Tell us about this.
Speaker: All our extensive number of listeners are going to sell the stock after listening to us.
Speaker: Yeah, for sure.
Speaker: Well, we have two interesting topics for today.
Speaker: The DOJ had an antitrust lawsuit that they filed against Apple a couple of weeks back.
Speaker: We'll spend some time diving into that.
Speaker: And then the second topic is what we loosely call the perplexity hypothesis.
Speaker: We can tee this up more in the second half, but...
Speaker: Perplexity is this new, well, they call themselves an answer engine and they're less of a search engine and more of an answer engine, kind of similar to ChatGPT in some ways.
Speaker: There's a pretty open question of, you know, is this alternative approach to search potentially going to replace Google search in the long run?
Speaker: So two pretty meaty topics and we'll dive right in.
Speaker: Cool.
Speaker: So I am teeing up the first topic, and that is the Department of Justice, U.S. government's case against Apple.
Speaker: They're into just case.
Speaker: They're joined by a couple of other U.S. states.
Speaker: I think the one that I most notably call out is Apple's home state of California is also a plaintiff in this suit.
Speaker: So at a high level, before we get into details, the overall allegation is that they're
Speaker: maintaining a monopoly of the smartphone market.
Speaker: They defined two markets, smartphone, all smartphones in the U.S., and then the performance smartphones in the U.S., which is a little different than in previous cases.
Speaker: So they allege that Apple has a 65% market share of all smartphones and, more importantly, a 70% share of performance smartphones in the U.S.,
Speaker: They allege that Apple maintains their monopoly with contractual restrictions and fees.
Speaker: You can kind of read that as App Store stuff via the App Store.
Speaker: And then also that they control how other companies can innovate.
Speaker: via restrictions on their private APIs.
Speaker: They give examples of that, of how smartwatches can integrate, what chat apps can actually message, and also digital wallets.
Speaker: So that's an overall view of the case.
Speaker: And before we get into the hot takes that you alluded to earlier, we should be clear that both of our views here that we express here are our own personal views and not those of our employer.
Speaker: Absolutely.
Speaker: So given that overview, Viggy, what are your initial thoughts when you read through this thing or skim through this 88-page PDF?
Speaker: So let me start with when I've written about the Google and the Amazon antitrust cases before, I did a little bit of tee-up of state of antitrust legislation in the U.S. Maybe it's helpful to just start with a brief summary of that.
Speaker: I won't go into too much detail, but just for our listeners who are maybe new to this topic,
Speaker: Some underlying concepts behind monopoly laws in the US.
Speaker: Firstly, it's not illegal to be a monopoly.
Speaker: What is illegal is to maintain your monopoly status through anti-competitive conduct.
Speaker: That is essentially whenever there is an antitrust lawsuit, the entire premise of the lawsuit is for the government to prove that
Speaker: a company is being a monopoly and not just that but they're maintaining their monopoly position through anti-competitive practices so that is an extremely important nuance when it comes to these cases it is really important to note that it is not illegal to be a monopoly
Speaker: So even if you could make the argument that Apple has whatever, 60, 70% of the market share, that doesn't make them guilty.
Speaker: So in this case, for any US-based listeners, they might look around and a lot of the friends have iPhones or, you know, maybe think back to the school days and how everyone was getting iPhones rather than Androids and
Speaker: And maybe it was representative or dependent on the demographics on the penetration of iPhone there.
Speaker: The fact that so many people had it, you might think that Apple has a monopoly.
Speaker: Maybe they did have a monopoly or maybe they do.
Speaker: But the problem isn't that everyone loves iPhones and they buy iPhones.
Speaker: So Apple has all this market share.
Speaker: That's okay.
Speaker: But it's what Apple might be doing to make sure that they continue to have this high market share.
Speaker: If they continue to have this high market share through innovation, attracting consumers to buy their products, that's fine.
Speaker: But if they do stuff that might impede others from doing that or impede users from switching, that's anti-competitive.
Speaker: Absolutely.
Speaker: And then the other nuance about antiracist history in the U.S., there was this idea of the consumer welfare standard.
Speaker: Essentially, if you've heard some of these takes around
Speaker: if a monopoly increases prices for its consumers.
Speaker: So if you can think about this in a very classic monopoly example, a lot of monopoly laws started with like these oil companies, you know, BP, Exxon, all of these companies.
Speaker: There was a major amount of consolidation that happened like a very long time back.
Speaker: What that means is,
Speaker: If a single oil company, for example, owns a lot of oil companies and everybody needs oil, you can spike up the cost of oil, spike up the gas of oil.
Speaker: So that's essentially like the more traditional view on, you know, the consumer welfare standard and is a monopoly actually hurting consumers.
Speaker: But the spirit of antitrust is essentially, it's not just about consumer welfare in the short term, but it's also about maintaining competition in the long term.
Speaker: You want to have free markets as much as possible, as long as, you know, there is enough competition in the market.
Speaker: It essentially helps the whole society move faster, all of that stuff.
Speaker: So the antitrust laws are meant to be laws that maintain competitiveness in markets.
Speaker: So with that starting point, what do you see in this case that's different than other cases?
Speaker: Yeah, I think that's an interesting one.
Speaker: I will say this.
Speaker: I spent a bunch of time analyzing the Google versus DOJ case a while back.
Speaker: I would say that in that case, there were a lot more.
Speaker: Again, I'm not a liar.
Speaker: I'm not a legal expert, but both of us have a decent amount of understanding about how some of these laws work.
Speaker: And more importantly, a willingness to read really long PDFs.
Speaker: Yeah, exactly.
Speaker: Both of us spent a lot of time reading the Google DOJ antitrust suit.
Speaker: The arguments in that case were, I would probably say that they were pretty slam dunk in a lot of ways.
Speaker: Like, for example, one of the anti-competitive practices the DOJ talks about
Speaker: is, you know, for most of you know that like Android is an open source operating system.
Speaker: So it's not just Google that makes Android phones.
Speaker: There's a lot of these, you know, the Xiaomi's and like the Realme's and all of these other manufacturers, the Samsungs that pick up Android and make these phones.
Speaker: So one of the things that Google actually forcefully made a lot of these manufacturers do is agree to very unfavorable terms.
Speaker: And essentially, they were arm-twisting these manufacturers to agree to these terms and stopping access to Android if they refused to agree to those terms.
Speaker: So that's a very clear anti-competitive conduct that Google engaged in to maintain their monopoly status.
Speaker: So I would say the Google one to me was a fairly clean argument.
Speaker: If you look at the Amazon versus FTC case,
Speaker: It's a slightly more nuanced argument than the Google one.
Speaker: But the FTC's argument, one of the parts of the argument was...
Speaker: Amazon forced how much sellers could charge for their products, both on their platform and outside their platform.
Speaker: They basically enforced that a seller could not sell for a lower price outside the platform.
Speaker: And because Amazon's stake rate was higher, what that resulted in, even if there were other marketplaces that were willing to take a lower stake rate,
Speaker: a seller could not actually price that product cheaper on another marketplace, which is price fixing in some ways.
Speaker: And so I would say in both of those cases, there were fairly clean, you know, anti-competitive conduct.
