Latest / 1 to 100 / E43: How to Survive The SaaS Sell Off with Dipti Desai

Transcript

E43: How to Survive The SaaS Sell Off with Dipti Desai · 1 to 100

• 0:00
  So that the, I don't know if you've seen any of the episodes,
• 0:02
  but we just talk about SAS basically.
• 0:04
  Yeah, I was going to say, is the name changing?
• 0:06
  It's getting hammered. So this is going to be the AI
• 0:11
  podcast now. Yeah, I got into a lot of
• 0:15
  trouble actually over the last 48 hours.
• 0:18
  Because of the name you mean. I publicly posted something
• 0:21
  about how the SAS sell off is not violent enough.
• 0:25
  What did people say when you said this?
• 0:28
  Yeah, I made a comment on acts about like how I just don't
• 0:34
  believe that traditional SAS pricing is like a way to
• 0:40
  basically it's like the sell off in the public markets for big
• 0:43
  trat SAS is like not severe enough because I just don't see
• 0:47
  how you can endure very large revenue over time.
• 0:52
  And there are like other, I think there are other better
• 0:55
  pricing models on a go forward basis.
• 0:58
  And then I made the comment that like given this reality, like
• 1:01
  even we are shifting. And then a bunch of our
• 1:04
  customers saw that and then they reached out to their account
• 1:06
  managers and they were like, your CEO said this thing, what
• 1:09
  does this mean for us? And then I had to do like
• 1:12
  internal cleanup and then external cleanup.
• 1:14
  And anyway, it was a whole thing.
• 1:16
  But yeah, I just don't see how like bold trad 1.0 SAS survives
• 1:22
  over the last over the next two years.
• 1:24
  Like I think you, you have to have like one of the following.
• 1:28
  You need to add fintech compute or data like those are the only
• 1:34
  ways to actually survive. But that's just my opinion.
• 1:39
  Specifically, so we don't assume different things.
• 1:42
  You're specifically referring to the per seat pricing aspect of
• 1:45
  it or is there more to it? I think, I think, well, I think
• 1:51
  three things. One is I think like perceived
• 1:55
  pricing is going away. The second is I think software
• 1:58
  that is like workflow software that is not underpinned with
• 2:04
  unique data of some variety is going to be hard to retain.
• 2:11
  And then I think the traditional way you sell SAS, so like you
• 2:15
  hire a big sales team and you do like calling and then you sell
• 2:20
  the SAS product. I don't think that's going to
• 2:22
  retain either. And so that that was actually
• 2:26
  what I was mostly talking about is like those things seem seem
• 2:31
  unlikely to like continue to hold like I don't think you can
• 2:37
  keep build a public company that way.
• 2:39
  I just don't think it's possible.
• 2:41
  And I think the public markets are of that view.
• 2:44
  Yeah, I don't know what you guys think.
• 2:46
  I actually think that your thesis is true if companies
• 2:51
  continue to operate like based on the status quo, just even
• 2:56
  from like an operational like expense point of view.
• 2:58
  But also what also is becoming true is that all of these
• 3:01
  businesses are becoming a lot easier to operate and a lot less
• 3:06
  cost effective to operate and maintain and keep live.
• 3:09
  And so like, I actually think there's probably like a bridge
• 3:11
  reality in which a well run sort of like 1.0 size company can
• 3:16
  just become super efficient over like the next few months and
• 3:21
  like years and become like a really, really viable like cash
• 3:23
  cow. And so like there's, there's
• 3:27
  survival thing. It kind of depends on like where
• 3:30
  you set your ambition to be. And I, I also think, I don't
• 3:36
  know, like I, I do just feel like we are so much like in a
• 3:39
  bubble. I, I was this last week.
• 3:41
  I was like a ski retreat with like one of it's like business
• 3:46
  groups have a part of called Hampton that has like a bunch of
• 3:48
  different industries in it. And I would say like most of the
• 3:51
  people that are in Hampton are also like pretty digitally
• 3:53
  native businesses, even if their core business is not super
• 3:56
  digital. And I just became like super
• 3:59
  like clear to me that there is just like a lot of noise still
• 4:04
  with this whole thing. And I, I don't think it's going
• 4:08
  to be like dramatic as long as the business is like, again, not
• 4:11
  just going to be like, oh, status quo as usual, like, can
• 4:14
  you just make mildly decent decisions?
• 4:16
  I don't think that there's like a massive issue towards the
• 4:18
  model. Yeah.
• 4:20
  I, I think we roughly agree, because my point was more about
• 4:25
  the public stock market value of these companies, which you have
• 4:29
  to discount future cash flows on, which means that like you, I
• 4:33
  think we're saying different things.
• 4:34
  You're saying like can you make money from these businesses?
• 4:37
  I think the answer is yes. And I actually wrote that in one
• 4:39
  of the responses. I think you can make a lot of
• 4:41
  money from these businesses if you run them well.
• 4:43
  I just don't think you should buy the stock.
• 4:46
  And so like I was just, I was mostly commenting on the sell
• 4:49
  off in the public markets because I think it's really hard
• 4:53
  to discount the future value of these types of companies because
• 4:58
  you don't know over what time horizon it just slowly depletes
• 5:02
  basically. So if you believe that to be
• 5:04
  true, do you think that potentially Fermat has been
• 5:07
  capitalized in a really non efficient way?
• 5:10
  Because like you've got you capitalize future cash flow for
• 5:12
  not based on like not public market comms, but like the VC
• 5:16
  route that public market comms got to you know, from that path.
• 5:19
  And so like, how do you think about that?
• 5:21
  Yeah, Yeah, That was what I was saying.
• 5:22
  Yeah. Yeah.
• 5:24
  This was literally the question that I was asked that got me in
• 5:26
  trouble. And I responded, I'm going to
• 5:29
  give a response that's going to get me in less trouble.
• 5:32
  I think that if all we did was continue to build SAS, then
• 5:35
  yeah, it's not, it is. It is incorrectly capitalized
• 5:41
  and you cannot underwrite the valuation.
• 5:43
  That is, that is true. Oh, you got to finish.
• 5:46
  That makes sense. Where do you go?
• 5:49
  I literally can't. I literally can't because this
• 5:52
  is how we get it. This is how I get in trouble.
• 5:54
  But like, Zach's point is correct is all I'm trying to
• 5:57
  say. Like like what we've raised the
• 6:02
  beyond the world has changed. We raised our B last year
• 6:06
  roughly at this time plus or minus, right.
• 6:09
  And the world has moved and there were elements of what we
• 6:13
  raised on that are still true and are going to produce the
• 6:18
  future enterprise value of the company.
• 6:20
  But the sort of like SAS sales motion that we grew very fast on
• 6:25
  leading up to the B does not hold the same enterprise value
• 6:29
  anymore. That is for sure the case, yeah.
• 6:32
  And both in my opinion, and I think like, I don't think like
• 6:37
  anybody would disagree with that, basically.
• 6:39
  What do you all think? What is your view of like the
• 6:41
  future of SAS 50? I mean, to me still frankly, SAS
• 6:45
  is a delivery model, right. I think, I think that the legacy
• 6:50
  companies that don't rest on their laurels will, will make it
• 6:55
  whether they're a SAS company or not.
