Latest / SEA of Startups / 🎙️ EP 10: The $40B GenAI Reality Check: Why 95% of Enterprise AI Projects Are Failing (And What the 5% Winners Know)

Transcript

🎙️ EP 10: The $40B GenAI Reality Check: Why 95% of Enterprise AI Projects Are Failing (And What the 5% Winners Know) · SEA of Startups

• 0:06
  SPEAKER_02: Uh-oh, what's coming?
• 0:08
  SPEAKER_02: Yeah.
• 0:40
  SPEAKER_00: Yeah,
• 0:41
  SPEAKER_00: it's like all these slides be sliding around and you're like,
• 0:45
  SPEAKER_00: not actually sure what's going on.
• 0:47
  SPEAKER_00: I guess it's like the emperor's new AI clothes.
• 0:59
  SPEAKER_01: IT just dropped it.
• 1:00
  SPEAKER_01: It freaked out a lot of the people in the industry.
• 1:04
  And basically what it talks about was like massive tens of
• 1:07
  billions of dollars of enterprise AI.
• 1:11
  SPEAKER_01: And the headline figure that they throw out there,
• 1:14
  SPEAKER_01: like leading off with it is 95% have zero ROI.
• 1:19
  SPEAKER_01: Zero.
• 1:21
  SPEAKER_01: I mean, we can pick across it, but like headline 95% zero.
• 1:28
  SPEAKER_00: That's huge, huge chunk.
• 1:30
  And it's like MIT coming up with a study that chat GPT reduces
• 1:34
  our brain power.
• 1:34
  SPEAKER_00: So wait, back to this hold up.
• 1:38
  SPEAKER_00: I mean, that can be real, right?
• 1:40
  SPEAKER_00: 95% is crazy.
• 1:42
  SPEAKER_00: Seriously.
• 1:44
  SPEAKER_01: When you dig into it,
• 1:45
  there's a lot that makes sense in regards to the reasoning why
• 1:49
  it actually gets
• 1:51
  SPEAKER_01: there.
• 1:51
  And to be honest with you,
• 1:52
  I wish I was making it up because I'm in the arena.
• 1:55
  SPEAKER_01: I'm looking at AI tech companies right now.
• 1:57
  SPEAKER_01: We're looking at investing in SaaS.
• 2:00
  SPEAKER_01: But like,
• 2:01
  SPEAKER_01: while you have a lot of this public fake facing social media,
• 2:05
  SPEAKER_01: LinkedIn posting about the AI journey,
• 2:09
  SPEAKER_01: you know,
• 2:09
  SPEAKER_01: really a small percents are actually really generating tangible impact or
• 2:15
  SPEAKER_01: transformation or ROI out of it.
• 2:18
  SPEAKER_01: And,
• 2:18
  SPEAKER_01: you know,
• 2:19
  SPEAKER_01: I like the term that the report uses,
• 2:21
  SPEAKER_01: which is that there's the Gen AI divide and,
• 2:25
  SPEAKER_01: you
• 2:27
  SPEAKER_01: Taking these numbers, it's a pretty wide divide.
• 2:30
  Considering it's coming from the U.S.,
• 2:31
  we'll call it the Grand Canyon of Gen AI divides.
• 2:35
  SPEAKER_00: Yeah, no doubt.
• 2:37
  SPEAKER_00: The way I see it,
• 2:38
  SPEAKER_00: even when the whole lovable thing came out,
• 2:40
  SPEAKER_00: I got so overwhelmed that everywhere was personally AI.
• 2:43
  And now it's like with this study,
• 2:45
  it's kind of making me wonder,
• 2:47
  is it really alters hype?
• 2:49
  SPEAKER_00: I guess what we're trying to get at is,
• 2:51
  are we really saying that we might need to have some
• 2:55
  uncomfortable conversations
• 2:56
  SPEAKER_00: today?
• 2:57
  SPEAKER_00: Seems...
• 2:58
  SPEAKER_00: I came in for a ride.
• 3:00
  SPEAKER_01: It's a little uncomfortable,
• 3:01
  especially if you're one of the players that are sitting there
• 3:04
  with zero ROI,
• 3:05
  SPEAKER_01: but lots of money spent.
• 3:07
  SPEAKER_01: But if you're building,
• 3:09
  SPEAKER_01: if you're funding,
• 3:10
  SPEAKER_01: or if you're betting your career on AI,
• 3:13
  SPEAKER_01: I hope this conversation might save you from becoming part of the 95%.
• 3:17
  SPEAKER_01: Let's be part of that 5%.
• 3:19
  SPEAKER_01: So let's dig into it.
• 3:30
  SPEAKER_00: And I'm Kim Yeo,
• 3:32
  SPEAKER_00: and welcome to the Sea of Startups,
• 3:34
  SPEAKER_00: where we dig into the stories that matter,
• 3:37
  SPEAKER_00: not just the ones that trend,
• 3:38
  SPEAKER_00: like this one.
• 3:39
  SPEAKER_01: So today,
• 3:40
  SPEAKER_01: we're digging in,
• 3:41
  SPEAKER_01: as we kind of referenced,
• 3:42
  SPEAKER_01: we're going to do a deep dive into the AI revolution,
• 3:46
  SPEAKER_01: how it's having an identity crisis,
• 3:49
  SPEAKER_01: and,
• 3:50
  SPEAKER_01: you know,
• 3:50
  SPEAKER_01: what the hell are we doing about it?
• 3:51
  SPEAKER_01: We're taking this report into context later.
• 3:55
  SPEAKER_01: What do we do?
• 3:56
  SPEAKER_01: So before we dive into the actual data and the report did come out,
• 4:02
  SPEAKER_01: so we could go full nerd today,
• 4:04
  SPEAKER_01: get deep into the details.
• 4:06
  SPEAKER_01: You know, I love doing that anyways.
• 4:08
  SPEAKER_01: But I want to talk about something a little bit more personal.
• 4:11
  How many of you that are listening right now have kind of felt
• 4:15
  that pit in your
• 4:17
  SPEAKER_01: stomach when somebody asks you about like,
• 4:19
  SPEAKER_01: so how's your AI initiative going?
• 4:22
  Like the dinner party conversation that for whatever reason has
• 4:25
  become like super
• 4:26
  cliche at this point as a bunch of C-levels get around that may
• 4:31
  not actually know
• 4:32
  SPEAKER_01: AI,
• 4:32
  SPEAKER_01: but are like,
• 4:33
  SPEAKER_01: so how's your AI initiative going on?
• 4:35
  SPEAKER_00: Everyone's asking that question at this point.
• 4:38
  SPEAKER_00: It's all over the internet.
• 4:39
  SPEAKER_00: Oh my, oh man, that question again.
• 4:42
  SPEAKER_00: It's like asking how someone,
• 4:44
  SPEAKER_00: how their diet is going,
• 4:45
  SPEAKER_00: or at least their workout is going,
• 4:46
  SPEAKER_00: when you know they just had pizza for breakfast.
• 4:48
  SPEAKER_01: A little bit of a universal pretending going on right now.
• 4:53
  SPEAKER_01: You know, I'm a tech optimist.
• 4:55
  We're all supposed to be like,
• 4:57
  the promise that's out there is like,
• 4:59
  we're going to be AI first.
• 5:00
  SPEAKER_01: There's this disruption.
• 5:01
  SPEAKER_01: We need to be AI native.
• 5:03
  SPEAKER_01: But like,
• 5:05
  I think we're overestimating the short term,
• 5:07
  maybe underestimating the long term.
• 5:09
  SPEAKER_01: We're a little bit confused.
• 5:11
  SPEAKER_01: Maybe there's a little panic after this report came out.
• 5:13
  SPEAKER_01: So there's a little bit of like sorting out what's going on.
• 5:17
  SPEAKER_00: Yeah,
• 5:17
  and it's funny because I feel like sometimes when I'm using it
• 5:20
  as a person,
• 5:21
  SPEAKER_00: you're a huge avid user of AI as well.
• 5:24
  I've been using it and it's like sometimes you don't feel like
• 5:27
  they're actually
• 5:28
  SPEAKER_00: moving the needle much.
• 5:29
  SPEAKER_00: And as much as it does help a lot,
• 5:32
  SPEAKER_00: But in terms of its consistency and reliability,
• 5:35
  SPEAKER_00: it begs the question of,
• 5:36
  SPEAKER_00: is it really changing things?
• 5:39
  SPEAKER_00: And, you know, it's normal to be confused and slightly panicked.
• 5:43
  SPEAKER_00: It's only human.
• 5:45
  SPEAKER_00: But here's what's not okay.
• 5:46
  Companies are wasting billions and billions of dollars and
• 5:49
  years of talented
• 5:50
  people's lives because we're just too proud to admit we don't
• 5:53
  know what we're doing
• 5:54
  SPEAKER_00: yet.
• 5:54
  SPEAKER_00: And even JP Morgan has his own analyst program and it's like,
• 5:57
  SPEAKER_00: you're robbing talented people's jobs off as well.