Speaker: The Apple one to me feels...
Speaker: A little bit more fuzzier, the crux of their argument seems to be Apple is a closed ecosystem, which Apple claims is by design.
Speaker: Apple claims, you know, their privacy first.
Speaker: They want to create a great consumer experience.
Speaker: And a lot of times opening up the ecosystem to other players, other companies, and it dilutes that experience.
Speaker: And the case basically argues that by having this closed ecosystem and by not allowing a lot of other features to interoperate well with other, especially like non-Apple products.
Speaker: And I know you talk more into this, James, so you can talk about this more.
Speaker: But the case essentially argues that that interoperability, the lack of that interoperability is essentially an anti-competitive conduct to maintain their monopoly position.
Speaker: That, I don't know, feels it's an argument.
Speaker: I just don't know if it's as slam dunk as some of the previous other cases that I've seen.
Speaker: Yeah, and our pre-recording banter, it came clear that we had contrasting opinions on how we felt about the case, but we did have something in common, which is what you described at the beginning, the contrast with the...
Speaker: other antitrust cases and like their attention to detail in their complaints.
Speaker: You know, there are some other ones that you didn't list there.
Speaker: But again, we both read these in our personal capacity and we're like interested in it with those details are pretty juicy.
Speaker: And it was, you know, that was, you know, half the fun.
Speaker: And and they were longer.
Speaker: But, you know, I joked about this complaint being 88 pages long.
Speaker: But you like you remove some stuff like a page and a half of table contents and like signature pages and like kind of repetitive formulaic stuff that they have to put towards the end.
Speaker: it actually ends up being fairly short compared to other ones.
Speaker: And so I think that is an interesting dynamic to, they put this first complaint forward.
Speaker: And I feel like I wish I had some, could be on a fly on the wall in the room at the DOJ because the important context here is they spent two years developing this case.
Speaker: It's a long time.
Speaker: Yeah.
Speaker: I guess maybe legally not that long.
Speaker: Yeah.
Speaker: For an individual, that's certainly a long time.
Speaker: I mean, compared to the length of the cases predicted to run, it's supposed to run for three years anyway.
Speaker: So two years prep, three years of ongoing casework, I guess it's all five years to get this thing done.
Speaker: And I do agree it feels light for that amount of effort that
Speaker: for those previous cases, the government had access to tons of documents, were able to pull things out.
Speaker: They had existing stuff, and then they were able to get access to documents, and they were able to update their findings and stuff like that.
Speaker: Presumably, the DOJ already has some of that things, but they were very light on the details here.
Speaker: The only numbers I saw referenced were those 65 and 70% market share numbers.
Speaker: And so it kind of is...
Speaker: If I was on Apple's legal team, I'd be waiting for the other shoe to drop or I'd be feeling very confident that they have nothing.
Speaker: I guess.
Speaker: So let me let me push back on a couple of things.
Speaker: One is that you mentioned performance smartphone market.
Speaker: And again, like nuance around antitrust, like one of the things you have to do in an antitrust case is you have to define a relevant market, which usually also includes like a geographical thing.
Speaker: So in this case, it's the U.S. market.
Speaker: But then it also includes, you know, what is the market that you're talking about?
Speaker: And in this case, they talk about the regular or not regular, like just the smartphone market.
Speaker: And then they talk about the performance smartphone market.
Speaker: And this is always a contentious point in, I think, a lot of cases where you can always counter-argue the Letterman market.
Speaker: This happened in the Google versus DOJ case also.
Speaker: Can you actually argue that Android itself or in-app payments within Android is just like a market, things like that.
Speaker: But performance smartphones to me...
Speaker: I don't know.
Speaker: It seems like a slightly made up market.
Speaker: Like you're basically saying, you know, I'm taking the smartphones market and then slicing it to expensive phones, which seems like a fairly arbitrary way market.
Speaker: I mean, I think my pushback would be we split markets for cars this way.
Speaker: You know, how different is a phone versus a car market there?
Speaker: That's fair.
Speaker: Yeah.
Speaker: Like, I do think for the longest time, people love to make fun of Apple making expensive phones.
Speaker: They've...
Speaker: had some expansion of the types of phones they offer, but they still are fairly at the higher end.
Speaker: So, you know, Mercedes doesn't really make cheap cars.
Speaker: I mean, I guess their lower end models do bleed into some other space, but they're still expensive for the type of car they sell.
Speaker: But I would say that in terms of price points, for example, you know, the...
Speaker: Price difference between a cheap car and a luxury car is large.
Speaker: It's just objectively large in terms of dollar a month versus with an iPhone.
Speaker: Yes, an iPhone is expensive, but, you know, maybe a slightly cheaper Android phone in the US costs.
Speaker: I don't know.
Speaker: I haven't bought one of the non-performance phones, I guess.
Speaker: But like $300, $400 is like what I'm guessing an Android phone would cost in the US versus maybe the lower end of the iPhones cost like $800, something like that.
Speaker: I mean, honestly, I feel like that what you just said furthers my argument here.
Speaker: I do feel like a luxury car in the US is 2X, a regular car, and an iPhone is 2X, an Android.
Speaker: Sounds about right.
Speaker: I do feel like these are different markets.
Speaker: But I guess objectively in terms of the dollar amounts, but fair, I get the argument.
Speaker: I think for me, like objectively it's like 400 versus 800, that's like a $400 difference versus the $30,000 car versus the $100,000, $70,000.
Speaker: Yeah, yeah, yeah.
Speaker: And I think, so I was talking about like the concept of maybe waiting for the other shoe to drop if I was at Apple.
Speaker: And that's kind of something that I'm, I was theorizing before we started.
Speaker: recording, which was that there's no way that they have nothing else.
Speaker: I almost feel like they're waiting to see how Apple will respond.
Speaker: Because I think there's a couple different ways that
Speaker: Apple might try to defend themselves.
Speaker: And we've seen some result of recent court cases that might influence both these parties here.
Speaker: You talked about how there's existing antitrust cases and they can talk about, I think most notably, Apple's closed loop or closed ecosystem versus Android's more open one.
Speaker: And you saw how that dynamic was important in comparing the Epic Games' case about Fortnite versus both Android and Android.
Speaker: and Apple and how because at what I'm citing specific cases here but they don't have necessarily Google and Apple don't have a duty to deal with Epic and and and offer them what they want but the fact that Google at one point did offer you know a different like a a non-integrated experience that meant that Android was actually open to this kind of
Speaker: legal argument, whereas iOS never did.
Speaker: Yeah, I think that's fair.
Speaker: But what I like about these market definitions is that they're broad, actually, because I think what got hung up in the Epic lawsuit is that we
Speaker: When you talk about the smartphone, the performance smartphone markets being made up, it was, you know, I felt even more made up in the Epic lawsuits where they're talking about, oh, there's a market of in-app payments and in-app mobile application distribution, you know, like these are, it's like everyday people are like, what the hell are you talking about?
Speaker: And so I feel like in those, it was even more made up.
Speaker: So I actually do like the broadness of this market definition.
Speaker: Yeah, that's fair.