• 6:56
  Companies that fundamentally can't get off of.
• 7:00
  I actually think a lot in terms of at least when I look at our
• 7:03
  our world and our space, I actually evaluate a lot of
• 7:07
  legacy companies based on their ability to morph their data
• 7:11
  models towards an agenda future and assign sort of a likelihood
• 7:16
  estimate on can they or can't they, if that makes sense that
• 7:20
  that's a lot. Completely, I completely agree
• 7:23
  with that. I completely, by the way, just
• 7:25
  yeah, what what is sorry, before we jump ahead, like what does
• 7:28
  the business do? And yeah, how, how like what,
• 7:31
  what world does Crystal work in? And yeah, maybe share a little
• 7:36
  bit about the kinds of data that you're thinking about.
• 7:39
  Yeah, for sure. So I'll keep it brief.
• 7:41
  And then, right, so, so Crystal, depending on who we're
• 7:45
  explaining kind of to what Crystal does different contexts
• 7:48
  make sense. But for this audience, one way
• 7:50
  to think about what Crystal is, is it's a system of action for
• 7:54
  beauty transactions. So we are focused very much on
• 7:57
  the B2B trade, on the B2B commerce side of things and sort
• 8:02
  of our wedge and our hero use case was great.
• 8:05
  This is going back a few years. I was 14 etcetera had played out
• 8:09
  customer acquisition costs were rising.
• 8:12
  Covad was still a thing and brands were basically starting
• 8:17
  to old, became old was becoming new again, right?
• 8:20
  Brands were starting to diversify.
• 8:22
  They were looking to go on in general, but what they were
• 8:24
  finding is a lot of the tools at their disposal were old.
• 8:26
  And this was even pre ChatGPT, right.
• 8:28
  So this was the API economy was now, you know, sort of front and
• 8:34
  center low code. No code was was a thing, which
• 8:37
  obviously has changed a lot obviously in the world of AI,
• 8:42
  but that's that was sort of the the genesis of Crystal, right?
• 8:45
  Is, is back kind of that 2022 time period.
• 8:50
  And so a lot of our initial customers have been customers
• 8:53
  that started out on Shop 5, Shop 5 Plus and sell them to Costco
• 8:58
  and you know Walmart, Target distributors and things like
• 9:01
  that. And then today really we've
• 9:04
  become a really powerful platform for buyers and sellers
• 9:10
  of all shapes and sizes, but squarely focused on on B2B.
• 9:15
  And so we obviously see some very interesting data and sort
• 9:19
  of the context layer and the mission critical workflows
• 9:21
  around depending again on who's sort of looking at this, they
• 9:25
  may resonate with sort of the order, the cash flow side of it
• 9:29
  or the procure to pay, you know, side.
• 9:32
  And so, so there's all of those elements really blended in and
• 9:35
  and sitting at the core of. One of the things that when we
• 9:38
  last caught up that at least clicked for me was that it is
• 9:43
  like fundamentally the hard part here is that there's a lot of
• 9:47
  data that is like held in a bunch of different places that
• 9:50
  are that's hard for people to get access to.
• 9:52
  And then you want to like take advantage of that.
• 9:55
  Let me ask, just sort of attaching this back to like the
• 9:58
  massive sell off that we've seen over the last week.
• 10:03
  Like if, if, if. I guess the way I think about
• 10:05
  this is if it was workflow only, do you feel like it would be
• 10:10
  just as valuable versus the fact that it is workflow and
• 10:15
  proprietary data? Yeah, the the proprietary data
• 10:20
  is, is very much at the heart of what we do.
• 10:24
  But then the workflows is what enables that insight into, I
• 10:29
  don't want to call gathering the data that the that could sound a
• 10:32
  little funky, but but yeah, I mean the workflows is what gives
• 10:37
  us access to the data. And the, the more workflows we
• 10:39
  enable, the richer, the deeper the data set is.
• 10:43
  So that's sort of we have therefore kind of very specific
• 10:48
  insights into into that enterprise data and the context
• 10:51
  who does business with whom, how much, when, how.
• 10:55
  Etcetera. Yeah.
• 10:56
  I think like my extreme example is, I mean, this is the one that
• 11:00
  everybody talks about on X, which is not fair, is like
• 11:03
  Monday, OK, Like no proprietary data and it's like down 65% over
• 11:10
  the last year, right? Like purely A workflow thing.
• 11:15
  Yeah, I don't know. I don't know how you would
• 11:17
  interact with that. Yeah.
• 11:19
  My thought on that kind of stuff is that there's a a price that
• 11:22
  somebody is going to be willing to pay forever to do something
• 11:26
  right. And so like if there's a, if
• 11:28
  your option is use Monday or build your own system, that's an
• 11:33
  equivalent of Monday. There's a cost that is
• 11:37
  associated with both those options.
• 11:39
  And so it just depends on what price, how much can the price of
• 11:43
  Monday come down ultimately to compete against you vibe coding
• 11:48
  your own version of it, because there is, there is a price.
• 11:51
  And I don't know what that price looks like, but I think the the
• 11:53
  reason why Monday it's falling so much to your point, Rishab,
• 11:55
  there's no if, if you don't have something that enables a value
• 12:00
  that a business couldn't get on their own by by coding this
• 12:03
  thing and it's just a workflow, then there's going to be some
• 12:06
  sort of price that that drops to.
• 12:08
  And I actually don't think that I'll ever go to 0.
• 12:11
  But it could be that users are look at Monday and they're like,
• 12:13
  OK, well, I'm not willing to pay more than $3 a month per user
• 12:16
  for this because of this, right. And, and so, yeah, at that
• 12:21
  point, if, if it cost my business 60 bucks a month,
• 12:24
  that's not worth me trying to vibe code this thing.
• 12:26
  This works well enough. But if it's $3000 a month, then
• 12:30
  a business owner may sit there and say, well, I'm just going to
• 12:32
  my team just going to build this thing and we'll run it
• 12:34
  internally. So there's, there's going to be
• 12:36
  some sort of trade off where I, I think software even without AI
• 12:40
  as a part of it or not, not strictly AI driven can survive,
• 12:45
  but just sort of a very low commoditized price point.
• 12:48
  You guys heard about how much a phone contract costs in India?
• 12:52
  It's like next to nothing because the consumer's like now
• 12:54
  I'm not going to pay it and they just switch right all the time.
• 12:57
  Apparently that's what drives it.
• 12:59
  Whereas if you look at like Canada or or the UK, people are
• 13:02
  switching all the time. Providers make it pretty easy to
• 13:05
  switch. I wonder if we're going to see,
• 13:09
  you know, different software make it even easier to switch
• 13:12
  from Monday to, I don't know, Google Chat, which is free or
• 13:15
  from Slack to Google Chat. They just you just click 1
• 13:18
  button and your whole team's moved over.
• 13:20
  Yeah, this is, I think this is the exact problem.
• 13:23
  The exact problem is you have to ask what is the direction of
• 13:26
  travel of two forces. One is the, the, the, the cost
• 13:32
  to switch and the cost to buy code.