• 6:00
  SPEAKER_00: And
• 6:01
  SPEAKER_00: Where is that direction?
• 6:02
  SPEAKER_00: And it's unclear.
• 6:03
  SPEAKER_01: Yeah, 100%.
• 6:05
  So like this MIT report,
• 6:07
  I think it came out of Project Nanda,
• 6:08
  which sits under the media lab.
• 6:10
  SPEAKER_01: So it basically did this.
• 6:12
  Let's think of it as like they're doing the therapy session
• 6:15
  that we all need.
• 6:16
  SPEAKER_01: Tell me what you really think.
• 6:18
  SPEAKER_01: They went around, did these interviews, was part of their methodology.
• 6:24
  And it's essentially like trying to get the people to say the
• 6:27
  quiet part out loud,
• 6:28
  SPEAKER_01: which is like,
• 6:29
  SPEAKER_01: we're failing.
• 6:31
  SPEAKER_01: We're not failing because the technology is bad.
• 6:33
  We're failing because the approach that most companies are
• 6:36
  taking is wrong.
• 6:37
  SPEAKER_01: There's some really interesting takeaways that we can pull out of this.
• 6:41
  SPEAKER_01: We can still be tech office.
• 6:43
  SPEAKER_01: We just need to get things straight in the way we're approaching it.
• 6:48
  SPEAKER_00: Yeah,
• 6:48
  SPEAKER_00: and I think,
• 6:48
  SPEAKER_00: again,
• 6:49
  SPEAKER_00: that comes down to perspective,
• 6:51
  SPEAKER_00: whether we have the right view of what's working for us and what's not.
• 6:55
  SPEAKER_00: And, well, talk to me about that.
• 6:57
  SPEAKER_00: What does wrong look like?
• 6:59
  SPEAKER_01: If you unpack some of it and you go around, there's a huge consumer case.
• 7:03
  SPEAKER_01: So you have a lot of these people that are sitting at these enterprises.
• 7:09
  SPEAKER_01: I can't remember the exact figure that came out of it,
• 7:11
  SPEAKER_01: but like 80% have like experimented with the generic tools out there,
• 7:16
  SPEAKER_01: which would be like your chat GPT,
• 7:18
  SPEAKER_01: maybe they've got a copilot.
• 7:20
  SPEAKER_01: But like when you look at,
• 7:22
  SPEAKER_01: when you look beyond and start getting into like this specialized tools,
• 7:27
  SPEAKER_01: Really a much smaller cut that in half have actually deployed it.
• 7:31
  SPEAKER_01: So we're trying, which makes sense.
• 7:34
  SPEAKER_01: We have this evolutionary aspect of tech adoption,
• 7:36
  which is starting in the consumer side and migrating over to
• 7:40
  the enterprise side.
• 7:41
  SPEAKER_01: But we're still early innings in this game.
• 7:44
  SPEAKER_01: And so like we've got a long way to go.
• 7:47
  SPEAKER_00: Then again, long way to go, but we are not dancing yet.
• 7:50
  SPEAKER_00: The dance floor is still warming up.
• 7:53
  When you're at a party and there's a dance floor and there's
• 7:56
  just one or two people
• 7:57
  SPEAKER_00: having the guts to show up,
• 7:58
  SPEAKER_00: that kind of thing.
• 7:59
  SPEAKER_00: It seems to me like it's that way.
• 8:02
  SPEAKER_01: Yeah.
• 8:02
  SPEAKER_01: I don't know whether or not this reference will land with anybody,
• 8:05
  but there's an old movie called where the people are just like
• 8:08
  standing by the wall
• 8:09
  SPEAKER_01: with their drinks,
• 8:10
  SPEAKER_01: nodding their head.
• 8:11
  SPEAKER_01: Your comment on dancing just kind of made me think of that.
• 8:15
  SPEAKER_01: But like,
• 8:16
  it's true because like you got a lot of folks out there that
• 8:18
  are just kind of like
• 8:19
  SPEAKER_01: watching the game and trying to figure out like,
• 8:21
  SPEAKER_01: how do I get out there and start being part of all of this?
• 8:24
  And then you've got the folks that are actually out on the
• 8:26
  dance floor that are
• 8:27
  SPEAKER_01: like,
• 8:27
  SPEAKER_01: am I doing this right?
• 8:29
  SPEAKER_01: Meanwhile,
• 8:29
  you've got a couple of people that look like superstars that
• 8:32
  should be on some sort
• 8:34
  SPEAKER_01: of program,
• 8:35
  SPEAKER_01: ball dance,
• 8:35
  SPEAKER_01: whatever,
• 8:36
  SPEAKER_01: that are actually doing it right.
• 8:39
  SPEAKER_01: And they're dancing like nobody's watching, to extend the analogy.
• 8:46
  SPEAKER_01: There's a bit to unpack in regards to like,
• 8:50
  where we need to navigate towards and what the insights out of
• 8:54
  this report should
• 8:55
  SPEAKER_01: end up meeting towards.
• 8:57
  SPEAKER_00: Yeah,
• 8:58
  SPEAKER_00: and I think at the end of the day,
• 9:00
  SPEAKER_00: for that 5% who's dancing,
• 9:02
  SPEAKER_00: I don't think they're actually getting things right either.
• 9:04
  SPEAKER_00: They're figuring things out as they go along.
• 9:07
  And the main thing that sets them apart is more of the courage
• 9:10
  to get things going,
• 9:12
  SPEAKER_00: to make that mistakes and to have that silver shield down,
• 9:16
  SPEAKER_00: that vulnerability out,
• 9:17
  that they're willing to make mistakes in a way that set them
• 9:21
  apart for being that
• 9:22
  SPEAKER_00: 5% to dance on the dance floor.
• 9:24
  And I think the courage to measure what matters,
• 9:27
  I think that's not easy.
• 9:28
  So I think I take my hat off to them and the courage to fail
• 9:32
  fast instead of failing slow.
• 9:34
  SPEAKER_00: Because it's always easy to talk about things.
• 9:36
  SPEAKER_00: It's easy to point things out.
• 9:37
  SPEAKER_00: But it's not easy to actually do things and just stand up and say,
• 9:40
  SPEAKER_00: hey,
• 9:41
  SPEAKER_00: I'm really doing it,
• 9:41
  SPEAKER_00: butt off.
• 9:42
  SPEAKER_00: And let me get my mistakes done and we'll see again what happens next.
• 9:49
  SPEAKER_00: And let's get real.
• 9:50
  SPEAKER_01: If we're going to be innovative, we've got to lead from the front.
• 9:54
  SPEAKER_00: It's like,
• 9:55
  SPEAKER_00: you know,
• 9:55
  SPEAKER_00: you got to get real about what failure looks like,
• 9:58
  SPEAKER_00: because I think we're all picturing robots and explosions.
• 10:01
  Like that's how we were influenced by the Hollywood then about
• 10:05
  how AI looked.
• 10:07
  SPEAKER_00: But at this point, when I look back, it's like, it's pretty different.
• 10:10
  Although there is some resemblance about what Hollywood did
• 10:13
  visualize about how AI looks like at
• 10:16
  SPEAKER_00: from the 2000s?
• 10:17
  SPEAKER_01: I'm waiting for Jarvis myself.
• 10:19
  I want to be Tony Stark in Iron Man where I'm just talking to
• 10:23
  my Jarvis and it's
• 10:24
  SPEAKER_01: doing everything for me.
• 10:26
  SPEAKER_01: Meanwhile, I'm moving my hands and pulling strings.
• 10:29
  SPEAKER_01: We've got a lot of building and for the most part,
• 10:32
  SPEAKER_01: when we think about most of what we're talking about now,
• 10:36
  SPEAKER_01: Even though the technology has been around for some time,
• 10:39
  SPEAKER_01: artificial intelligence is not necessarily new.
• 10:42
  SPEAKER_01: But this latest wave and the progress that we made is very much new.
• 10:49
  SPEAKER_01: And within this wave, we're still very early...
• 10:54
  SPEAKER_01: early days.
• 10:55
  SPEAKER_01: While we're talking about approving budgets for AI initiatives,
• 11:00
  SPEAKER_01: for the most part,
• 11:01
  SPEAKER_01: it's just going to be expensive screensavers.
• 11:03
  It's the equivalent of buying an expensive paperweight or
• 11:06
  something that is tucked
• 11:07
  SPEAKER_01: into the folder of what didn't work.
• 11:11
  SPEAKER_00: And MIT calls it the pilot purgatory.
• 11:14
  SPEAKER_00: And, you know, I felt that.
• 11:16
  SPEAKER_00: And only 5% of,
• 11:17
  SPEAKER_00: if you really think about it,
• 11:18
  the study also shows that only 5% of custom enterprise AI tools
• 11:22
  ever make it to
• 11:23
  SPEAKER_00: production.
• 11:24
  SPEAKER_00: Think about that.
• 11:25
  SPEAKER_00: 95% of these projects die in a pilot phase.