Speaker: What's your read on some of the specific anti-competitive conduct that they've talked about?
Speaker: I think they've talked about a bunch of things around wallets, around the Apple Watch.
Speaker: Yeah, I'm curious if you might be helpful to walk through some of those examples and maybe also share your read on how egregious or not you think these are.
Speaker: Certainly.
Speaker: So I think we, to mark that transition, you know, we talked earlier about being a monopoly is okay.
Speaker: You're allowed to be a monopoly.
Speaker: It's the maintaining that's not okay.
Speaker: And so the previous, the market definition and those percents, you know, 65% of all smartphones in the U S and 70% of all performance smartphones,
Speaker: That's kind of where they'll have an initial challenge where Apple will try to say that, A, that's not a lot of market share.
Speaker: They'll try to change the market to be global sales.
Speaker: But the thing is, we've demonstrated they have a monopoly by the 65% or 70% number, which I'm not an expert enough to say if that demonstrates monopoly or not.
Speaker: But like fair, let's say that that argument of the relevant market is accepted.
Speaker: Yeah.
Speaker: And so the next thing is, the allegation is that they're maintaining this monopoly through some anti-competitive behavior.
Speaker: And so the DOJ gave some examples, which honestly, I felt like when I was doing some research and I was reading other people's opinions on it, they were outraged or dismissive of it.
Speaker: And maybe this comes from my previous spirit where I was saying that they got to have more than this.
Speaker: Because I agree that the examples they give aren't exactly slam dunks.
Speaker: It feels like picking at nits at how Apple operates.
Speaker: But that they do demonstrate that they affect the behavior in their market.
Speaker: So they talk about the iMessages one as an example where...
Speaker: Yeah, so in their sub-stings, they're saying they control innovation in their market by restricting access to APIs.
Speaker: So that's where the iMessage one comes in.
Speaker: And what was interesting about their argument about that messaging is they say they don't let other apps do SMS on iOS.
Speaker: Yeah.
Speaker: Which is different than saying they don't develop iMessage for Android.
Speaker: I feel like those are two conceptually different things.
Speaker: They're saying no other apps can handle your SMSs on your iPhone.
Speaker: And that kind of inhibits, well, obviously, if you're working at Apple, you're like, well, duh, I want the stickiness of iMessage.
Speaker: You want to keep them coming back to iMessage.
Speaker: But then folks can't make competing messaging apps, right?
Speaker: That's clearly the anti-competitive stuff.
Speaker: effect there.
Speaker: And I think they did a good job of also showing that, well, the impact is broader than that because, and I think this, I was sympathetic to this allegation is, is they're saying as far as impact in their market,
Speaker: they control iOS.
Speaker: Only iOS can send SMSs on iPhone.
Speaker: It doesn't support next generation texting standards.
Speaker: So it's raw, you know, 1980s, 1990s SMS that's running here.
Speaker: And despite any advances they made in iMessage with encryption and things like that, raw SMS doesn't have that.
Speaker: And so that means that Apple is forcing all communication between iPhones and
Speaker: Androids that would use texting to be unencrypted.
Speaker: And I feel like that's showing an example of they're harming the general market where I'm not told, I mean, obviously it's annoying if people make fun of your green text, but I think the actual harm here is, look, there's unencrypted communications out there and it's because Apple doesn't support this.
Speaker: And if they allowed other messaging apps to support this, the consumer might say, I care about my SMSs being encrypted.
Speaker: I would change apps.
Speaker: But they can't change apps.
Speaker: And they don't change apps.
Speaker: And then they're stickier to iOS.
Speaker: And so I think it is showing how they're having an effect on the broader smartphone market by controlling that.
Speaker: Of course, I should say that they said they're going to roll out RCS, which is the next generation of that.
Speaker: But the fact that they are the only ones that can have access to the API is that you're going to be dependent on Apple updating iMessage to port
Speaker: you know, next versions of RCS and keeping up to date.
Speaker: And I think that is kind of an example of they're still going to hold that choke point and they're still going to be able to affect the smartphone market with that.
Speaker: Yeah, I think that's fair.
Speaker: And I do think that both of us maybe have a somewhat unique perspective, you as an active Android user and me as a past Android user that's moved to iOS.
Speaker: Yeah.
Speaker: I do think that the iOS, Android message communication stuff is like annoying as a user, for sure.
Speaker: I think I agree with that.
Speaker: Would you say it's consumer harm?
Speaker: There's no dollar amount associated, but I'm an annoyed consumer.
Speaker: Oh, you know.
Speaker: Yeah, that's a good question.
Speaker: Yeah, I mean, I do think it's inconvenient.
Speaker: I don't necessarily know if, you know, that single thing was a decision factor in terms of, you know, me having moved from Android to iOS or, you know, not deciding to go back to Android, for example.
Speaker: But I think I get that it's like,
Speaker: inconvenience so and and i do think they also bring up another example of the apple watch and smartwatch support as something else which is maybe both for the apple watch on other operating systems doesn't work as well and then other watches on ios also don't work as well so
Speaker: I guess the crux of their argument is not necessarily that each of these things are egregious by themselves, but potentially all of these things have added up in a way that's enabling Apple to have maintained their monopoly.
Speaker: Which is, yeah, why I like the market definition, because if you...
Speaker: I feel like when those previous cases went too deep and they're just talking about in-app payments and at distribution, you're just talking about one part about this, but this is a broader picture.
Speaker: For your example, on the smartwatches, you even kind of verge into a new market, which is smartwatch sales.
Speaker: Obviously, I have a lot of friends that love their Apple Watch, but the specific private APIs that Apple Watch has...
Speaker: you know, access to or like special settings you have to prove as a user when you use a non-Apple Watch with an iPhone.
Speaker: You know, I don't know how familiar you are, given that I can see you're a Fitbit user with your iPhone.
Speaker: I am.
Speaker: That like you appreciate being able to use a different smartwatch with your device.
Speaker: And the fact that there's fewer features, whether you use them or not, is like an example of them, you know, holding things back because they want people to buy their Apple Watch with their iPhone.
Speaker: That's fair.
Speaker: I mean, I wonder, I know we were talking earlier about what is the rationale behind why this case is framed this way.
Speaker: I wonder if this is just, you know, one part of like multiple things that the government is taking to Apple.
Speaker: I know the payments issue with Fortnite that's been going on for a while.
Speaker: I don't think it's fully gotten resolved yet.
Speaker: I do think that there was a ruling where it was basically ruled that Apple does need to support third-party payments, which Apple is doing questionable things by actually supporting it, but not really discounting.
Speaker: They still charge...
Speaker: They charge 4% less or something like that.
Speaker: They charge 30% for their own payments and 26%.
Speaker: So I wonder if this is more kind of arrows being shot at Apple to get them to move maybe both on the payments thing and potentially maybe this is like, maybe they think this is not necessarily a case that the government would win, but maybe this would incentivize Apple to settle quicker and agree on some interoperability.
Speaker: I think certainly that on the Apple's end, it's a centralized pain point of the policy and legal team are feeling this pain.
Speaker: Though I do feel that when it comes from, I guess we characterize that as defense, on the offense side, I do think the actors are different.