• 13:36
  And given the direction that those are both moving, your,
• 13:40
  your first point Jeremiah is the exact right point.
• 13:43
  The altar like you have to measure it against an
• 13:45
  alternative. These things don't live in
• 13:46
  absolutes. But the problem is that those
• 13:48
  two things are moving at incredible rates right now.
• 13:53
  And so when you are doing that like now I'm like a finance
• 13:56
  jockey sitting behind my like cloud assisted Excel sheet
• 14:01
  because even I don't really need to do anything, right.
• 14:04
  And I'm like trying to do this discounted cash flow model.
• 14:07
  I'm like looking forward three years.
• 14:10
  I'm like what price am I putting on the product?
• 14:13
  And that's why I don't think like that like software is
• 14:16
  revenues we go back to like the Monday example.
• 14:19
  I'm I don't know what the revenue and profitability looks
• 14:21
  like, but I'm sure it's fine. And I'm sure it looks very
• 14:26
  different than the stock price does.
• 14:27
  Like I'm sure it's not down 65%. It's just that you know that
• 14:31
  five years from now it's not going to be the same business
• 14:34
  that it is today. And that's what that the the
• 14:37
  market is looking at five years out, not today.
• 14:40
  Yeah. I mean, I think to your point,
• 14:42
  it's it's whether for our business or in the context of
• 14:45
  SAS sell off, it's all about enterprise lock in, right.
• 14:47
  And that's directly a function of yes, the workflows, the
• 14:51
  underlying context, the surface area of what it's traversing
• 14:56
  almost right. I am, I'm less optimistic that
• 15:00
  and, and maybe maybe I'm thinking about this all wrong,
• 15:03
  but I mean, there's, there's a plethora of software we all use,
• 15:06
  right? Like take, take the example of a
• 15:07
  feature flag software. Like do we really have time to
• 15:11
  build and maintain our own feature flag software?
• 15:14
  Like is that where I'm going to invest any of our time,
• 15:17
  especially to the extent that let's say teams become more lean
• 15:20
  and efficient because of AI, that means there are fewer
• 15:24
  people to write code and maintain.
• 15:28
  And then are you going to go through just a bunch of
• 15:30
  consultants or maintain like I don't know who does?
• 15:32
  That so I don't think, yeah, I mean, I think it's the right
• 15:35
  question. I don't think you would ever
• 15:36
  build a feature flag software, but I think it would be wild to
• 15:39
  assume that Cloud Code doesn't automatically feature flag.
• 15:42
  Like if I asked you 5 years from now, do you think Cloud Code
• 15:45
  will automatically feature flag? That's a high chance, yeah.
• 15:48
  Or two years. In six months, maybe.
• 15:52
  Yeah, like so now we're debating timing, right?
• 15:55
  Right. Not a, not an.
• 15:56
  It's not an if it's a Len. Yeah.
• 15:58
  I, I obviously, I don't know Crystal super well, but I
• 16:00
  actually think that's a good example of a software
• 16:02
  potentially that's not as impacted because how like are
• 16:05
  you going to go do all those things on, on your own?
• 16:08
  If you're a business? I think it's a, it's a good spot
• 16:10
  to be in. It's a complex software.
• 16:12
  It's a niche, A niche. It's like a, it's, there's a
• 16:15
  type of software where it's like, it's actually much harder
• 16:20
  to go the vibe coding route than to just work with somebody who's
• 16:23
  doing that on your behalf. So.
• 16:26
  We we got to kill the vibe trying to build this like for
• 16:30
  sure yes, I. Agree.
• 16:31
  You said what? I'm I'm biased, but I agree,
• 16:35
  yeah. Well, in her case.
• 16:38
  Software. Yeah, in her case, it's
• 16:40
  impossible because there's a proprietary data asset that
• 16:43
  underlies the ability to do the workflow right.
• 16:45
  So, so like it's really good news.
• 16:48
  So yeah, because everybody else is going to like try to vibe,
• 16:52
  you know what I mean? Like people are going to try to
• 16:54
  vibe code like Fermat from a year ago.
• 16:57
  Like that is way more likely, right?
• 16:59
  Like that's like a like a like a thing that is very possible.
• 17:04
  Rich, I'm going to take this clip to future investors and
• 17:06
  like you don't have to hear it from me, Rich that it.
• 17:08
  So thank you. Ready material, Slide 2.
• 17:13
  I wouldn't say proprietary data and then also knowledge and
• 17:17
  expertise matter. Like I, I told you guys a couple
• 17:19
  weeks ago, I was trying to vibe code a trading bot.
• 17:22
  It's not going well. And it's not because the vibe
• 17:24
  coding isn't successful. It's because I don't know how to
• 17:28
  actually get like the, the math right.
• 17:30
  So there's like A and it AI doesn't either.
• 17:34
  So it's not like, yeah, like there's certain things that are
• 17:39
  that are just way harder to do And, and maybe like that gets
• 17:43
  better eventually. But I think like the person
• 17:45
  who's done it for 20 years and knows what they're doing is
• 17:47
  probably going to be AI in that context that has like the battle
• 17:52
  scars of losing, probably lost a lot of money on their way to
• 17:56
  getting this like, right. So yeah, I think there's just
• 17:59
  like that. Domain knowledge and expertise
• 18:00
  will still matter. I think that it taste is like
• 18:03
  the only thing that's going to matter like pretty soon.
• 18:07
  I used to think that product managers were the most useless
• 18:10
  people in the entire world. And like the traditional version
• 18:13
  of a product manager, I feel like is still like pretty
• 18:16
  useless. Like they're kind of just a
• 18:17
  middleman bridge between talented people.
• 18:20
  But most but like the really, really good product managers
• 18:22
  have taste. You know, a designer without
• 18:25
  really good sort of like knowledge of the problem is like
• 18:29
  pretty useless. And like a developer without
• 18:33
  really good like UX ability is like pretty useless.
• 18:36
  But like the one role that can actually transform into being
• 18:39
  like to killing both of them, I think is like a really solid
• 18:42
  product manager that knows enough technically and maybe has
• 18:47
  a few people on their team that they can like lean on and like
• 18:50
  knows how things should be done. And then can kind of just like
• 18:53
  execute and iterate at like really, really incredible speed.
• 18:57
  Whereas I think they used to be kind of like an anchor on teams
• 19:00
  that would be like they needed to get sign off for everywhere
• 19:01
  to push anything forward. And it was like the opposite.
• 19:03
  So I feel like that position been like completely 180 and
• 19:07
  could be like, and like in a weird way, it's like what my
• 19:10
  role had become in like the last year was like a product manager
• 19:15
  because I wasn't in Sigma anymore.
• 19:17
  I never had developed. And then now I found that like
• 19:20
  most of my time is spent like 5 coding and it's incredible.
• 19:24
  It's way faster than if we were to take any sort of traditional
• 19:27
  route. But the reason why it's good is
• 19:29
  because I have like the context from both and I feel like I have
• 19:32
  like pretty decent taste. And so that's my biggest in
• 19:35
  terms of like building software companies.
• 19:37
  I think it's just like people with taste that can execute.