• 11:28
  Then again,
• 11:29
  it begs the question,
• 11:30
  why Lovable is getting such noticeability?
• 11:34
  Is it because of this,
• 11:35
  whereby people are actually actively putting them ideas in,
• 11:39
  SPEAKER_00: and actually getting their ideas out.
• 11:42
  SPEAKER_00: Is that reason why Liverpool is coming out?
• 11:44
  SPEAKER_00: Is it why Liverpool is so loveable?
• 11:47
  SPEAKER_01: No,
• 11:47
  SPEAKER_01: so like,
• 11:48
  I'm going to get into a little bit of the controversial and
• 11:51
  some of my viewpoint on
• 11:52
  SPEAKER_01: this because I think there's a couple of things that are going on.
• 11:55
  Where you see a lot of these pilots taking place are really
• 12:00
  unlike the front of the
• 12:02
  house sort of operations where a lot of the implementations are
• 12:07
  looking at sales,
• 12:08
  SPEAKER_01: marketing,
• 12:09
  SPEAKER_01: automation.
• 12:10
  SPEAKER_01: It's looking at like the sexy top line sort of efforts.
• 12:16
  And what ends up happening most times is a lot of this is not
• 12:20
  really generating the
• 12:22
  SPEAKER_01: expected outcomes.
• 12:24
  SPEAKER_01: We've all gotten these horrible emails, these mass AI-generated emails.
• 12:29
  We've all seen the marketing content of really horrible stuff
• 12:34
  that's obviously not
• 12:36
  SPEAKER_01: crossing the uncanny valley.
• 12:38
  SPEAKER_01: And we see a bunch of that.
• 12:40
  And then you have a handful of companies that are trying to
• 12:43
  build like really solid
• 12:45
  SPEAKER_01: enterprise products that can do well.
• 12:47
  SPEAKER_01: Some of them are in those like sexy top line areas,
• 12:50
  SPEAKER_01: but a lot of like the real ROI sits on the back office side.
• 12:54
  SPEAKER_01: And we don't really see a lot of adoption on, on those levels.
• 12:59
  SPEAKER_01: Um,
• 13:00
  SPEAKER_01: The other aspect is we talk about the workflows.
• 13:05
  And so when you look at one of the things that this report
• 13:09
  highlighted as well was that
• 13:12
  SPEAKER_01: Even when you have a vertical AI solution purpose-built for a company,
• 13:19
  SPEAKER_01: maybe it's an in-house project,
• 13:20
  SPEAKER_01: maybe it's a third party,
• 13:22
  but people have a tendency of still reverting back to these
• 13:26
  generic tools because
• 13:28
  SPEAKER_01: it's a habit,
• 13:29
  SPEAKER_01: it's familiarity,
• 13:30
  and likely there's a falling apart in regards to the way that
• 13:34
  these purpose-built
• 13:35
  SPEAKER_01: ones work.
• 13:36
  are actually integrating with the workflows the way they're
• 13:39
  intuitively designed.
• 13:40
  So it's not really a technology problem per se because they're
• 13:44
  using it somewhere.
• 13:47
  SPEAKER_01: It's a matter,
• 13:47
  SPEAKER_01: to me,
• 13:48
  it's a lot of like product design for the enterprise software
• 13:52
  builders or for the
• 13:53
  SPEAKER_01: in-house projects.
• 13:57
  SPEAKER_00: Right.
• 13:57
  So again,
• 13:58
  it's what you said earlier,
• 14:00
  it's very much front-based rather than back-end.
• 14:03
  That's why we have these issues of back-end not being able to
• 14:07
  connect to the front-end.
• 14:10
  SPEAKER_00: That could be also a people problem again,
• 14:13
  because why I'm saying that is like all these technology
• 14:16
  solutions are built by
• 14:17
  SPEAKER_00: people.
• 14:18
  right and they focus so much on the front end that the back end
• 14:22
  loses out but
• 14:23
  ultimately if you really drill down to it it's still a people
• 14:26
  problem it's a trust
• 14:27
  problem it's a it's a this doesn't fit how i actually work the
• 14:31
  fact that people use
• 14:32
  so much of ai and yet they still go back to the vertical um
• 14:37
  purpose-built uh
• 14:38
  systems says a lot you know uh break it down for us like how
• 14:42
  does that you know why
• 14:43
  SPEAKER_00: why is that biggest issue
• 14:46
  SPEAKER_01: Yeah,
• 14:46
  SPEAKER_01: so like one of the examples that they throw out into the report,
• 14:49
  SPEAKER_01: and they didn't say what company or what software they were using,
• 14:53
  SPEAKER_01: but they referenced a law firm that instead of using the actual legal,
• 14:59
  SPEAKER_01: the design product for legal services,
• 15:02
  SPEAKER_01: and there's a few out there,
• 15:03
  SPEAKER_01: like the most recognizable name brand is Harvey,
• 15:06
  SPEAKER_01: but like who knows which one they were talking about.
• 15:08
  SPEAKER_01: Maybe it was in an household one.
• 15:10
  But it referenced that the people that they spoke to apparently
• 15:13
  are still going
• 15:13
  SPEAKER_01: over to Ope Chat GPT and querying there.
• 15:17
  SPEAKER_01: You don't have the trust factor because it's not purpose-filled.
• 15:21
  SPEAKER_01: It's not ring-fenced on the data set.
• 15:24
  SPEAKER_01: So you have a little bit of proneness to error that is there.
• 15:29
  And so it draws the question that people in existing processes
• 15:33
  are still getting in the way.
• 15:35
  SPEAKER_01: The explosion of the generic tools like ChatGPT and these,
• 15:41
  they've already started developing habits out of the
• 15:45
  individuals that when they're
• 15:46
  SPEAKER_01: in their workplace,
• 15:49
  they're reverting back to the familiar and querying in these
• 15:54
  places as opposed to
• 15:55
  SPEAKER_01: in the tools.
• 15:56
  SPEAKER_01: So there's
• 15:58
  SPEAKER_01: So it's really just kind of a first wave problem in my viewpoint.
• 16:03
  SPEAKER_01: And it's also a product design issue.
• 16:06
  that the products need to become more intuitive and they need
• 16:09
  to be built more into
• 16:10
  the workflows where it's harder for you to go outside the
• 16:13
  system to ask that trashy
• 16:15
  etc instead of just sitting inside of your existing workflow
• 16:20
  your existing systems
• 16:22
  your existing windows etc so there's like this integration
• 16:26
  piece that goes along
• 16:28
  SPEAKER_01: with it like
• 16:29
  You can't just have these standalone tools and think that
• 16:34
  they're going to end up
• 16:35
  SPEAKER_01: breaking out these other habits.
• 16:37
  You have to kind of integrate further within the workflows of
• 16:44
  users in order to be successful.
• 16:47
  SPEAKER_00: So in essence,
• 16:47
  SPEAKER_00: it's trying to back it up with stronger,
• 16:50
  more robust back-end to make sure that it's a wholesome
• 16:54
  experience as one user uses
• 16:56
  SPEAKER_00: it,
• 16:56
  SPEAKER_00: not just the front-end,
• 16:58
  SPEAKER_00: like what we face with chatshipping nowadays of hallucination problems.
• 17:03
  at least they're able to churn out from a black box about how
• 17:07
  the users use it so
• 17:09
  that there's some reliability and consistency in the results
• 17:12
  that need to extend
• 17:14
  SPEAKER_00: the user's expectations,
• 17:16
  SPEAKER_00: not like you put in something and then you get a total different answer.
• 17:19
  SPEAKER_00: And it gets frustrating,
• 17:21
  especially when you have the expectation of it trying to
• 17:24
  smoothen or streamline
• 17:25
  SPEAKER_00: your workflow and instead you're not getting what you want.
• 17:27
  SPEAKER_00: I don't have any live analogy for this, but anyway.
• 17:30
  SPEAKER_00: But that sounds...
• 17:31
  SPEAKER_00: exhausting for everyone involved, really.
• 17:36
  There's that trust factor involved there in regards to can you
• 17:41
  really rely upon it
• 17:43
  SPEAKER_01: in order to produce what you actually need
• 17:47
  And so that's an area that we need to overcome from a product
• 17:52
  design standpoint,
• 17:54
  SPEAKER_01: from the aspect in regards to how it's being implemented.
• 17:59
  And we also need a lot of the implementations that they've been
• 18:03
  talking about.
• 18:04
  So there's a couple of additional points that I think are
• 18:07
  useful to highlight out
• 18:09
  SPEAKER_01: of this,
• 18:10
  SPEAKER_01: which is the in-house build process.
• 18:13
  SPEAKER_01: versus the third party tools.
• 18:16
  And one of the things that the report highlighted is that a lot
• 18:20
  of companies are
• 18:20
  SPEAKER_01: trying to go it alone and build their own internal tools.
• 18:25
  SPEAKER_01: That has a significantly higher failure rate.
• 18:28
  SPEAKER_01: than if they just bought a third party tool.