Speaker: So while they do care about the outcome, I do think they probably...
Speaker: don't coordinate as much as we would expect them to, you know, except if they're on the same case, like the attorneys general here and, and the DOJ obviously are coordinating, but even in the press coverage, you can see within this case that, that there was different, different opinions being expressed publicly.
Speaker: Like the attorney general, general of California was talking about how they were expecting certain remedies.
Speaker: Whereas the DOJ was saying, uh, that they don't, they don't have yet to, to, to
Speaker: They don't want to talk about what the remedies they want to see are, you know, because the remedies could be something as drastic as breaking up iPhone or Apple, or it could be something that is more similar to Microsoft's monopoly case at the beginning of 2000s, where they had to sign some legal agreement that how they would change their behavior.
Speaker: You know, both those are kind of two ends of the spectrum.
Speaker: So I do see, I think,
Speaker: Apple certainly will feel the pressure in a coordinated way, but I don't know how coordinated the offense is, so to speak.
Speaker: It doesn't make that sense.
Speaker: I do want to talk about one other side of the example because I think there's two other examples.
Speaker: There's two sides to two supporting components to how they're saying they're maintaining a monopoly.
Speaker: One was the controlling innovation.
Speaker: We talked about that a little bit like with the messaging and the smartwatches.
Speaker: The other one is how they are.
Speaker: I want to say how they let me look at my notes to say how they phrase it specifically.
Speaker: They say they harm, so the previously was like control behavior and innovation, and this one is how they harm competition by imposing contractual restrictions and fees.
Speaker: One example they give is that they've more or less outlawed quote unquote super apps.
Speaker: Especially if we target the market in the US, we know in China there's apps like WeChat that have all-encompassing, have lots of things in them, you know, chat, shopping, you know, the equivalent of food delivery, things like that.
Speaker: And so those don't exist in the US.
Speaker: I think I'm more team this is a crap argument on this kind of thing.
Speaker: It's tough, but it's true.
Speaker: I would say if I were to take it further, I would say this is more about, oh, look, their app store restrictions are causing people to not be able to compete with them or they're
Speaker: their placement within iOS and developing apps is harming competition.
Speaker: Yeah, that's fair.
Speaker: The super apps themselves, I'm a little skeptical.
Speaker: Yeah, that's fair.
Speaker: It's really interesting.
Speaker: I was listening to something recently, which was one of the founders in India.
Speaker: He's founder of a pretty large company called Cred.
Speaker: He's founded a bunch of other large companies before.
Speaker: He was basically talking about how...
Speaker: brands, like larger brands.
Speaker: And super apps, I think, is kind of in that angle of it's essentially one larger brand that's trying to do a lot of different things.
Speaker: He was talking about how in countries where a lot of your laws are still evolving, in the US, for example, there's a lot of financial regulation and less evolved legal systems.
Speaker: There's lesser protections, things like that.
Speaker: So in a lot of those countries, there's actually more incentives for larger brands to build trust and have super apps.
Speaker: So while I get the super app argument for why this is limited super apps in the US, there's probably a broader question of does the concept of super apps even make sense to the more evolved tech ecosystem in the US?
Speaker: Yeah, I think I had a similar thought independently where I almost was imagining how if we were restricting it to the US, it's a little tough to say, well, we're bringing concepts of how a certain culture might develop.
Speaker: Culture is still developed today.
Speaker: The fact that certain places might always have two-hour meals versus in the US we eat while walking.
Speaker: That could also exist in how we interact with technology.
Speaker: And so
Speaker: I'm not too surprised that some places are a certain way and some places are other.
Speaker: But I do feel that, well, I'm not necessarily looking to have a super app here in the US, that there is an element of, well, Apple doesn't want a super apps here because they want to be the super app.
Speaker: The OS is the super app.
Speaker: And everything else is like the add-on app that the super app doesn't have because it's too small.
Speaker: So that's kind of what I feel when I see it.
Speaker: MARK MANDELSKI- Yeah, that makes sense.
Speaker: Cool.
Speaker: So we'll see how this evolves.
Speaker: I think there's definitely interesting elements as part of the case, but maybe less as less convincing compared to some of the previous ones that we've seen.
Speaker: Yeah.
Speaker: I say initially, this is a feigned attack to see how Apple will blink or how they'll respond.
Speaker: And so I'm excited to see more or be totally let down when it's just an empty case.
Speaker: Well, with that, we'll dive into the second topic for the week.
Speaker: So if you've been following what's been happening with ChatGPT and related products, there's a new set of products that have come up recently, Perplexity being one of the more popular of these products.
Speaker: There's probably a few different names by which they're being referred to now.
Speaker: There are essentially like these chat-based conversation products.
Speaker: And when you typically talk about chat-based products, you're usually referring to ChatGPT or like Google's Gemini, which used to be called Bart before.
Speaker: Or for my friend out there, Microsoft Copilot.
Speaker: Microsoft Copilot, too.
Speaker: Yep.
Speaker: And then I think there's like a slightly different variation of this that's coming up, which...
Speaker: I would roughly call it answer engines.
Speaker: So perplexity is one of them.
Speaker: They kind of work pretty similar in a lot of ways.
Speaker: Like when you go into ChatGPT, you're asking a question.
Speaker: It responds with an answer.
Speaker: You ask more follow-up questions, and kind of the conversation goes.
Speaker: Or maybe we should say that's one way you might use ChatGPT, whereas for these answer engines, it's more about answering your question.
Speaker: ChatGPT is right.
Speaker: Yeah.
Speaker: Pretend to be your mom and talk to it, right?
Speaker: Yeah, exactly.
Speaker: You could have like a really open-ended thing, but the answer engine is like literally answer my question.
Speaker: That's right.
Speaker: I think the answer engines are specifically built with that purpose of, you know, if you wanted to ask a very specific question, it's, hey, I want a recipe for X or, you know, what?
Speaker: I was cooking something that day and I was like, okay, I want to know what ingredients exist in this particular food.
Speaker: And it kind of gave me like a quick answer or...
Speaker: I had some random podcast I was listening to and somebody claimed that Ukraine is the largest country in Europe.
Speaker: And I was like, wait, is that true?
Speaker: It turns out it is true, like after Russia.
Speaker: So just things like that where you need answers to questions and not necessarily, you know, like a more...
Speaker: a list of search results, for example.
Speaker: So perplexity.
Speaker: MARK MANDELYIERIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIENIEN
Speaker: not exactly an imaginary friend, but like a personality that you can chat with.
Speaker: So I don't know if it would help you like get your feelings out or have help with willingness.
Speaker: Yeah, yeah.
Speaker: And that's like another direction you could go where like chat GPT is like a good human interface and maybe you want to go dive further on the
Speaker: I can talk to you like a human direction and not necessarily answer questions, but like an answer engine, it's like you could start from ChatGBT and you go towards like, this is a differentiated product here, just like an AI therapist could be a differentiated product, but it's more specific than ChatGBT.
Speaker: It's an important point.