• 19:40
  It's going to be like the only thing that's like, really
• 19:42
  useful. Yeah, I, I as a product manager
• 19:47
  prior to becoming a founder, I could not agree more.
• 19:51
  I think, I mean, today, I don't know, this is like a
• 19:53
  controversial slash hot take, but like a product management
• 19:57
  might be the most important job of the IR.
• 20:00
  I I would agree. Do you guys see the Mexican
• 20:03
  standoff reference from Andreessen on?
• 20:05
  That's floating around on a couple days.
• 20:08
  Back, yeah. Tasteful Trump and then ability
• 20:11
  to look around corners and and also like to the extent that
• 20:14
  people are still the ones using software.
• 20:15
  So versus like a world where only agents are buying from
• 20:18
  agents and only agents are going to work.
• 20:19
  And that that was sort of the other related to that which was
• 20:24
  like I forget who it was maybe was adjacent like somebody on
• 20:28
  Twitter said and which I completely agree.
• 20:31
  I actually sent it to some folks on our team, which is like the
• 20:34
  other important job of the IRI really think will be the
• 20:36
  enterprise seller. Yeah, my opinion, but yeah, I
• 20:41
  said this like a year ago to our team at our last off site is
• 20:44
  like with the end outcome with AI is that we all become pretty
• 20:48
  faces on the screen. Like we're just here to make the
• 20:51
  other person feel good, which is an important job.
• 20:53
  Like I was like recommending to a bunch of people to read save.
• 20:57
  Have you guys read Sapiens? By any chance?
• 21:00
  Yeah, I think SAPI, it's an excellent book and actually more
• 21:03
  relevant given what's happening with AI than ever before.
• 21:08
  In my opinion. Sapiens and Guns, Germs and
• 21:12
  Steel are the two that I would recommend.
• 21:14
  But it like reminds you that we are a like we aren't, we are an
• 21:20
  animal of like, like we, we require social constructs
• 21:25
  basically. And we require like we require
• 21:28
  to be able to talk to each other and do things together and
• 21:30
  organize together and things like that.
• 21:34
  And so that part of the human condition is not changing.
• 21:38
  And so and like, yeah, so a bunch of the tasks that we do
• 21:43
  today will change, but that part of it is not going to change.
• 21:46
  So it's like a not unreasonable way to like think about what
• 21:49
  jobs will persist. I'm going to show you guys this
• 21:52
  tweet and get your thoughts on it.
• 21:55
  So this guy went from zero to 1 in like a year and he said this
• 22:01
  apps are about to be worthless. It took me an hour or two to
• 22:05
  write this article. What happens when building a
• 22:08
  consumer app becomes something you can vibe code in an hour?
• 22:12
  What happens when there are 1000 quit smoking or quit drinking
• 22:15
  apps? The answer is that software is
• 22:17
  going to turn into media. Consumers aren't going to pay a
• 22:22
  subscription. They're going to consume like
• 22:24
  they read articles and like they consume apps.
• 22:26
  Five years from now, or maybe even one app's going to become a
• 22:29
  content marketing strategy piece of value put out into the world
• 22:32
  in order to sell your real product.
• 22:33
  And they're saying selling attention or selling physical
• 22:36
  atoms. What do you guys think?
• 22:38
  I mean, maybe, but like, mobile consumer apps were kind of
• 22:43
  already here. Like there were already just
• 22:46
  commoditized things that were being, you know, coded in
• 22:50
  Eastern Europe and they only were businesses that and the
• 22:54
  many of them are our customers where their entire business is
• 22:57
  performed. It's a performance marketing
• 22:58
  company, not like an app company.
• 23:02
  And so like I don't necessarily know much has like changed in
• 23:05
  that category outside of like I still don't believe that the
• 23:10
  average person is going to download raplet on their mobile
• 23:14
  phone. So like we're talking about
• 23:20
  consumer apps, you know, So like, no, I don't think that
• 23:25
  that's like very accurate. Like the only thing would be
• 23:29
  like if if Apple came out with like an incredible vibe coding,
• 23:34
  but it's not really, but it's more of just like a all in one
• 23:37
  agents, you know, create things for yourself thing.
• 23:42
  But I don't think this happens from like a third party outside
• 23:47
  of the hardware. And from an AI point of view, I
• 23:50
  feel like all Apple's done has made something really annoying
• 23:52
  when you accidentally squeeze your phone.
• 23:54
  Oh fuck this thing don't work. Turn it off.
• 23:57
  I mean, I'll even say something that might get me like Banished
• 24:01
  vibe coding or not like end to end, like building end to end
• 24:05
  solutions, taking care of customers like any of this stuff
• 24:09
  and doing it consistently at a high bar is work.
• 24:13
  And I don't know that everyone wants to work hard, truly.
• 24:18
  And to your point, Zach, I, I agree.
• 24:21
  I don't think the average person is going to download Rockwood
• 24:25
  and stick to it and do the work that's needed.
• 24:29
  So I, I, in some ways, I think it's, you know, they say like
• 24:31
  the more things change, the more they stay the same.
• 24:33
  It's, it's going to be a probably a case of that, which
• 24:36
  is like AI or not like it's whoever puts in the work
• 24:41
  consistently and shows up every day.
• 24:43
  And by that I want to be like an individual person, a team,
• 24:46
  whatever it might be, is still the one that's going to win.
• 24:49
  And I don't know that that goes away in the age of AI.
• 24:53
  It changes, I'm sure, but I don't think it like goes away.
• 24:57
  The only thing I think is maybe accurate about his tweet is is
• 25:00
  his comment about people's willingness to pay for
• 25:03
  subscriptions. Like I'm always blown away by
• 25:04
  like the frugalness of the consumer when it comes to like
• 25:08
  tools and apps, whereas like I'm the opposite.
• 25:10
  Just like this is the cheapest thing ever.
• 25:12
  And so the only thing I can see that maybe drives this in some
• 25:15
  reality is that like consumers are, I think increasingly more
• 25:20
  frugal over time about like digital sort of things and
• 25:25
  consumption. And so like I can see
• 25:27
  potentially there being some sort of like price erosion or
• 25:30
  something, but I don't think that the average person is going
• 25:34
  to be bad coding their own stuff.
• 25:35
  Yeah, I agree with that. I also think the other buyer
• 25:39
  that is highly impacted is tech companies because they are going
• 25:44
  to download replet. And so like I like broadly
• 25:50
  agree. I think like the average
• 25:51
  consumer is more frugal than ever.
• 25:53
  They are not an East Coast or West Coast person who has seen
• 25:56
  their wealth go up for X because the stock market has gone up
• 26:00
  notwithstanding the last two weeks, right?
• 26:03
  And so like the consumer is more frugal than ever.
• 26:07
  And I think tech companies are not more frugal than ever, but
• 26:11
  actually have more demands than ever because there are more
• 26:16
  demands being placed on them than ever.
• 26:18
  And so like a traditional SAS tool just does not produce
• 26:22
  features fast enough that the tech company is going to be
• 26:28
  satisfied with the rate at which that vendor is moving.