• 18:31
  And it's the same as when we saw like the early iterations of
• 18:35
  software as a service.
• 18:37
  A lot of companies were trying to do these Frankenstein's
• 18:40
  monsters of assembling
• 18:42
  SPEAKER_01: their own patch together.
• 18:44
  SPEAKER_01: Here's what we specifically need sort of creations.
• 18:48
  And oftentimes it just didn't work for a multitude of reasons
• 18:52
  that we can dig into if we want.
• 18:54
  But it failed then,
• 18:56
  taking the same approach now of trying to go it alone and build
• 19:00
  it yourself.
• 19:01
  There's the difference between specialization,
• 19:04
  scaled problems,
• 19:05
  non-scale problems.
• 19:07
  There's instances where it will make sense for companies to
• 19:10
  build in-house.
• 19:12
  SPEAKER_01: But for the most part,
• 19:13
  SPEAKER_01: a lot of the difference,
• 19:15
  because we're not all the way there in later waves of AI where
• 19:19
  maybe it will
• 19:19
  SPEAKER_01: change.
• 19:20
  SPEAKER_01: But at the present moment,
• 19:22
  a lot of companies need to switch up their manner and start
• 19:26
  thinking about
• 19:27
  SPEAKER_01: third-party tools because the success rate there is significantly higher.
• 19:33
  SPEAKER_01: So that's the one thing.
• 19:34
  SPEAKER_01: The second thing is coming back to this front versus back.
• 19:40
  SPEAKER_01: A lot of the intention,
• 19:44
  SPEAKER_01: oftentimes when you're talking about initiatives,
• 19:47
  SPEAKER_01: it's very top line focus.
• 19:49
  SPEAKER_01: But what is AI really good at?
• 19:53
  SPEAKER_01: It's efficiency,
• 19:54
  SPEAKER_01: it's productivity,
• 19:56
  SPEAKER_01: it's doing more with less,
• 19:57
  SPEAKER_01: it's automations,
• 19:59
  SPEAKER_01: it's making us move faster than we did before.
• 20:02
  SPEAKER_01: And so sure,
• 20:04
  SPEAKER_01: it may make sense to really look at the top line revenue generation,
• 20:09
  SPEAKER_01: sales,
• 20:10
  SPEAKER_01: marketing,
• 20:10
  SPEAKER_01: et cetera.
• 20:12
  But when we think of efficiency,
• 20:14
  productivity,
• 20:15
  all of that,
• 20:16
  that's automate my back office.
• 20:19
  SPEAKER_01: Maybe I can cut my accounting department in half.
• 20:22
  SPEAKER_01: Maybe my supply chain management team can get cut in half.
• 20:27
  SPEAKER_01: Maybe my warehouse ops team can get cut in half.
• 20:32
  SPEAKER_01: Maybe this team, maybe that team.
• 20:34
  SPEAKER_01: A lot of these areas in which I can equip and better tool my team's
• 20:39
  SPEAKER_01: and start integrating and changing some of the workflows.
• 20:45
  There's a lot there that companies presently aren't putting
• 20:49
  enough attention on.
• 20:50
  That is one of the areas where the report highlighted is that's
• 20:55
  an area where
• 20:55
  SPEAKER_01: there's a lot of ROI potential,
• 20:58
  SPEAKER_01: but not a lot of these pilot projects are really going after present.
• 21:03
  SPEAKER_00: But then again, the way that I'm hearing it is like,
• 21:07
  You know how banks,
• 21:08
  there's so much of legacy problems that they don't get solved?
• 21:11
  It's because they integrate all these new different pieces and
• 21:15
  then they get to it
• 21:16
  SPEAKER_00: at some point in their lifespan of the bank and it doesn't get done.
• 21:23
  SPEAKER_00: It doesn't get solved.
• 21:24
  SPEAKER_00: It feels like it's a repetition of it.
• 21:26
  just that it's newer age with ai i think that's why a lot of
• 21:31
  tools are out there a
• 21:32
  lot of them don't get through the pilot phase it's also maybe
• 21:35
  because of this issue
• 21:37
  whereby banks have come legacy companies have gone through that
• 21:40
  and they're afraid
• 21:41
  to take this step to make sure that everything's integrated
• 21:44
  because it's a lot of
• 21:45
  work even getting things to change you you have with like
• 21:49
  having the you know
• 21:50
  SPEAKER_00: machine to talk to each other like
• 21:52
  SPEAKER_00: Your AI can't possibly talk to your CRM.
• 21:55
  SPEAKER_00: It doesn't know your customer history.
• 21:57
  SPEAKER_00: It can't access your financial data inventory system.
• 22:00
  So you've got this incredibly powerful tool that knows
• 22:02
  absolutely nothing about
• 22:04
  SPEAKER_00: your actual business.
• 22:05
  SPEAKER_00: And then you're having it run.
• 22:06
  SPEAKER_00: And then it's like, how do you integrate these pieces?
• 22:08
  SPEAKER_00: Because it takes a lot of work to get them moving all in one go.
• 22:12
  SPEAKER_00: So I'm just wondering how would that look like?
• 22:15
  SPEAKER_00: Because...
• 22:17
  SPEAKER_00: As it is,
• 22:18
  SPEAKER_00: as much as we're trying so hard to move away from that legacy issue,
• 22:22
  SPEAKER_00: it feels like we're falling back again in another way.
• 22:27
  SPEAKER_01: Just a footfall thought.
• 22:28
  SPEAKER_01: This is where we're going.
• 22:30
  SPEAKER_01: This is where we're getting to,
• 22:32
  SPEAKER_01: where you essentially have your knowledge base within your organization,
• 22:39
  SPEAKER_01: which are these systems.
• 22:40
  SPEAKER_01: It is your CRM.
• 22:42
  SPEAKER_01: It is the history that you've captured there.
• 22:45
  SPEAKER_01: It is your financial data, your inventory systems.
• 22:49
  SPEAKER_01: It is all of that.
• 22:51
  SPEAKER_01: If you think about from an AI perspective,
• 22:56
  SPEAKER_01: the OpenAI,
• 22:58
  SPEAKER_01: ChatGPT,
• 22:58
  SPEAKER_01: these sort of models,
• 22:59
  they're basically crawling the web and digesting all of this
• 23:03
  outside information to
• 23:05
  SPEAKER_01: where you can essentially query it and get answers,
• 23:08
  SPEAKER_01: get feedback,
• 23:09
  SPEAKER_01: get basically this nice conversation going.
• 23:13
  There's a lot of ability in order for us to start implementing
• 23:17
  AI systems that can do the same.
• 23:20
  SPEAKER_01: Instead of crawling the web for public data,
• 23:24
  SPEAKER_01: there's companies out there that are starting to ring fence these things.
• 23:28
  SPEAKER_01: focusing in specifically on internal company knowledge base.
• 23:33
  SPEAKER_01: Link up your Slack conversations.
• 23:36
  SPEAKER_01: Link up your CRM.
• 23:37
  SPEAKER_01: There's AI-first CRM systems that are out there right now.
• 23:42
  SPEAKER_01: We're piloting one ourselves.
• 23:44
  SPEAKER_01: It's coming.
• 23:46
  SPEAKER_01: There's still a lot of sensitivity around it,
• 23:49
  SPEAKER_01: especially for bigger companies because they are worried about the data.
• 23:53
  SPEAKER_01: Is it training your models?
• 23:54
  SPEAKER_01: I don't want my data ending up out there.
• 23:57
  SPEAKER_01: But
• 23:58
  I do think that we're going to start seeing a lot of
• 24:02
  interesting innovations,
• 24:03
  SPEAKER_01: but these are the areas where not enough attention
• 24:07
  SPEAKER_01: in these pilots has it really been driven at,
• 24:10
  SPEAKER_01: is going after these areas where your operations teams,
• 24:16
  SPEAKER_01: yes,
• 24:16
  SPEAKER_01: your sales teams,
• 24:17
  if you can get quick call sheets on your upcoming meetings
• 24:22
  generated by a sensor
• 24:25
  SPEAKER_01: device while you're in transit to go meet a client,
• 24:28
  SPEAKER_01: It gives you the data digest out of your CRM.
• 24:31
  SPEAKER_01: These are all cool things that can end up improving.
• 24:33
  SPEAKER_01: I mean, you may not see tangible ROI at the beginning.
• 24:36
  SPEAKER_01: Where you do see quicker is when you start thinking of AI agents,
• 24:42
  SPEAKER_01: AI implementations,
• 24:44
  SPEAKER_01: AI automation.
• 24:45
  It's where we're going that we can start being able to get the
• 24:50
  productivity gains.
• 24:52
  SPEAKER_01: And that's where, in my mind, the...
• 24:56
  SPEAKER_01: the lower hanging fruit of AI is really going to come from.
• 25:00
  But it's a different workflow in regards to how do you manage
• 25:04
  the implementation at
• 25:06
  SPEAKER_01: an enterprise level to really extract it.