Speaker: Yeah, I think on other examples also, if you were putting together, you know, some kind of a document and you needed help writing a document, I think ChatGPT has been useful, not effective, but I guess like useful at those kind of use cases where it's kind of like a sequence of instructions.
Speaker: You're kind of giving feedback and you're like, hey, I want X, write it in like bullet format, stuff like that.
Speaker: So I think for those kinds of experiences, something like ChatGPD has been like more useful.
Speaker: It's general purpose, like its strength is it can do anything because it is general purpose.
Speaker: Whereas these answer engines are, sorry to belabor the point here.
Speaker: These answer engines are more directed towards answering a question or like you're trying to pop on the internet and find something.
Speaker: The answer engines can answer Ukraine, biggest country in Europe.
Speaker: And I think you bring up a good point, which is a lot of the answer engines are predominantly online.
Speaker: So they're essentially using web pages that are online right now and then trying to extract answers out of the content available there.
Speaker: Versus a lot of ChatGPT, for example, is trained using past data.
Speaker: They do have a version that's available for the plus users that is using some combination of ChatGPT plus Bing to have a more live experience.
Speaker: But Perplexity, for example, is primarily online and their primary source of answers is extracting information from web pages.
Speaker: So the broad question that's come up a lot recently is...
Speaker: you know, can something like perplexity potentially replace, you know, Google search?
Speaker: You know, Google search has... Or Bing, again, for my friend out there.
Speaker: Or Bing.
Speaker: Yep.
Speaker: Search has been around for a really long time, Google search.
Speaker: Like the PageRank algorithm, that was probably like Google's single biggest innovation.
Speaker: It's been their biggest cash cow.
Speaker: maybe it's helpful to revisit a little bit of the past attempts at when other products have tried to replace google search and mostly fail so i would say a couple of
Speaker: Instances in the past that's interesting to look at.
Speaker: One is what I would call like the privacy hypothesis.
Speaker: This was essentially like DuckDuckGo, Mozilla.
Speaker: I know Firefox was like primarily like a browser, but I think they were also trying to do some search, searchy things.
Speaker: So I think that was probably like one of the first attempts, which is, can we provide you something that doesn't harvest your data?
Speaker: That's very private, you know, uses kind of like the information within that context and not necessarily who you are, what you like, that kind of stuff.
Speaker: I think that experiment is like mostly failed or at least it's not scaled beyond the point.
Speaker: I tried switching to DuckDuckGo for a while.
Speaker: It lasted like two weeks and then I was like, okay, this is not a significant enough consumer experience that, you know, we're going to give up Google search.
Speaker: So that was probably like the first meaningful attempt at trying to replace Google search.
Speaker: The second one, I know we've talked about this, James.
Speaker: There's been attempts at like subscription-based engines.
Speaker: Neva being one of the more recent ones, which was by the guy that used to run Google ads.
Speaker: They tried this for, I think, three, four years before they decided to shut shop.
Speaker: And they are now rolled into Snowflake and things like that.
Speaker: So that was kind of like the second hypothesis, which is primarily a business model hypothesis of can we essentially turn this into a subscription-based, more highly valuable search.
Speaker: And I think now we're in the third phase of it.
Speaker: And I mean, I would probably say that off all of these things in the past, like this one,
Speaker: feels like a meaningful hypothesis where the perplexity hypothesis or the answer engine hypothesis is essentially that there are certain things that search is good at.
Speaker: Search gives you a ton of results, but there are some things where you just need answers.
Speaker: Google famously introduced the one box, which has been controversial, I guess, a little bit, but mostly useful.
Speaker: And the answer engine essentially just takes that forward significantly where...
Speaker: they are link's second answers first.
Speaker: So that's maybe a bit of the summary of the history, James.
Speaker: I'm curious if you have takes on any of these past or current hypotheses.
Speaker: I think what I liked about your retelling there is like how you said you tried to use DuckDuckGo, but it only lasted a couple of days.
Speaker: And I would...
Speaker: I feel, and clearly I'm an optimist in this scenario, but I feel that I've had the opposite experience with
Speaker: Let's keep saying answer engines.
Speaker: Predominantly, I've mostly been using Google Gemini.
Speaker: But then over the past week, knowing that we were going to talk about perplexity, I've been using perplexity to answer my questions.
Speaker: And ever since I started using Bard like a month ago or something like that, and I've switched now this past week using perplexity, honestly, when I want to ask a silly question or I just want to answer something, my example was I've been watching Suits.
Speaker: And at one point I was like, one of the main characters, Harvey Specter, I was like, he's acting like such a hotshot.
Speaker: And like one episode was about how much he got paid, but they never said the number.
Speaker: And so I was just like, ask perplexity, how much money is Harvey Specter supposedly paid?
Speaker: And then, you know, crawled around and gave me an answer.
Speaker: I didn't really care that like it'd be a hundred percent accurate, but it was like, well, low millions.
Speaker: That's pretty crazy.
Speaker: And then done, you know, but like in the past, I would have to like on that, that kind of query, it'd be like, all right, search engine.
Speaker: type that in, scroll through the choices.
Speaker: This one looks good.
Speaker: Click on it, read what it says and then be like, well, did that answer my question?
Speaker: Yeah, maybe, maybe not, but I don't care enough, so I'm just going to go with whatever it said.
Speaker: So I do feel like at least for
Speaker: like a broad section of the queries that are questions I have that I would jump to a search engine for, that the answer engine is clearly superior and very sticky for me in that sense that I'm like, wow, I'll just ask you, yeah, why would I bother clicking search results?
Speaker: What year is this, you know?
Speaker: Yeah, that's a great point.
Speaker: Yeah.
Speaker: And I think it's also good at follow-up questions and remembering the past context and kind of giving your answers directly.
Speaker: I think I've definitely had a lot of instances where I would scroll, open a link, and then just do a control F for whatever I was looking at.
Speaker: Yeah, because I also feel like even if you are able to handle, you go to a link, you find something you didn't quite give you want, you go back to your search results page, you're still at the previous context of your search without everything you learned at that new page.
Speaker: Whereas an answer engine, it's like it looked at stuff, it gave you the result, and it like...
Speaker: you understand what it says and maybe you have a follow-up question.
Speaker: It takes into account what you supposedly learned with the previous response in its own response.
Speaker: And it's now you don't have that context switching or context backtracking that you have to do mentally today.
Speaker: So it's clearly an improved product in that sense.
Speaker: That's a good point.
Speaker: I guess, do you agree that of all of the past attempts at replacing search engines, you know, maybe this one...
Speaker: comes the closest to actually solving a consumer problem and potentially having a step function change in terms of what value they can add to consumers.
Speaker: I think so.
Speaker: I mean, I think maybe using Gemini and trying perplexity makes me a little less of a normie than I am about the character myself, characterizes myself as.
Speaker: But I do feel, despite my interest in the business of tech and competition and stuff like that, I am still pretty much a normie.
Speaker: I'm like, yeah, Google could be my search engine.
Speaker: Google's on my photos and stuff like that because that's what's easy.
Speaker: That's what's there, you know, that the other products out there.
Speaker: aren't differentiated enough to draw me to them, right?
Speaker: I mean, why they aren't is maybe a more complex subject.