• 26:32
  And so like, I mean, we have churned more traditional SAS
• 26:36
  tools in the last four weeks than I ever thought possible.
• 26:39
  And there's like more coming. And none of it is because of
• 26:43
  price. All of it is because of feature
• 26:46
  problems. And I think this, I think we are
• 26:48
  not unique, yeah, in, in, in this regard.
• 26:52
  Yeah, I think it's business software that's impacted by
• 26:54
  this. I mean, the reality is like if
• 26:56
  you're a consumer and you care about cost vibe coding your own
• 27:00
  app is actually the most expensive thing you could do.
• 27:03
  It's cheaper to just go buy the the thing for $2.00 a month.
• 27:07
  But for a business it's a totally different equation.
• 27:09
  If you're not spending $2.00 on a piece of software, you're
• 27:11
  spending 202 thousand 20,000, whatever that is.
• 27:15
  So I think the in businesses are structured in a way where you
• 27:21
  have just more ability to actually produce something and
• 27:23
  maintain it. I still think though, it's going
• 27:25
  to be mostly tech businesses that do that.
• 27:27
  Like my, my dad owns a grocery store.
• 27:30
  He's going to be vibe coding a software.
• 27:32
  There's no way he wants to touch that.
• 27:35
  So if you are selling software to, yeah, if you're selling Apos
• 27:39
  software, like you're probably pretty safe.
• 27:41
  Just figure out how to make it better.
• 27:43
  That's not a category that you might have more competition, but
• 27:46
  your customers aren't going to be vibe coding their own
• 27:48
  software. There there are some categories
• 27:51
  that I I do think are going to be surprisingly vibe Cody like 1
• 27:56
  is really unsexy markets that no super solid founder ever wanted
• 28:02
  to get into in the 1st place. An example is my brother runs a
• 28:05
  home services company and like the people that build CRMS for
• 28:10
  like window cleaning companies are like the Z class SAS
• 28:14
  entrepreneurs and so like the products are just horrible.
• 28:18
  Like I remember when I was trying to get him onto another
• 28:21
  CRM and we were looking at like specific CRMS in like this
• 28:24
  category. I was like, this is insane,
• 28:25
  like, you know, and so and I have another friend that runs
• 28:29
  like a a deck sealing company and he had showed me recently
• 28:32
  they have like a door knocking app that will show you, you
• 28:35
  know, for their, their door to door sales people.
• 28:37
  Like when was the last time we knocked on this door and all
• 28:39
  these types of things. And they were using this this
• 28:43
  one that everybody hated, but it was like kind of the one in the
• 28:45
  industry. It was kind of like the HubSpot
• 28:47
  and he had Vibe coded their own version with a companion mobile
• 28:51
  app that was like incredible because like they were so
• 28:55
  connected to the problem. And so like, that's like an
• 28:59
  industry where you're like, Oh yeah.
• 29:00
  Is the average like painter gonna vibe code?
• 29:03
  It's like actually like I think they might because they're
• 29:06
  pretty underserved. Like as tech companies, we get
• 29:09
  really great software from like the best teams in the world.
• 29:12
  There might be this like the unexpected category actually
• 29:17
  might be disrupted a lot. As well.
• 29:19
  I, I actually think this is the underestimated insight, Zach, is
• 29:23
  that vibe like everybody thing. And especially indeed to see
• 29:27
  ecom Twitter like like this, this is like the rhetoric is
• 29:31
  that it's going to make the price go down.
• 29:34
  This is like not the point at all.
• 29:35
  Like the point is that with AI and vibe coding, you can get
• 29:39
  better features. And so actually where you're
• 29:43
  going to see it the most is exactly in the situations that
• 29:46
  you're describing the way if I bought it from a vendor, I'm
• 29:49
  going to guess something that I don't like.
• 29:51
  It could be a duck sealing company, it could be home
• 29:53
  services. It in our case, it's ETL like we
• 29:57
  built something better than 5 Tran for ourselves because five
• 30:00
  Tran couldn't do what we wanted it to do, right.
• 30:04
  And and so I think it like the shape that this takes is going
• 30:09
  to be a feature problem, not a cost problem.
• 30:11
  Yeah, yeah, Zach, you just reminded me, my friend, he
• 30:15
  didn't vibe code his app, but he has one of the biggest apps in
• 30:18
  like Brazil and Paraguay for like restaurants.
• 30:22
  So it's everything from ordering ingredients to serving tables to
• 30:26
  take out. It's like everything, all the
• 30:28
  money didn't vibe code it, but it's still like a fairly big app
• 30:32
  does very well. So it's even just that
• 30:34
  geographic component and I think we see that in E com as well.
• 30:38
  You've got all the Australian brands and British brands that
• 30:40
  just see what's working in the US and then, you know, go into a
• 30:43
  fit, you know, DTC brand and a different country.
• 30:47
  We've been talking or there's there's been this trend lately
• 30:49
  around buying boring businesses. Maybe the the new trend is build
• 30:53
  software for boring businesses. Vibe cut software from boring
• 30:57
  businesses. Yeah, I actually just thought
• 30:59
  too when I was like watching my brother work on this and my
• 31:02
  other friend who are like people that spend a lot of time on
• 31:05
  computers and when I open up a piece of software, like learning
• 31:09
  the software, there isn't really like a learning curve for most
• 31:12
  softwares that I buy. But it was like, well, what is
• 31:15
  the best on? And I thought about like
• 31:16
  onboarding flows and it was like, well onboarding flows like
• 31:19
  slowly show you different steps in the journey and you kind of
• 31:23
  like play with them and interact with them.
• 31:24
  I was like, well, the best onboarding flow for any software
• 31:27
  would be to like ask it to do something and then it does it.
• 31:30
  And then you see like a new button and then you know that
• 31:33
  when I want this thing to be done in the future, I would
• 31:35
  click this button. And so it was like outside of
• 31:38
  the features, outside of the price, like could also just like
• 31:41
  custom made software if it got to that point, also just be like
• 31:45
  the best training tool for the person that's using it or the
• 31:48
  teams that are using it. It's like there isn't a learning
• 31:51
  curve. It's like as you get it to the
• 31:53
  place in which you need it to be, like you've actually learned
• 31:56
  100% of the of the tool. Yeah.
• 32:00
  Doing the background onboarding flow is, is where there is no
• 32:03
  onboarding flow because it only presents itself when needed, as
• 32:08
  needed. Because we we've got this, I
• 32:10
  mean we sort of fit into a bit of that sort of category where
• 32:13
  our customers are, they've been doing DTC OPS, but now they're
• 32:18
  doing retail OPS. It's a totally different muscle.
• 32:21
  The standard operating procedures, the language, the
• 32:23
  like just the wiring is completely different and they're
• 32:27
  not always prepared or maybe they haven't always correctly
• 32:32
  estimated what that shift may look like.
• 32:35
  And that actually morphs itself into support tickets and
• 32:39
  questions largely around Dolby and expertise like somebody was
• 32:43
  saying earlier and how that then please?
• 32:46
  We're in the fire, we're in the belly of it in ECOM.
• 32:50
  We, we foreplay, Zach doesn't know about this because we're on
• 32:53
  a ski trip, but we just lost a customer recently, big customer
• 32:57
  because they built their own thing, right?