• 25:09
  So there's this delicate balance of you need it to kind of fit
• 25:13
  seamlessly in with
• 25:15
  SPEAKER_01: the way you're currently doing it.
• 25:17
  Because otherwise you have to do this whole change management
• 25:20
  in an organization,
• 25:21
  SPEAKER_01: retraining your employees,
• 25:23
  SPEAKER_01: changing workflows,
• 25:24
  SPEAKER_01: all of this.
• 25:25
  So like the challenge right now is more so how can we start
• 25:29
  getting this sort of
• 25:31
  symbiosis between these two streams to where we're not
• 25:34
  necessarily shocking the
• 25:36
  SPEAKER_01: system too much,
• 25:37
  SPEAKER_01: but we're being able to amplify these waves that we're on.
• 25:43
  So it's like what I'm hearing as well is like instead of
• 25:46
  looking big scale
• 25:47
  SPEAKER_00: enterprise transformation,
• 25:48
  we're looking at filtering it at each touch point so that it
• 25:52
  helps streamline the
• 25:53
  SPEAKER_00: operation and make life a bit easier is what I'm hearing.
• 25:57
  SPEAKER_00: It's like we start with individual value and then we scale up from there.
• 26:01
  SPEAKER_00: I mean, we can't try to boil the ocean, right?
• 26:03
  Make one person's Tuesday or Monday better and then worry about
• 26:07
  the whole big thing going on.
• 26:09
  SPEAKER_00: And what about for leaders who are panicking about
• 26:13
  security risk because as it is like when we look at from an
• 26:18
  individual standpoint
• 26:19
  it's like they are already proving demand right like what
• 26:21
  you're saying they're
• 26:22
  showing that why AI actually helps instead of hinders
• 26:27
  eventually the take up rate
• 26:28
  SPEAKER_00: or the people will actually embrace it even more and that's when um
• 26:33
  SPEAKER_00: AI is going to take shape even more.
• 26:35
  SPEAKER_00: So we will actually see results that we're actually moving the needle.
• 26:38
  SPEAKER_00: And I think that's a playbook right there.
• 26:39
  SPEAKER_00: Like,
• 26:40
  SPEAKER_00: you know,
• 26:40
  SPEAKER_00: for any startup founders listening,
• 26:42
  SPEAKER_00: your customers are already using AI and,
• 26:44
  SPEAKER_00: you know,
• 26:45
  SPEAKER_00: they're just using a free generic versions.
• 26:47
  SPEAKER_00: Perhaps you could start thinking about building them something better.
• 26:50
  SPEAKER_00: Yeah.
• 26:50
  SPEAKER_00: Back to my question again.
• 26:51
  What about leaders who are panicking about shadow IT and,
• 26:54
  you know,
• 26:54
  security risk?
• 26:55
  SPEAKER_00: What does that look like?
• 27:05
  SPEAKER_01: Yeah, so I mean, the security risk is obviously an issue.
• 27:13
  SPEAKER_01: A lot of companies are operating off of sensitive private information.
• 27:18
  They don't want their employees going on to a public LLM and
• 27:24
  typing it into
• 27:25
  SPEAKER_01: something that we're getting trained off of every single problem.
• 27:29
  SPEAKER_01: There's a lot more solutions.
• 27:31
  This is one of the reasons why companies are doing these pilots
• 27:34
  and why they're
• 27:35
  SPEAKER_01: trying to implement solutions specific to their enterprise,
• 27:40
  because they're ring-fenced to where the data that's going into
• 27:43
  them are not
• 27:43
  SPEAKER_01: training outside models.
• 27:45
  SPEAKER_01: They don't want that leakage.
• 27:46
  They don't want the security risks that potentially come from
• 27:50
  that as well.
• 27:52
  That is an issue and that's going to continue to be an issue as
• 27:56
  long as we have
• 27:57
  this adoption challenge in regards to are people using outside
• 28:05
  generic models
• 28:06
  instead of the ones that the companies are trying to adopt and
• 28:12
  integrate in with
• 28:13
  SPEAKER_01: the organization.
• 28:15
  SPEAKER_00: Right, right.
• 28:16
  SPEAKER_00: And to a certain extent,
• 28:18
  SPEAKER_00: if you draw parallels between Southeast Asia and the West,
• 28:22
  SPEAKER_00: like Southeast Asia,
• 28:24
  SPEAKER_00: I think we do have that cultural willingness to experiment.
• 28:27
  SPEAKER_00: And this is a massive advantage, whereas...
• 28:32
  SPEAKER_00: Western companies are forming AI ethics committees.
• 28:35
  SPEAKER_00: All right, so enough about the 95%.
• 28:37
  Let's talk about the 5% who are absolutely crushing it because
• 28:40
  they exist and
• 28:41
  they're making millions and they're doing things the rest of us
• 28:44
  can learn from.
• 28:45
  SPEAKER_00: So what do they look like?
• 28:48
  SPEAKER_01: I mean,
• 28:48
  SPEAKER_01: first off,
• 28:49
  SPEAKER_01: they're not looking at it as just like some sort of magic bullet that is
• 28:53
  SPEAKER_01: automatically going to magically end up creating ROI.
• 28:59
  SPEAKER_01: What they're doing is they're treating it like it is.
• 29:02
  SPEAKER_01: It's a powerful tool that requires an actual strategic approach.
• 29:08
  SPEAKER_01: It requires integration into the organization,
• 29:11
  SPEAKER_01: which means that there's adoption,
• 29:14
  SPEAKER_01: etc.,
• 29:14
  SPEAKER_01: and requires ongoing optimization.
• 29:17
  SPEAKER_01: It's not a one-time, here you go, now we're done.
• 29:20
  SPEAKER_01: It requires ongoing optimization.
• 29:24
  SPEAKER_00: And what would be the examples?
• 29:25
  SPEAKER_01: When you look at the ones that are really building off of it,
• 29:29
  SPEAKER_01: you know,
• 29:30
  SPEAKER_01: first off,
• 29:31
  a lot of the ones that I see that are most successful are
• 29:35
  really the smaller
• 29:35
  SPEAKER_01: companies because they're at the benefit.
• 29:37
  They don't have the hierarchy,
• 29:40
  the process,
• 29:41
  the rigidness in order to be able to move forward.
• 29:45
  SPEAKER_01: Because the larger you are,
• 29:47
  the harder it is to really adopt and move forward because you
• 29:51
  have legacy of
• 29:54
  SPEAKER_01: people,
• 29:54
  SPEAKER_01: processes,
• 29:56
  SPEAKER_01: etc.
• 29:57
  SPEAKER_01: Whereas smaller organizations,
• 30:00
  SPEAKER_01: instead of making that extra hire as you're growing,
• 30:03
  you simply adopt more technology and you get more efficiency
• 30:07
  out of your existing
• 30:08
  SPEAKER_01: team.
• 30:09
  SPEAKER_01: Whereas larger organizations,
• 30:11
  they're trying to figure out how can they end up managing off
• 30:14
  of the players that
• 30:15
  SPEAKER_01: they have.
• 30:16
  SPEAKER_01: When you look across any tech department,
• 30:19
  SPEAKER_01: everybody's using different sorts of code writer tools,
• 30:23
  SPEAKER_01: copilots.
• 30:24
  SPEAKER_01: There's tons that are out there.
• 30:26
  SPEAKER_01: You know, you ask a bunch of folks out there.
• 30:29
  SPEAKER_01: They've all got their favorites.
• 30:30
  SPEAKER_01: They kind of generally sink around a handful of them.
• 30:34
  SPEAKER_01: But like your 10x dev is now like 100x.
• 30:39
  There's becoming like one of the things that I didn't talk
• 30:41
  about is there's a Gen
• 30:42
  AI divide between like the developers who are using it and the
• 30:45
  ones that aren't.
• 30:46
  SPEAKER_01: Because like the bottom tier is still like a super far bottom tier.
• 30:49
  SPEAKER_01: They haven't changed very much.
• 30:51
  SPEAKER_01: But the top tier people, they are doing so much more.
• 30:55
  They are becoming so much more efficient and they're building
• 30:58
  so much more.
• 30:59
  there's still the early stage flaws of you may be accumulating
• 31:06
  technical debt if
• 31:07
  you're not careful and if you're not making sure that you're
• 31:10
  fulfilling all of the
• 31:11
  SPEAKER_01: context that it necessarily needs to build.
• 31:14
  SPEAKER_01: But then if you transition over into organizations,
• 31:18
  there's still a lot of opportunity where companies are starting
• 31:20
  to implement it
• 31:21
  SPEAKER_01: into their processes.
• 31:22
  SPEAKER_01: Some are not the sexy things that you see all over the place,
• 31:27
  SPEAKER_01: but a lot of them are just kind of like,
• 31:28
  SPEAKER_01: Figuring out the roles that could end up being automated.
• 31:33
  Figuring out ways that they can better leverage the knowledge
• 31:36
  bases within the organizations.
• 31:38
  SPEAKER_01: How can they implement in order to get better communications?