Speaker: But for these, they are.
Speaker: I am not clicking the Google search bar at the bottom of my Google Pixel phone to answer these questions.
Speaker: I am clicking a different app that's several hundred pixels up my phone screen to ask this question.
Speaker: I think that's
Speaker: I think clear evidence that if a normally had their search bar swapped out for an answer engine, I think they might not notice for a while.
Speaker: That's a good point.
Speaker: I'm curious.
Speaker: So I was also digging into what other similar products exist.
Speaker: I found something called u.com, which kind of does something similar to perplexity.
Speaker: Oh, yeah.
Speaker: I remember reading an article about their founder.
Speaker: Okay.
Speaker: I think it was around the time I was digging into Neva.
Speaker: So yeah.
Speaker: Interesting.
Speaker: Okay.
Speaker: Because I think you.com might have been previously a search engine or like maybe they made this evolution.
Speaker: I think they've moved to this one.
Speaker: Yeah.
Speaker: I found something called find, P-H-I-N-D, which seems kind of similar to the perplexity model of being an answer engine in a lot of ways.
Speaker: Over here, I just try to call myself a normie, but I'm writing these down so I can check them out later.
Speaker: I'm curious.
Speaker: Tell me what's your take on this since, you know, you...
Speaker: have more engineering experience than I do.
Speaker: It does seem like perplexity started in 2022, something like that, you know, give or take plus or minus two years, you know, maybe they worked on some stuff before that.
Speaker: I was looking at Neva started in 2019.
Speaker: I think they sold sometime in 2023.
Speaker: So that was like a four-year stint.
Speaker: U.com started in 2020.
Speaker: So we're basically talking about like
Speaker: a two to four year timeline where some of these products have been like fairly quickly.
Speaker: I guess maybe I questioned for you if the hypothesis is true that we're moving forward
Speaker: more towards answer engines away from search engines, or maybe at least like the balance is changing of the use of search engines.
Speaker: Is it easier to build an answer engine compared to search engine?
Speaker: Like my hypothesis was like,
Speaker: A search engine is like you index the entire internet.
Speaker: You literally need to like index everything.
Speaker: Like there's a lot of like search infrastructure that goes into that.
Speaker: There's like crawling infrastructure that goes into that.
Speaker: And you kind of need to parse the internet and like rank whatever of billions of pages that are out there versus with the answer engine.
Speaker: It's not like a boil the ocean problem.
Speaker: It's almost pick the few things that you need so that you can answer this problem.
Speaker: I think...
Speaker: I mean, this is dodging the question because you asked, is it easier to build an answer engine than a search engine?
Speaker: I think that what you need to build either of these products is very similar.
Speaker: I think what's changed is these large language models being prevalent.
Speaker: And the reason I say that is, you know, going back to how search engines worked for the past 20 years.
Speaker: If I would vastly simplify their algorithm, it would be, you know how when you open a page or a doc and you do Command F or Control F and you type something?
Speaker: It's less crude than that because it doesn't have to be a perfect match, but you type, you know, pizza and your PDF reader, your website says one out of 15 things found.
Speaker: So there's 15 mentions of pizza on that.
Speaker: And so like the presence of these words
Speaker: or maybe some other words that you might say are similar to pizza within a page influences how previously influences the relevance people would say a page would have to a given query, my example, pizza.
Speaker: And that's how things would work in the past.
Speaker: And then, you know, at some point, you know,
Speaker: how do you make that better without understanding the page?
Speaker: Well, you could understand user behavior.
Speaker: And so that was the next step.
Speaker: And I think a difficult thing for other startups to differentiate themselves from Google would be that Google, because of the amount of user activity on their page, they could learn that
Speaker: Not only could they have a relevance component based on what they could understand from the page without really understanding it, but they could also look at user behavior to help adjust their rank by saying, oh, past users really like this page, so we should try this one.
Speaker: So broadly, I would say previously search engines would be you crawl the whole internet and have an index of the internet.
Speaker: And then when someone asks something, look for, try to understand.
Speaker: You don't fully understand the page, but you can kind of look for what they searched for and then also look at what past users clicked on.
Speaker: And that's how you choose the best things in the order.
Speaker: And then we come to these large language models.
Speaker: And what's different about them is it's important to remember that they don't really think.
Speaker: But they do have a, they're really good at that inner human computer interface.
Speaker: So as a human, we're, we're looking to understand something like, and so it can both, the human computer interface could both be a page on the internet, understand, like reading the whole thing and then trying to be able to summarize it until it covers X, Y, Z topics, or maybe like a much larger, but you can, you actually understand it rather than saying,
Speaker: you would run it on that page and generate some sort of intermediate representation of it rather than just saying, well, pizza shows up five times.
Speaker: So it's probably rather than... So like keyword base anymore.
Speaker: So you actually understand the pages now.
Speaker: So I think that's something that is new that search engines could actually differentiate from Google now is they don't actually need to combine is that word there and user activity.
Speaker: And that's where Google could be better than them.
Speaker: They could actually understand the page and say, this page is better.
Speaker: Google could still say...
Speaker: well, this page is better and user activity, but the user activity component is a lot less important now because really the user activity was a proxy for was this page actually what they're looking for?
Speaker: So again, I dodged your question saying, is it easier to build?
Speaker: I would say both are easier to build now.
Speaker: It's just now that we can actually understand the page,
Speaker: We don't have to depend on the user to click through to do stuff.
Speaker: The answer engine can go understand the page.
Speaker: Is this page talking about something that I think the user wants?
Speaker: Yes, no.
Speaker: Again, it's really guided thinking because it's not actually thinking itself, but it can understand certain concepts or internalize certain concepts and regurgitate them.
Speaker: So now that we have this ability to understand the page, you can make this product leap from the user picks what they're looking for to
Speaker: This is what the internet says about this topic, user, you know?
Speaker: Yeah, that's a really good point.
Speaker: Maybe one way to think about it.
Speaker: I remember while reading like the Google DOJ lawsuit, like one of the things that they talked about was they tried, Google tried to maintain as much monopoly position in search as possible because the more users used search, there was more data on, you know, what kind of results were relevant or not and which in turn kind of made them better and better and better and better.
Speaker: So what you're saying is like in this new construct, you don't necessarily need that volume of user data to actually have user results.
Speaker: So there's maybe less of an incumbent advantage.
Speaker: Like I'm sure Google has been doing this for a while, but you probably don't always compete on like the scale of usage that you have.
Speaker: Yeah, I would say so.
Speaker: Maybe if we'd imagine that mix of your understanding of relevance of something, plus the previous user behavior, maybe before you would have to, maybe it turns out only 20% relevance and 80% user behavior, and now that shifted to the 80% relevance, 20% user behavior.
Speaker: That's a really good point.
Speaker: Yeah.
Speaker: That makes a lot of sense.
Speaker: So you wouldn't be surprised if another five perplexities showed up in the next three years?
Speaker: As like an enthusiast of using you to answer my simple questions and like the kind of questions previously I would use Accenture in for, and they've been totally replaced by these, I do feel like it's a natural evolution.
Speaker: In the pre-show banter, I said that perplexity would be what Google built if they weren't cowards.