• 32:59
  And these are businesses, agencies in ECOM that are
• 33:03
  heavied up with OpEx. They spend so much on labor and
• 33:08
  most of this labor is just sitting around, right, doing,
• 33:12
  you know, running ads, sending emails, doing fairly basic
• 33:15
  routine tasks that I imagine are as close to getting fully
• 33:18
  automated as they can be separate from the the creative
• 33:22
  work. So they're building their own
• 33:24
  stuff, right? So we're in that replacement
• 33:28
  category. Reminds me of my first brand,
• 33:30
  actually my first brand, as soon as we got like 500 a thousand
• 33:34
  sales reliably, someone posted it online and it ended up on all
• 33:37
  of these piracy websites where you could get my products for
• 33:40
  free. And then people would resell the
• 33:42
  products by tweaking them slightly and so everyone would
• 33:45
  have them for free. So I had to release new stuff
• 33:47
  all the time. But one of my brands now, knock
• 33:51
  on wood, it's been a year, but my best seller is still my best
• 33:53
  seller. So it's like, OK, let's hope
• 33:56
  this continues. But with my first brand, I was
• 33:59
  really in the the belly of it. I had to continually adapt and
• 34:04
  make new products. Otherwise I was, you know, sales
• 34:06
  were going to drop off the side of a Cliff.
• 34:08
  So maybe we're uniquely in that category because we don't, you
• 34:11
  know, build software for grocery stores.
• 34:13
  We build software for people who are, you know, OpEx heavy
• 34:16
  businesses and technically their labor can and can and will go
• 34:21
  start building their own stuff. You know, I was asked this by
• 34:24
  one of our investors last week, and I'm curious if you guys feel
• 34:28
  this way, that the biggest thing that has shifted is that you
• 34:31
  have to like ship product at an unprecedented speed And like,
• 34:36
  that is what the new expectation is.
• 34:38
  It sounds like everybody is OK. What what, like if you had to
• 34:43
  describe the shift, how would you describe it like in your own
• 34:47
  organization? And what are you doing in your
• 34:49
  own organization to keep up with what you think the new
• 34:51
  expectation is? I think we covered this a little
• 34:54
  bit couple weeks ago maybe, but the idea of things like annual
• 34:58
  planning and I, I mean, Rishabh, sounds like you, you were
• 35:01
  operate on a little bit faster timeline than we do even.
• 35:03
  But we kind of look at, I mean like a quarterly goal.
• 35:08
  But it really it's like, OK, what's the next?
• 35:10
  What's the next thing that we're doing?
• 35:11
  And let's just do that thing. And that's, that's the
• 35:13
  objective. And then I think the, the other
• 35:17
  piece of it that I think about constant, I've been thinking
• 35:20
  about this for a long time is how do we like that is true.
• 35:24
  If you're in the, in the just like pure software business,
• 35:28
  it's all about shipping features as fast as possible.
• 35:31
  So how do we actually get out of just being like a pure software
• 35:35
  play so that we don't the odds that we're going to be the able
• 35:41
  to ship software faster than everybody else is actually small
• 35:44
  in my opinion. I don't know that we're geared
• 35:48
  that way. We're not a bunch of like 20
• 35:50
  somethings on a, in a, a basement or an Airbnb or
• 35:55
  whatever, shipping as fast as possible.
• 35:57
  So I'm, that's just the reality. Like I don't want to work 80
• 36:00
  hours a week. I've got a family.
• 36:02
  And so like, what does that look like for me to think about that
• 36:06
  and, and try to like, I, I like to work hard.
• 36:09
  I, I've always worked hard, but I, I can't put the kind of time
• 36:12
  in that I used to. So it can't just be about
• 36:14
  volume. It has to be about strategy and
• 36:16
  thinking beyond just the just shipping features.
• 36:22
  Yeah. I mean, I think for us, we, I
• 36:23
  really think in terms of problem surface area, right?
• 36:25
  I mean this is sort of not a new concept, but I think what was it
• 36:29
  a couple years ago, 22 to four years ago and sort of this
• 36:33
  notion of compound startups was making the rounds, right,
• 36:36
  Whether you're rippling and companies like that that really
• 36:39
  very grand expanded the product suite.
• 36:41
  And so I think AI really enables that and in a very, very real
• 36:46
  way. So how can you solve adjacent
• 36:48
  use cases that maybe normally would have taken you 2 years to
• 36:52
  get to, if ever? How do you really increase the
• 36:55
  surface area of the problems that you solve for your
• 36:57
  customers with higher quality? I love what you were saying
• 37:01
  earlier about work and how, you know, if you compare this to
• 37:06
  consumer apps where there's like 10,000 variations of the weight
• 37:10
  loss, smoking, quitting smoking, routine building apps or
• 37:14
  whatever happens to be, there is going to be a tendency or to
• 37:19
  what Zach was saying. Most people aren't going to
• 37:21
  build apps and even the people who do build apps will probably
• 37:23
  build something expedient because it's become easier.
• 37:26
  We were talking about this last week actually, in terms of, you
• 37:29
  know, AI ad creative, there's an expedient path that you can take
• 37:32
  that's very lazy and actually the probably the worst quality
• 37:36
  content. And that's not to say that the
• 37:37
  tools that you're using to make that are bad.
• 37:40
  It's just the optimizing for speed over quality over actually
• 37:45
  solving a problem over actually working to make things better
• 37:49
  might not be the right goal. There are going to be a few
• 37:51
  people who actually, whether it's taste or it's forecasting
• 37:55
  consumer demand well, are going to solve real problems for their
• 37:59
  customers. And that's going to be the
• 38:01
  alpha, not shipping features as fast as possible.
• 38:05
  I don't know, maybe right. But yeah, I I can agree with you
• 38:09
  more. I think about putting in the
• 38:11
  work right, not doing the easy thing.
• 38:13
  Yeah, I just my, the thing that has me rocked is like, and yeah,
• 38:19
  and we talked about it like a version of this Jeremiah a
• 38:21
  couple of weeks ago. But I think the thing that's
• 38:23
  rocked me is anthropic or clawed is like releasing something
• 38:28
  every two days that's useful. And I just, I've just never, I
• 38:36
  mean, have you any of have any of you guys like experienced
• 38:39
  anything like that from a single company?
• 38:41
  I mean, it's just crazy. I think of like, I, I think of
• 38:44
  intercom as a good example of like a software business and
• 38:47
  ability to ship new things over, let's say 10 years ago, right?
• 38:52
  What intercom was doing, where they went from just like a help
• 38:54
  desk to like in app messaging. And I was a bunch of other
• 38:57
  things too, but like, that's a good example.
• 38:59
  But that was like every six months there was a new drop,
• 39:03
  right? Which that felt good at that
• 39:05
  time. Like, oh wow, there's something
• 39:07
  new every as a as an intercom customer that felt really great
• 39:10
  and there was new value being created.
• 39:12
  But the speed of that is totally different than Windows now.
• 39:16
  Yeah, Anthropic maker open AI looks slow.