• 31:42
  SPEAKER_01: There's tons of companies out there.
• 31:44
  SPEAKER_01: Robotic process automation is an old form.
• 31:48
  There's long been an industry built around simple logic of if
• 31:55
  this,
• 31:56
  then that.
• 31:57
  if i get an inbound email that said that comes from it comes
• 32:01
  into this email
• 32:02
  account then send it over to this crm pipeline and put them on
• 32:08
  this uh email
• 32:10
  campaign for example and that's that's that's been around and
• 32:14
  that's that's just
• 32:14
  one example there's a there's a bunch of like if then then if
• 32:18
  if this then that
• 32:19
  SPEAKER_01: sort of scenarios but
• 32:22
  SPEAKER_01: When you think about AI implementations,
• 32:24
  SPEAKER_01: you can start implementing it at a higher degree of complexity.
• 32:31
  SPEAKER_01: And there are companies out there that are being more innovative,
• 32:34
  SPEAKER_01: that are taking the steps and being more strategic about the way they're
• 32:37
  SPEAKER_01: implementing,
• 32:38
  SPEAKER_01: integrating,
• 32:39
  SPEAKER_01: and optimizing over time.
• 32:41
  And you see them doing massive amounts of automation of routine
• 32:45
  inquiries.
• 32:45
  Sure,
• 32:46
  we can talk about how in the long run there may be job
• 32:49
  displacement.
• 32:50
  SPEAKER_01: Being a tech optimist,
• 32:52
  I still look at it and say humans are incredibly creative at
• 32:55
  figuring out more work
• 32:56
  SPEAKER_01: to do.
• 32:57
  If we get a productivity tool,
• 32:59
  we're just going to figure out more stuff to do.
• 33:02
  SPEAKER_01: What happened when we got calculators or computers or like any,
• 33:05
  SPEAKER_01: like name any sort of technological innovation.
• 33:08
  The first time somebody was able to do a Google search and the
• 33:12
  internet was at your fingertips.
• 33:13
  Like we've always figured out more stuff to do and we're still
• 33:17
  going to do it.
• 33:18
  The key thing coming back to like what separates the winners
• 33:20
  and losers is the
• 33:22
  SPEAKER_01: focus around the use case.
• 33:24
  And being very intentional about the value that you're trying
• 33:29
  to extract out of it
• 33:30
  and looking at it from the lens of where are the areas which I
• 33:34
  can extract
• 33:35
  SPEAKER_01: productivity gains that really move the needle.
• 33:40
  And making sure that you're strategically implementing it and
• 33:44
  that you're putting
• 33:44
  SPEAKER_01: it into the culture of the organization.
• 33:48
  But there's this aspect of round like change management that
• 33:52
  does come into play.
• 33:53
  SPEAKER_01: So you're trying to not shock the system too much in the beginning.
• 33:57
  But trying to build those habit formations amongst your people
• 34:01
  so that they're not
• 34:02
  SPEAKER_01: going out to those generic tools of like,
• 34:05
  SPEAKER_01: instead of using the one that's on my office laptop,
• 34:08
  SPEAKER_01: I'm just going to pull up my web browser and go to ChatGPT.
• 34:12
  SPEAKER_01: Sorry, I'm throwing them under the bus just referencing them.
• 34:14
  SPEAKER_01: But they got that brand value.
• 34:16
  SPEAKER_01: So they're the first name that pops into my head.
• 34:19
  SPEAKER_01: But it is that.
• 34:20
  So we're looking at companies that are really augmenting,
• 34:23
  not necessarily replacing.
• 34:26
  SPEAKER_01: We still see call centers are actually got an uptick in employment,
• 34:30
  SPEAKER_01: one of the last statistics that I saw.
• 34:32
  So we still have not necessarily full out replacement,
• 34:36
  but we're seeing augmentation for sure.
• 34:39
  And I think this brings it to another perspective,
• 34:42
  I would say,
• 34:43
  like augmentation.
• 34:45
  It depends,
• 34:45
  again,
• 34:46
  on how much you use it,
• 34:47
  that it benefits you in the long run.
• 34:50
  And then to another end,
• 34:51
  for example,
• 34:53
  JP Morgan,
• 34:54
  maybe,
• 34:54
  they do have their own...
• 34:58
  SPEAKER_00: in-house AI tool which replaces analysts.
• 35:02
  SPEAKER_00: So on that end, it's like they're kind of displacing future.
• 35:09
  SPEAKER_00: So this is another divide again.
• 35:12
  SPEAKER_00: So on the second layer, what I'm trying to get at is that
• 35:16
  SPEAKER_00: I'm trying to re-emphasize it's also a people problem as well.
• 35:20
  SPEAKER_00: It,
• 35:20
  SPEAKER_00: again,
• 35:20
  SPEAKER_00: depends on how the company looks at it,
• 35:23
  the perspective they have of how AI would help them in the long
• 35:27
  run to take
• 35:28
  SPEAKER_00: advantage of AI.
• 35:29
  SPEAKER_00: For example,
• 35:30
  SPEAKER_00: like VBS with a new CEO,
• 35:32
  SPEAKER_00: Tan Sushan,
• 35:34
  SPEAKER_00: she recently had a press conference,
• 35:37
  but to quote her is that she was saying that she looks for
• 35:40
  people now who have...
• 35:43
  SPEAKER_00: a positive attitude towards their work.
• 35:45
  And they would want people who come on new hires to be AI
• 35:49
  competent,
• 35:49
  to be AI capable.
• 35:51
  And I think this says a lot about how progressive you are in
• 35:54
  your thinking in the
• 35:55
  sense that how much of AI that you are going to use in your
• 35:58
  workforce that's going
• 35:59
  SPEAKER_00: to help
• 36:00
  SPEAKER_00: augment the success of the company in short,
• 36:04
  SPEAKER_00: rather than using it to displace some future talent.
• 36:11
  I think for a company to actually succeed,
• 36:14
  it needs to marry both talent and tech.
• 36:19
  So I think re-emphasizing that it has to be augmentation rather
• 36:22
  than anything else
• 36:23
  SPEAKER_00: to actually see how AI is moving the needle.
• 36:27
  SPEAKER_01: I mean,
• 36:28
  I find that definitely interesting in regards to what the
• 36:32
  future of the workforce
• 36:33
  SPEAKER_01: will end up coming out as.
• 36:35
  I mean,
• 36:35
  when you look at Asia in general,
• 36:37
  we've got a very young population.
• 36:40
  SPEAKER_01: And for the most part,
• 36:41
  SPEAKER_01: because they're young,
• 36:42
  SPEAKER_01: they've been experimenting with a lot of the technologies.
• 36:44
  SPEAKER_01: So like...
• 36:46
  SPEAKER_01: I do believe that given the right context,
• 36:51
  that they can be experimental and that we'll be capable of
• 36:55
  moving fast forward.
• 36:58
  SPEAKER_01: We just need organizations that have the willingness to experiment,
• 37:03
  SPEAKER_01: that have the willingness to move for faster decision making.
• 37:07
  SPEAKER_01: Because that is what that adoption is going to end up requiring.
• 37:12
  So if we can marry up the idea of looking for the people that
• 37:17
  have the right
• 37:18
  SPEAKER_01: attitude,
• 37:21
  SPEAKER_01: combining that with organizations that also have the attitude of speed,
• 37:26
  SPEAKER_01: experimentation,
• 37:27
  SPEAKER_01: and moving things forward,
• 37:30
  SPEAKER_01: You could end up seeing Southeast Asia move very,
• 37:33
  SPEAKER_01: very fast and be a compelling market in regards to adoption.
• 37:39
  SPEAKER_00: Yeah, and I guess that's corporate marriage in a way.
• 37:44
  Marrying people competent and having the right tech to move the
• 37:49
  companies,
• 37:49
  SPEAKER_00: propel the companies forward really.
• 37:51
  SPEAKER_00: And Asia Pacific is leading in growth.
• 37:53
  SPEAKER_00: AI app downloads are up 80% versus 51% in Europe and 39% in the States.
• 38:00
  SPEAKER_00: That's funny,
• 38:02
  SPEAKER_00: because Chad GPT is from the States,
• 38:04
  SPEAKER_00: and then you have Liverpool from Europe,
• 38:06
  SPEAKER_00: and then there's Figma from the States.
• 38:09
  SPEAKER_00: That's interesting.
• 38:10
  SPEAKER_00: And there's this hunger to leapfrog that I find so energising.
• 38:16
  SPEAKER_01: Yeah, I mean, it's not uniform across the board.
• 38:18
  SPEAKER_01: So, like, we can't necessarily paint Asia with one single brush.
• 38:23
  SPEAKER_01: But certainly, like, Singapore, Hong Kong, the business centers there.
• 38:29
  When you start thinking about the rest of Southeast Asia,
• 38:32
  Indonesia,
• 38:32
  Thailand,
• 38:33
  Malaysia.
• 38:35
  SPEAKER_01: Also, like, Philippines, young populations.