Speaker: Meaning that Google has a lot to lose from the search engine being replaced by something else because it's their bread and butter.
Speaker: It's where they make all their money.
Speaker: And perplexity's only way of monetization to date is the pro subscription.
Speaker: So I could understand why Google would be reticent.
Speaker: But there is an opportunity here if you can find it.
Speaker: I forgot what concept it is, but if your technology is so good at drawing users, that's a good sign.
Speaker: That has to be 10, 20, 100x better to draw users.
Speaker: Maybe it was Satya Nadella code?
Speaker: I don't remember.
Speaker: And I think that's what these answers are, at least for the kind of queries I'm doing so far.
Speaker: So let me poke at this nuance.
Speaker: I think that there are different types of queries.
Speaker: There is informational queries, you know, everything that we were talking about, you know, you're in the largest country in Europe, like that kind of stuff.
Speaker: And then there is queries with actual commercial intent.
Speaker: So this is, you know, I'm looking for X products to buy or, you know, afford X and stuff like that.
Speaker: Like queries which have very clear commercial intent.
Speaker: And...
Speaker: Google today, primarily, I don't think they make a lot of money from the informational searches themselves, which I think is what perplexity and chat GPT and BARD and like all of these more answer engines, conversational products, whatever you want to call it.
Speaker: I think those are good at solving the information questions, but I don't think any of them have actually started solving or I don't see a clear path in which they would actually solve the commercial intent questions very well.
Speaker: I mean, there's also the question of like, how do you show ads and conversations and all that, but like even outside of that, just splitting by informational versus commercial intent queries.
Speaker: How do you see that playing out?
Speaker: Like it's hard for me to understand.
Speaker: If, you know, somebody like said, you would be able to do a good job at like product search, for example.
Speaker: Yeah, I agree.
Speaker: I mean, this is one of my keynotes here is I also wrote those exact same words, probably because we've seen them come up in other coverage where you might try to be defining what a search engine does.
Speaker: And I totally agree.
Speaker: I think that's an important question here.
Speaker: I think I would put, I would characterize monetization of commercial intent.
Speaker: I would say there's probably two ways I'm thinking of.
Speaker: One would be more like discovery monetization.
Speaker: And so that would be what we think about with ads.
Speaker: If you're searching for something like a mattress and you get mattress results on your search result page and that mattress query was monetized by the ad unit itself.
Speaker: So they got paid to show that page.
Speaker: And then the other part, the other part would be like the transaction element.
Speaker: So
Speaker: An integrated example would be like Google Flights.
Speaker: You're searching for a flight.
Speaker: When you buy it through Google Flights, Google gets a kickback on that, right?
Speaker: Which is different because it's at the end of the funnel, you know, the bottom of like your user journey.
Speaker: You get paid for that final action versus one when you're still doing your consideration discovery phase is when you're getting paid for the ad showing up.
Speaker: And so I'm thinking about that for the
Speaker: for the commercial intent and how you monetize them.
Speaker: I do feel like the discovery element is a little bit in threat.
Speaker: Like today, if Perplexity wanted to make money on commercial intent, they probably just needed to find all the shopping style questions and set up an Amazon account with affiliate links and generate affiliate links for
Speaker: for people when they want to buy shit, you know?
Speaker: If I could just say, hey, Perplexity, I need like XYZ cable for my phone and blah, blah, this link.
Speaker: And it's, okay, cool, here's an affiliate link, you know?
Speaker: And I would imagine that Amazon would be incentivized to actually share some of that like catalog with
Speaker: with exactly because yeah so and then i mean i think there's another further conversation of dynamics there like how much shifts there like the value capture because now perplexity is like taking these are straight to the purchase and amazon's not doing it much other than like being the entity that has a relationship with the seller yeah so like maybe there even shifts to be like they might get a higher share of the the sale than than in the past
Speaker: And so I feel like that's an obvious way they could immediately do this tomorrow.
Speaker: But I do, I do have been, I've been working my brain on how do you get ads in here?
Speaker: Because I think that's the main crux is
Speaker: Today, commercial intent queries do subsidize the silly queries, right?
Speaker: But it's a lot more expensive to run these LLMs.
Speaker: So it's doubly important to figure out how do you monetize this both in the earlier on in your answer engine experience with ads and then also at the very end when you finally make the purchase, you conceivably want to monetize all the way through.
Speaker: So those are both important.
Speaker: Yeah, that's a good point.
Speaker: Yeah.
Speaker: Yeah, I think I still find it hard to imagine how a buying experience, for example, could happen within like an answering engine or, you know, within the conversational experience.
Speaker: I mean, I know Baldr or like Gemini has tried some of these like multi-modal type interfaces.
Speaker: But it still seems like even booking Google flights, for example, I don't think it's easy to do.
Speaker: It maybe gives you some results, but you still eventually do actually make a decision.
Speaker: You probably need more information which doesn't fit into a single conversational screen window.
Speaker: So there's some kind of hybrid interface that maybe is like top of the funnel, but you still go down somewhere else for the bottom of the funnel.
Speaker: So I think I'm not fully convinced that like all of the commercial intent queries can be solved through a conversation interface or an answering engine.
Speaker: But I think what you're saying is right.
Speaker: Yeah.
Speaker: Maybe there is some kind of a referral kind of thing where you do send traffic somewhere.
Speaker: Right.
Speaker: So in the middle of you saying that you were saying, I don't know if you can, I'm like right now, I just told perplexity, buy me an iPhone.
Speaker: Buy an iPhone?
Speaker: It is not giving me an option to buy an iPhone.
Speaker: But I do agree with you.
Speaker: I think we had, I think previous conversations about this, where as far as AI is going to eat the world, to some extent, like...
Speaker: There's some things that you just like, give me some paper towels, you know, just buy those.
Speaker: I don't want those here now.
Speaker: But then as a human, we do want to be digging in and have a response.
Speaker: Maybe arguably my water heater is broken.
Speaker: Get me someone to fix it.
Speaker: Right.
Speaker: But then I'm going on a date, you know,
Speaker: I think the most often you care about the experiences is going to be good.
Speaker: Both you and the person you're going to date, they're going to like it.
Speaker: What's the vibe?
Speaker: Have I been there?
Speaker: Does it meet my dietary restrictions?
Speaker: And like some of those you could automate away with an answer engine, but some of it's kind of like the answer engine is never going to, not in the next 10 years, is it going to be able to like have your sense, be able to internalize your sense of vibe or like what I like on a date, right?
Speaker: So like you're going to want to do some of that digging yourself.
Speaker: And I think that
Speaker: And this is where the counterexample to AI Eat the World is.
Speaker: Well, to some extent, it's rediscovering what users want to do again, but through a different avenue.
Speaker: But we already kind of did that manually.
Speaker: And so how much will it actually display, displace like a vertical search engine?
Speaker: So you talked about Google Flights, but there's also Expedia and Kayak, blah, blah, blah, blah.
Speaker: And you can go through the whole checkout flow on a flight there, pick your seat and stuff like that, do everything on those sites.
Speaker: Right.
Speaker: And then similarly, you know, ordering food or finding a place to eat.