• 39:19
  And I mean, I guess they're slow compared, right?
• 39:25
  I I think Open AI is sort of like burdened by their customer
• 39:28
  base. It's like I, I don't know, I
• 39:31
  don't think that there was like a, it was like the way the game
• 39:34
  board just like kind of fell and the pieces fell.
• 39:37
  And it's like, I wouldn't want to be open AI right now because
• 39:40
  they are just like burdened by just like massive prosumer cons,
• 39:45
  more consumer like category that like makes shipping features
• 39:50
  fucking horrible. Like where do we have some of
• 39:54
  that kind of like drag, which is just like a bunch of people that
• 40:00
  use you for really service level stuff, which make it pretty hard
• 40:05
  to go deep sometimes. I think it's like an interesting
• 40:08
  place for them to be like strategically because they can
• 40:11
  either like just like fully pivot into the consumer, but I
• 40:15
  think that they know that that's like probably not great long
• 40:18
  term. But then it's OK then then you
• 40:21
  need to somehow like shed whatever the weight that's being
• 40:24
  caused by this like consumer. That's what I think it is.
• 40:27
  Yeah. It's.
• 40:28
  It's their version of innovators.
• 40:31
  It is just just look at us. I think it's interesting Clod,
• 40:37
  the chat version of clod has always been far inferior to to
• 40:43
  ChatGPT. And I think that's like it stops
• 40:46
  you from using it unless you're serious about like a, a real use
• 40:49
  case that goes just chatting. And that's actually probably the
• 40:53
  thing that's caused such a divergent where clod is very
• 40:58
  business focused. And I mean, I I use it for
• 41:00
  business where and I still use chat BT for a lot of personal,
• 41:04
  but I I'm baking clod into my business workflows in a way that
• 41:08
  I don't chat with open. Yeah, I, it's hard for me to
• 41:12
  tell like what Open AI does from from here.
• 41:16
  But yeah, I mean, if you wake up 1 morning with like 800 million
• 41:21
  weekly active or whatever the new metric is, it's like you
• 41:27
  kind of have to serve that. Yeah.
• 41:30
  So and then no matter what you do, the whole company gets
• 41:33
  pulled by that, right. So like, I'll give a really good
• 41:36
  example like that's relevant to commerce.
• 41:40
  They don't have a choice but to think about the consumer
• 41:44
  experience of agentic shopping over the merchant experience.
• 41:47
  And so like if you're a merchant and you're wondering how do I
• 41:52
  think about what to do about Digentec shopping?
• 41:56
  I'm like, I'm sure if you, you know, called people at Open AI,
• 41:59
  they would like be helpful and things like that.
• 42:02
  But fundamentally, Open AI has no choice but to actually be
• 42:07
  consumer first in how they answer that question because
• 42:09
  there are 800 million people who are reliant on them to be
• 42:13
  consumer 1st and how they answer that question, right?
• 42:16
  Thankfully they released an ads product, so now there is good
• 42:20
  aligned incentives for all parties, right?
• 42:23
  So like we should see more to like build a full ecosystem.
• 42:32
  But yeah, it just creates all of these weird dynamics, right?
• 42:36
  In terms of like, where do you spend the time and how do you
• 42:38
  spend it and where do you like what are the first problems you
• 42:43
  solve? Yeah.
• 42:45
  And there's a lot of opportunity, honestly, for B to
• 42:47
  B companies in support of the open AI ecosystem.
• 42:51
  So like an interesting outcome for B to B software builders
• 42:56
  like us is thinking about how those dynamics impact how we
• 43:01
  build relative to those companies, right?
• 43:04
  Because one of those companies has different incentives than
• 43:07
  the other, right? And so like an example of that,
• 43:10
  which I'm not saying that Anthropic is like a good actor,
• 43:15
  bad actor, but something as simple as Claudebot, they chose
• 43:19
  to go after him for naming an IP or they discontinued serving the
• 43:26
  X AI customer base through cursor, right?
• 43:29
  So this is like they told their customer to stop serving their
• 43:33
  end customer, right? And so there's all of these
• 43:36
  weird dynamics, whereas like Open AI, we haven't seen any of
• 43:39
  that from Open AI, right? But can you think of any
• 43:42
  examples of that? And so I yeah, anyway, I think
• 43:45
  it just like frames the B to B opportunity pretty differently
• 43:49
  because of the the things that you're talking about.
• 43:51
  Is it possible that the best outcome for chat or for open AI
• 43:56
  mean ChatGPT specific as products?
• 43:58
  Kind of what I'm thinking about here, But for that to actually
• 44:00
  be owned by a matter or Google where you actually have all the
• 44:03
  consumer data, because I think their challenge right now is
• 44:05
  they don't have like the browsing day.
• 44:08
  They don't know what you're doing and different apps.
• 44:11
  I mean, maybe Apple's a good example of that, but they don't
• 44:14
  have the same access to that consumer behavioral data.
• 44:19
  And that's ultimately like, what stops them from from being a
• 44:22
  valuable consumer only play? I mean, yeah, like if I could
• 44:28
  dream of anything, it would be for Apple to buy Open AI because
• 44:32
  of how much of an Apple fan I am.
• 44:34
  And also, I mean, we we still haven't, I think they haven't
• 44:37
  fully come out with like what the hardware products going to
• 44:39
  be, right. And so it's like, yeah, like
• 44:42
  Apple hasn't made a meaningful hardware product in a very long
• 44:44
  time instead of Airpods. And so, yeah, like it could be
• 44:48
  the personal compute device of the future.
• 44:51
  And I feel like you would probably better if that was like
• 44:54
  the in the hands of Apple than in the hands of open AI, like a
• 44:58
  long term product point of view. But yeah, like I, I, I feel like
• 45:01
  yes, but I mean, I also think we were in like the beginning of
• 45:05
  the first inning, so on where these companies end up going.
• 45:09
  I mean, don't, don't you guys think to to rish your point
• 45:11
  about having as large of a user base as they do?
• 45:16
  Don't they almost have access to a data set that normal sort of
• 45:20
  telematics can't give you through sensors and things like
• 45:23
  that? Like people are literally
• 45:24
  telling it what they think and how they feel.
• 45:27
  Yeah, they have this interesting problem where yes, in the
• 45:32
  context of the chat, but they don't know anything that happens
• 45:35
  off property. So one of one of open AI is like
• 45:39
  biggest challenges is like, and we know this because we come
• 45:45
  from the commerce world. Just think about like the click
• 45:47
  through rate to the end brands website where like a lot of the
• 45:50
  rest of the activity happens for purchasing, right?
• 45:53
  So like commercial activity on the Internet, they don't even
• 45:56
  have a pixel yet, right? There's no such thing as an open
• 45:59
  AI pixel, much less like the full browser data like when
• 46:04
  Google has because you you use Chrome.
• 46:07
  And so I think they know a lot about your behavior in any
• 46:12
  moment and how you're interacting with each other.
• 46:15
  And so totally they have like a depth of information in one
• 46:19
  dimension that is way deeper than anyone else has ever had in
• 46:22
  the history of technology. And they live in this funny
• 46:25
  duality where on the other hand, they don't know anything that is
• 46:28
  happening outside of their ecosystem.