• 38:38
  SPEAKER_01: Well, maybe Malaysia and Thailand are a little bit on the older side.
• 38:42
  SPEAKER_01: But generally, young populations that are eager.
• 38:47
  SPEAKER_01: are experimental.
• 38:51
  SPEAKER_01: Offsetting that with Japan, which has a phenomenal deep tech sector.
• 38:55
  SPEAKER_01: So there's some phenomenal innovations coming out of there.
• 38:59
  corporate culture is not well known for being flexible or
• 39:04
  fast-moving so there's a
• 39:05
  bit of a bureaucracy challenge that could end up happening
• 39:08
  there but actually they
• 39:09
  do tons of pilots like it's it's not it's not well publicized
• 39:13
  so actually there's
• 39:14
  SPEAKER_01: there's a lot of interesting stuff happening in that market as well
• 39:18
  SPEAKER_00: And they're just quietly building.
• 39:20
  SPEAKER_00: And then on the other end of Asia as well,
• 39:23
  SPEAKER_00: China is just,
• 39:24
  SPEAKER_00: well,
• 39:25
  China as usual operating at its MO is that it operates in a
• 39:30
  completely different
• 39:31
  SPEAKER_00: universe.
• 39:31
  SPEAKER_00: National AI initiatives,
• 39:33
  SPEAKER_00: massive investments,
• 39:34
  SPEAKER_00: companies like Baidu or Alibaba embedding AI into everything.
• 39:38
  SPEAKER_00: It's like
• 39:39
  SPEAKER_00: watching the future happen in fast forward.
• 39:41
  SPEAKER_00: Like you remember how COVID,
• 39:43
  SPEAKER_00: when COVID happened,
• 39:44
  SPEAKER_00: China built like a hospital in like three to five days.
• 39:48
  SPEAKER_00: I think that's a skill that you're looking at as well.
• 39:50
  SPEAKER_00: And China is a different species compared to, you know, to everyone else.
• 39:56
  It's like how in our last episode when Chi Chi said how the
• 39:59
  experience in Huawei is
• 40:00
  SPEAKER_00: just so different compared to the other companies as well.
• 40:04
  SPEAKER_00: So again, that's another level of AI adoption, I guess.
• 40:09
  SPEAKER_01: Yeah, completely, completely different.
• 40:11
  SPEAKER_01: Yeah, it's going into the future.
• 40:15
  SPEAKER_00: Yeah.
• 40:15
  SPEAKER_00: And OK, so time for some tough love.
• 40:17
  You know,
• 40:18
  if you're building an AI startup right now,
• 40:20
  what should you focus on?
• 40:21
  SPEAKER_01: I mean,
• 40:23
  when I look at it and kind of marrying this up with my
• 40:25
  takeaways from the MIT
• 40:27
  SPEAKER_01: report as well.
• 40:28
  SPEAKER_01: Yeah.
• 40:31
  SPEAKER_01: We really need to boil it down to specificity.
• 40:35
  SPEAKER_01: You need to be able to focus on what are you really trying to accomplish?
• 40:42
  We need that sort of narrow focus because the challenge in my
• 40:48
  mind is that a lot of
• 40:50
  SPEAKER_01: things out there are too broad and they're not able to deliver business.
• 40:54
  They're trying to do too many things at once and it's not
• 40:57
  delivering on the ROI.
• 40:58
  SPEAKER_01: So if you're building right now, get specific and narrow down your focus.
• 41:04
  SPEAKER_01: The second aspect is looking at the stickiness.
• 41:07
  This is the habit formation so that you don't want somebody
• 41:11
  transitioning off of
• 41:12
  SPEAKER_01: whatever you build,
• 41:13
  SPEAKER_01: going over to some generic tool because they're accustomed to doing it.
• 41:17
  You want to be able to get that habit formation and the habit
• 41:20
  formation is going to
• 41:21
  SPEAKER_01: happen if you have intuitive design.
• 41:24
  And so you need to think more in regards to not just like the
• 41:28
  functional aspect of
• 41:29
  SPEAKER_01: it,
• 41:30
  but you need to understand the form factor of how it fits in
• 41:34
  with the workflows of
• 41:35
  SPEAKER_01: the individual,
• 41:36
  SPEAKER_01: the behavior of the individual.
• 41:38
  the knowledge and intuitive understanding of the individual
• 41:41
  that is going to be using it.
• 41:43
  SPEAKER_01: And then lastly,
• 41:43
  SPEAKER_01: which is somewhat related to the specificity,
• 41:46
  we need to have measurable value for what you're actually
• 41:50
  trying to accomplish.
• 41:52
  We need to be able to draw a straight line or at least a
• 41:55
  relatively straight line
• 41:57
  SPEAKER_01: from if you implement this,
• 42:00
  SPEAKER_01: it will lead to this result.
• 42:02
  SPEAKER_01: And that result either needs to be saving money, lowering costs,
• 42:07
  SPEAKER_01: Or you need to be increasing revenue.
• 42:10
  And ideally,
• 42:11
  if you can do both of those at the same time,
• 42:13
  that's the measurable value.
• 42:15
  SPEAKER_01: I increased your revenue while decreasing your cost.
• 42:18
  But a lot of the things,
• 42:19
  as I said,
• 42:21
  maybe perhaps non-sexy that can be big ROI drivers,
• 42:25
  SPEAKER_01: are those back office functions.
• 42:27
  SPEAKER_01: They're the operations teams at the different departments of operations.
• 42:32
  SPEAKER_01: If you can end up shaving off percentage points of operations budget,
• 42:39
  SPEAKER_01: start being able to replace all of that,
• 42:41
  SPEAKER_01: there's massive ROI.
• 42:43
  SPEAKER_01: So specificity, stickiness, and measurable value.
• 42:47
  SPEAKER_00: And that's really great input.
• 42:49
  And what about VC listeners and how do they evaluate AI
• 42:53
  startups differently now?
• 43:00
  SPEAKER_01: Yeah,
• 43:01
  that's a challenge because like in the conversations that I've
• 43:03
  had with people,
• 43:04
  I haven't really seen people solidifying upon one single sort
• 43:08
  of perspective or
• 43:09
  SPEAKER_01: view.
• 43:10
  You've got a lot of folks out there that are just kind of like
• 43:12
  sitting on the
• 43:13
  sidelines waiting and seeing because they've seen kind of the
• 43:16
  flood of money that's
• 43:17
  going into it and they're going to be caught in another
• 43:21
  momentum chasing hype
• 43:22
  SPEAKER_01: bubble,
• 43:24
  SPEAKER_01: which may be true.
• 43:25
  And then you got folks out there that are a little bit more
• 43:28
  gung-ho and wanting to
• 43:30
  SPEAKER_01: kind of spray and spray and put their vets around.
• 43:32
  And they're running events and throwing some money out there,
• 43:36
  this,
• 43:36
  that,
• 43:36
  and the other.
• 43:37
  SPEAKER_01: And then you got the other folks that fall more into my category,
• 43:41
  SPEAKER_01: which are,
• 43:43
  SPEAKER_01: you know,
• 43:44
  SPEAKER_01: maybe not necessarily AI as a product,
• 43:46
  but AI enabled products that are delivering on measurable value
• 43:51
  that have real use
• 43:53
  SPEAKER_01: cases,
• 43:53
  SPEAKER_01: real workflow integrations,
• 43:55
  SPEAKER_01: and that have more deliverability in what they're proposing to companies.
• 44:01
  And they have in their product roadmap this understanding of
• 44:06
  where they believe the
• 44:06
  SPEAKER_01: technology is going and building for not the tech that we have today,
• 44:10
  SPEAKER_01: but for the tech that we'll have tomorrow.
• 44:14
  SPEAKER_00: So much more intentional in some ways to somewhat appearing.
• 44:19
  Connecting dots between what is needed now while still building
• 44:22
  for tomorrow.
• 44:23
  SPEAKER_00: Wow, navigating in between liners too.
• 44:28
  SPEAKER_00: Like being Superman,
• 44:29
  SPEAKER_00: like,
• 44:29
  SPEAKER_00: you know,
• 44:29
  SPEAKER_00: having the invisible hand and mindless touch,
• 44:31
  SPEAKER_00: I guess,
• 44:32
  SPEAKER_00: to the next extent.
• 44:33
  SPEAKER_01: Exactly, exactly.
• 44:35
  SPEAKER_00: What about the regional play?
• 44:37
  SPEAKER_00: Like, should Southeast Asian startups focus locally first?
• 44:41
  SPEAKER_00: What's that like?
• 44:43
  SPEAKER_01: Honestly, I think we can focus not only regionally, but globally.
• 44:48
  SPEAKER_01: It's always inherently been very challenging to hit the US market.
• 44:52
  SPEAKER_01: It's a huge market,
• 44:53
  but it's inherently difficult just because it's another level
• 44:56
  of competitiveness
• 44:57
  because there's lots and lots of other startups that are all
• 45:00
  trying to go after
• 45:00
  SPEAKER_01: those same dollars.