Speaker: Like there's websites out there that you can look at these things and you can, in the end of the day, make the own decision for yourself.
Speaker: So.
Speaker: I feel like for some of those, there's some that it's like, just get me it.
Speaker: I don't care.
Speaker: And then there's others where it's like an integration with another vertical search provider maybe, or maybe some of it will need to be homegrown.
Speaker: I think that's what's kind of interesting of comparing the, we previously talked about the retrieval augmented generation rag that was present in Gemini, where they would call out to Google Maps to give results for local queries.
Speaker: That was a really rough version of that.
Speaker: Plexi does something similar for local queries too, but still it's like a rough version.
Speaker: And I wonder how you monetize without monetizing a little bit and then shunting the user over to another.
Speaker: That's a very good point.
Speaker: Yeah.
Speaker: I mean, I think I agree with your sentiment.
Speaker: Maybe there is more top of the funnel, middle of the funnel things that they can do and then the user into where over the bottom of the funnel transaction happens.
Speaker: I do think that the risk for Google is, I think today or like before all of these products, they were the default both for informational queries and intent, like commercial intent queries.
Speaker: And I do think they kind of feed off into each other.
Speaker: If I had to guess, I would guess like maybe, you know, 60-70% of the queries are like informational queries, but 60-70% of the revenue is coming from the commercial intent queries.
Speaker: So there's probably some kind of like a flywheel that's happening.
Speaker: You ask Google dumb stuff.
Speaker: Google doesn't make money off of it, but gives you answers.
Speaker: So you like build a relationship.
Speaker: So when you actually...
Speaker: want to ask commercial intent questions, you're also going to Google.
Speaker: I do think that that's probably a behavior risk for them where, you know, if the user behavior changes where a lot of this informational questions is going first into non-Google places, they potentially start
Speaker: running the risk of losing all commercial inquiries.
Speaker: Yeah, to some extent, that's why I feel like perplexity and other examples are pretty liberated to do what they want compared to Google, where they, it's a classic, can a big company innovate?
Speaker: Because they have a lot to lose from messing with their current search experience.
Speaker: But these other companies have nothing to lose to try to find something that works.
Speaker: And so that's what I find that's really exciting.
Speaker: Yeah.
Speaker: Yeah.
Speaker: That makes sense.
Speaker: Yeah, I think it also probably explains why Google has some of their generative search experiences that they're piloting.
Speaker: I'm guessing they're doing it for a subset of queries.
Speaker: If I was Google, I would probably experiment doing that for the non-commercial intent queries and maybe figure out a more hybrid interface for the commercial intent ones.
Speaker: Last time I'll say it in the podcast, but Bing also does that.
Speaker: Something I did want to point out when I was digging into Perplexity a little bit, like you talked about earlier how they don't have their model.
Speaker: You're saying like, well, they have a crawler, they've indexed some stuff, they can scrape some websites when they need to to answer a query, but their model they use, and you can actually select it in the settings, is different.
Speaker: Like you can choose GPT-4, you can choose like some other ones that people probably won't know from other companies.
Speaker: This is for Perplexity.
Speaker: So which I find is interesting about that is like there is some customization there, but also as an
Speaker: as far as like the difference between chat GBT and like the placement as being a little bit more focused towards answering stuff is also it's commoditizing those models.
Speaker: And so really their play feels like, feels almost that their play is they want to, uh,
Speaker: provide a good consumer experience for answering your question stuff and capture that consumer habit of using them.
Speaker: And they don't really actually care much about the model.
Speaker: I mean, we're not in there knowing what they're actually thinking.
Speaker: But that's one sign I see where like ChatGPT, it's OpenAI makes the model and chat experience.
Speaker: Google makes the model and the chat experience.
Speaker: Same with Apple, right?
Speaker: But what's interesting here is they don't actually care about the model.
Speaker: And then maybe long term, you won't be able to
Speaker: to choose it, but they might forever be using OpenAI or some other company that they fine tune a model a little bit with.
Speaker: Yeah, and I did listen to their founders, I think, on one of the podcasts, and he talked about they do use the GPT models as like the underlying models and, you know, maybe some of the other ones too as a substitute.
Speaker: And I think they do some amount of their own model training, but it's less about the information itself and more about how they present the results and summarize the results and that kind of stuff.
Speaker: So there's like some modeling happening, but I think I fundamentally agree with you that there isn't much of a model per se and maybe a lot of the
Speaker: a lot of the upside is coming from the UX.
Speaker: Yeah, I agree.
Speaker: Yeah.
Speaker: And so, I mean, that's cool.
Speaker: That's their innovation is this UX and the UX brought them there.
Speaker: Fortunately, it's not, like you just said, it's not really a moat, but it would be interesting to see, like, they could still be, if they move fast enough, they could be doing enough to stick with people.
Speaker: I mean, I think another thing that the founder said is they're only aiming for, like, a subset of the world population.
Speaker: They're not trying to scale.
Speaker: And so, like, maybe if they're keeping their thing small, they could really be a tight ship.
Speaker: They don't actually need to scale.
Speaker: That's a great one.
Speaker: Yeah, I do feel it's both an advantage and a disadvantage for them.
Speaker: The advantage is their UX basically lets them do whatever they want and actually compete against Google, who has a lot more UX restrictions and more to lose from search.
Speaker: And at the same time, they also run the disadvantage of UX is not usually a more click of this stuff that you can copy from someone.
Speaker: You kind of need to move fast, get market share.
Speaker: That's more of your play.
Speaker: Yeah, I agree.
Speaker: And I want to bring up before we run out of time, as I feel like we've talked about it previously, but I do want to say that what's missing here in this experience is like, how is this not just in the terms of for all those iPhone users out there, how is this not just Siri that doesn't suck, right?
Speaker: Siri could have been this the whole time.
Speaker: And so that's kind of an indictment of Apple's ability to have this kind of experience and even longer term thinking about
Speaker: how much of this could be done locally or like the models that we're going to get such that we could run this whole thing experience locally.
Speaker: And then that's a kind of a differentiating factor a little bit.
Speaker: And we've also talked about like on-device experiences.
Speaker: Exactly.
Speaker: That could be like a great opportunity where, I know Apple has tried this before with shortcuts and stuff, but this kind of seems like something where,
Speaker: There could be net new UX experiences on your phone.
Speaker: Like I hate going, I hate like searching for apps and then you want to play a song.
Speaker: It's like search for Spotify, go to Spotify, like type a bunch of shit and then figure out what you want to play.
Speaker: That seems like something Siri should be doing from with one command, you know?
Speaker: I know LLMs have only been really popular in the past two years, but if you really wanted to prevent a DOJ investigation into like harm in stopping innovation, you could have released this a lot sooner.
Speaker: But you know, they are behind the curve on this and you know, there's just no data points out there.
Speaker: But that was a fun conversation.
Speaker: I want a lot of ground.
Speaker: We will be back again next week.
Speaker: Until then, if you liked the episode, share it with a friend or a colleague.
Speaker: You can follow us on Spotify or Apple Podcasts and we'll come to you again soon.
Speaker: Bye.