• 46:31
  And actually very few other tech companies have this problem.
• 46:34
  Like Meta knows you across multiple devices because you're
• 46:39
  on Facebook, on your desktop, you're on Instagram on your
• 46:42
  mobile phone, you're chatting on WhatsApp.
• 46:45
  So they actually have a more full view.
• 46:46
  Google has a more full view, Amazon has a more full view.
• 46:50
  And so like Open AI is uniquely a consumer, a large consumer
• 46:55
  company that does not have as full of a view, which is very
• 46:58
  odd the Google. Ones.
• 46:59
  Interesting software and no idea I haven't kept on.
• 47:03
  I'm not sure. I, I mean, if I had to guess
• 47:06
  like Fiji Simo is like going to solve this problem.
• 47:11
  By the way, I'm not saying that they won't solve the problem.
• 47:15
  I'm just saying that the, the shape of their problem today is
• 47:18
  very interesting. Yeah.
• 47:20
  And I think that they should, I think we should expect them to
• 47:23
  learn more and more into partnerships to help them solve
• 47:25
  this problem over the next two years because again, they don't
• 47:29
  have the same incentive structure that Cloud does, right
• 47:33
  and or Entropic does. And so I just think it's a very
• 47:36
  interesting yeah board, so to speak, for how these things end
• 47:41
  up landing, but. Even if they were to launch
• 47:44
  their own version of these things, they don't have the
• 47:46
  decade plus of context that these other companies would.
• 47:49
  And I think that matters ultimately.
• 47:51
  But I, I find chat GB TI use it a lot for and I was just
• 47:55
  thinking about, I probably have 1000 plus chats with in ChatGPT.
• 47:59
  And so like my wife and I were actually doing this last night.
• 48:02
  We're asking questions about ourselves and that the depth of
• 48:04
  information it has about me is far deeper than what my wife has
• 48:07
  because we use it in different ways.
• 48:09
  But even in that context, I find I, I'm often annoyed by how
• 48:15
  little it knows about the things I, I about certain things,
• 48:19
  right, because it doesn't, it has such a small piece of
• 48:22
  context ultimately. So I'll ask a question like, OK,
• 48:24
  well, that's totally wrong because it doesn't know anything
• 48:26
  about these other things. And then I would have to give it
• 48:28
  that context. And then at that point said,
• 48:30
  well, I don't want to. It's AI just not interested in
• 48:33
  doing that for, for whatever reason.
• 48:35
  But I think it's, it's so good at some things like it.
• 48:38
  It knows a lot about how I think in a work capacity but nothing
• 48:42
  about my personal on a life base.
• 48:43
  Not nothing but like very little about my outside of work self.
• 48:48
  And like I'm sure if I asked it for certain things to buy that I
• 48:52
  would be interested. It probably would actually do a
• 48:54
  terrible job because it's all focused around 1 aspect of who I
• 48:57
  am. You guys overthink about how the
• 49:01
  cost to solve the problem first might not be worth the benefit
• 49:05
  if you think about Apple, for example, not really taking part
• 49:09
  in the whole AI race right now. But maybe they don't have to
• 49:12
  because the cost to solve that problem a little bit further
• 49:14
  down the line is going to be fine.
• 49:16
  And they already have the penetration.
• 49:17
  They already have the user base or open AI falling behind and in
• 49:21
  some ways B to B. If they can have the user base,
• 49:24
  then maybe they make it out on top.
• 49:27
  I remember back with my first brand again, this guy told me I
• 49:31
  was an idiot. So why are you making your
• 49:33
  products? Because everyone was just like
• 49:35
  reselling other people's products, right?
• 49:38
  And you know, I used to look at some of the copycat brands that
• 49:41
  popped up after and I imagine maybe you guys might feel this
• 49:44
  way with about know foreplay copycats.
• 49:47
  It's like, oh wow, it's become so cheap to solve the problem.
• 49:51
  Second, right? Maybe you've missed out on
• 49:54
  market share. Maybe you have a your tall
• 49:58
  mountain to climb in terms of getting that user base and
• 50:01
  solving a problem that makes switching worthwhile.
• 50:04
  But yeah, I wonder if there's a as a competitive advantage to, I
• 50:08
  don't know, staying in second place until it's the right time.
• 50:11
  Yeah, there's like a mountain of success stories of like waiting
• 50:16
  for someone else to prove like, Oh my God, I can't believe I
• 50:19
  forgot that. But is it bending spoons the the
• 50:21
  European like software company that literally just sees what
• 50:24
  happens in America and then they just like copy it in Europe?
• 50:27
  And they're thinking it was Rocket and.
• 50:29
  Rocket Internet Sorry, Benny Smith is like a is APE fund,
• 50:33
  right I. Know.
• 50:34
  Yeah, they just bought Vimeo. Yeah.
• 50:36
  Yeah, OK, yeah, yeah, sorry. Yeah, Rocket Internet or like,
• 50:40
  Oh my God, who are the guys? Drata.
• 50:43
  They just like saw what Vanda did and said we're gonna out
• 50:45
  execute them. I mean like I think this is, I
• 50:47
  mean, there's two. There's two things here though.
• 50:50
  It's like 1 is what is the, we're talking about public
• 50:53
  companies. So like what is the objective of
• 50:55
  public company? It's like to create short
• 50:57
  shareholder value most of the time in a smaller window than
• 51:01
  which in which the asset is being under like underwritten
• 51:04
  for which is like a weird dichotomy of the entire
• 51:06
  situation. And then like the other piece is
• 51:11
  like all of the companies that are deciding to go into this
• 51:15
  race is like their CapEx spend is like insane.
• 51:17
  And then you have the underlying strategy of your CapEx.
• 51:20
  So it's like like, for example, like I actually saw like a
• 51:23
  really good breakdown yesterday on like why Amazon share prices
• 51:26
  kind of like, you know, seemingly has like a very shaky
• 51:29
  bottom to it. And it's that like their CapEx
• 51:31
  is actually spread across like so many different verticals that
• 51:34
  the idea that it's like the the amount of CapEx is well has
• 51:40
  enough density in anyone vertical in which they're
• 51:42
  deploying it that it's going to like essentially win any of
• 51:46
  these races. And so it's like, I think it's
• 51:48
  not just deciding on whether you want to stay in second place,
• 51:51
  but it's also deciding on which game you want to win in this
• 51:54
  situation. Because like even public markets
• 51:57
  are are like signaling the idea that like, we understand that
• 52:01
  this is a race. We understand this is going to,
• 52:04
  you know, inquire like a crazy amount of capital expenditure
• 52:07
  over the past 2028. And like, participation isn't
• 52:11
  just enough in this situation. Like you actually need to win
• 52:14
  because of how high the stakes are.
• 52:16
  Yeah, so. You could also end up being like
• 52:18
  under a mountain of like infrastructure in which you can
• 52:21
  never get out of as well. Like that is also like a a
• 52:24
  future that exists in trying to sedate in second place.
• 52:28
  Crazy time I got to go guys, but this was super fun.
• 52:33
  But you guys? Thanks for coming.