• 45:03
  SPEAKER_01: But it is a big market.
• 45:05
  And the positive for startups is the barrier for the price of
• 45:12
  admission is
• 45:13
  significantly lower because we all have access to a lot of the
• 45:18
  same code writing
• 45:19
  SPEAKER_01: tools,
• 45:19
  SPEAKER_01: a lot of the same utilities that we can now build faster,
• 45:22
  SPEAKER_01: stronger,
• 45:23
  SPEAKER_01: better.
• 45:24
  SPEAKER_01: And so there is...
• 45:26
  SPEAKER_01: More so than I believe before,
• 45:28
  opportunity in order to build global products from the cost
• 45:32
  base of a place like
• 45:34
  SPEAKER_01: Southeast Asia.
• 45:36
  And there's a number of examples of companies that are out
• 45:38
  there doing it.
• 45:39
  SPEAKER_01: We've got some in our portfolio.
• 45:41
  SPEAKER_01: It's not exclusive to us.
• 45:43
  SPEAKER_01: There's plenty out there.
• 45:44
  But there is a big opportunity for companies to be building
• 45:47
  globally from here.
• 45:49
  SPEAKER_00: And speaking of which,
• 45:50
  SPEAKER_00: you know,
• 45:50
  SPEAKER_00: even on this part as well,
• 45:52
  on the other end of the causeway is that there's so many AI
• 45:56
  events have popping up
• 45:58
  so much like mushrooms nowadays that it's even harder to keep
• 46:01
  track through.
• 46:02
  So I guess you can see the ripple effect about how on this end
• 46:06
  of the world,
• 46:07
  SPEAKER_00: at least Southeast Asia is really taking this AI wave like
• 46:11
  by its horns and you know and so the next question backs you
• 46:15
  know looking ahead the
• 46:16
  future of this divide where does this leave us really is the
• 46:20
  gen i divide i don't
• 46:22
  think it's permanent is it because everything is temporary i
• 46:24
  mean you have bubbles
• 46:26
  and then the bubble burst and then it's like so what's your
• 46:28
  view on this gen i
• 46:30
  SPEAKER_00: divide at this point
• 46:32
  SPEAKER_01: Yeah,
• 46:32
  SPEAKER_01: so I mean,
• 46:34
  the gen AI divide at this point is we need to keep it in the
• 46:37
  context that we're
• 46:39
  SPEAKER_01: still early days.
• 46:42
  SPEAKER_01: Even though it may end up seeming like a lot has ended up taking place.
• 46:48
  SPEAKER_01: We're still quite early days in building this out.
• 46:53
  SPEAKER_01: We just need to keep into context that we have this divide presently.
• 46:58
  SPEAKER_01: We need to understand it,
• 47:00
  SPEAKER_01: build towards it,
• 47:01
  resolve it and keep on moving forward because innovation is
• 47:05
  going to stumble a few
• 47:06
  SPEAKER_01: times.
• 47:06
  We'll end up finding these divides,
• 47:08
  but we just need to figure out a way to cross the chasm.
• 47:11
  So I hope the takeaway and looking at it was basically giving
• 47:17
  feedback in to what
• 47:19
  SPEAKER_01: we need to do better.
• 47:22
  SPEAKER_00: I think as well is that what we talked about earlier is also founders,
• 47:26
  SPEAKER_00: like the message is pretty clear.
• 47:28
  SPEAKER_00: You can't just write the hype wave anymore.
• 47:30
  Like it's better to get more specialized because the market's
• 47:33
  getting more
• 47:34
  sophisticated about distinguishing between AI that demos well
• 47:38
  and the AI that
• 47:39
  SPEAKER_00: really actually delivers value.
• 47:42
  And it's complex because some of the biggest ROI wins come from
• 47:47
  automation,
• 47:47
  SPEAKER_00: like fewer customer service agents,
• 47:49
  SPEAKER_00: less manual contract review.
• 47:50
  SPEAKER_00: That's real impact.
• 47:52
  SPEAKER_00: But it comes to social costs.
• 47:54
  And the winners might be companies that use AI to augment
• 47:57
  rather than replace,
• 47:59
  SPEAKER_00: make existing employees more productive rather than eliminate positions,
• 48:03
  SPEAKER_00: which we've covered earlier.
• 48:05
  SPEAKER_00: I guess that's more...
• 48:07
  SPEAKER_00: I think that's more sustainable long-term.
• 48:09
  SPEAKER_00: Plus, the augmentation model often creates better user adoption.
• 48:13
  People are more willing to embrace tools that make their jobs
• 48:16
  easier rather than obsolete.
• 48:19
  SPEAKER_00: So I think it's really,
• 48:21
  the way I see it is trying to find that fine line that enables
• 48:26
  people to embrace AI
• 48:27
  SPEAKER_00: even more effectively.
• 48:31
  SPEAKER_00: For example, like how iPhone came into play.
• 48:33
  SPEAKER_00: iPhone is the phone then and now is because of its intuition.
• 48:37
  SPEAKER_00: So I think we need more AI tools that are more intuitive,
• 48:40
  SPEAKER_00: that are able to pick up users once and giving them what...
• 48:47
  SPEAKER_00: they didn't think they need to quote Steve Jobs.
• 48:50
  So I think that would enable the AI adoption,
• 48:53
  even quicker than the AI adoption.
• 48:55
  SPEAKER_00: Okay, so let's wrap this up.
• 48:57
  SPEAKER_00: If someone's driving home from this episode, what should they remember?
• 49:03
  SPEAKER_01: So I would say first and foremost, the AI revolution is real.
• 49:08
  SPEAKER_01: It's going to be messier and slower than the hype may suggest.
• 49:13
  SPEAKER_01: It's not going to be perfect, but it is here and it's not going anywhere.
• 49:19
  So companies are still figuring out,
• 49:21
  which means that there's still time to get it right.
• 49:24
  SPEAKER_01: And we're early innings, so we haven't even really scratched the surface.
• 49:27
  SPEAKER_01: We're not done yet.
• 49:28
  SPEAKER_01: So if anyone's like tinkering with an idea, now is the time.
• 49:33
  I think that's the best part about being in this sector is that
• 49:36
  everything is
• 49:37
  SPEAKER_00: always moving and we are always,
• 49:38
  it feels like we are always learning and it feels like we are
• 49:41
  always scratching the
• 49:41
  SPEAKER_00: surface.
• 49:42
  SPEAKER_00: And I think, you know, the second one is focus on value over flash.
• 49:46
  SPEAKER_00: Again, it doesn't apply to this business.
• 49:48
  SPEAKER_00: It applies to people as well.
• 49:50
  SPEAKER_00: You want someone who has more substance.
• 49:53
  rather than just being flashy or just on a surface level
• 49:56
  because you can't get deep with them.
• 49:58
  And the next winners will be those who solve real problems with
• 50:02
  measurable impact.
• 50:03
  SPEAKER_00: Again, substance, not those with the coolest demos.
• 50:06
  SPEAKER_01: Exactly, exactly.
• 50:08
  SPEAKER_01: And then we need to overcome the learning gaps,
• 50:10
  SPEAKER_01: which is going to end up being a big opportunity.
• 50:14
  And one aspect of it is just making the tools smarter to where
• 50:18
  they're able to
• 50:19
  SPEAKER_01: absorb the knowledge base of organizations.
• 50:21
  But it's also the learning gap on the adoption from the
• 50:23
  individual users of
• 50:25
  being able to get it integrated into their workflows and
• 50:28
  overcoming that kind of alteration.
• 50:30
  So there's the learning curve on both the technology side
• 50:34
  that's being implemented
• 50:35
  SPEAKER_01: as well as on the user side.
• 50:38
  And for the C audience specifically,
• 50:41
  you've got advantages in speed and agility.
• 50:43
  The smaller you are,
• 50:44
  the easier it is for you to move and navigate,
• 50:46
  like use them.
• 50:47
  While bigger companies are stuck in community hell,
• 50:50
  you can be deploying and learning as you go.
• 50:56
  SPEAKER_01: Exactly.
• 50:57
  SPEAKER_01: This Gen AI divide isn't permanent.
• 50:59
  SPEAKER_01: It's a natural part of technology adoption.
• 51:02
  SPEAKER_01: It's just the way that it ends up going.
• 51:04
  SPEAKER_01: We've got early chaos.
• 51:05
  SPEAKER_01: We'll get a little bit of consolidation.
• 51:07
  SPEAKER_01: We'll see what works.
• 51:08
  SPEAKER_01: We see what doesn't.
• 51:09
  And then we just keep on moving from chaos phase into the
• 51:13
  development phase to see
• 51:14
  SPEAKER_01: new companies emerge out of it.
• 51:17
  SPEAKER_01: And ultimately, we'll start seeing real tangible ROI.
• 51:21
  SPEAKER_01: Next iteration of this MIT report may end up being flipped,
• 51:24
  SPEAKER_01: where it says only 5% didn't get ROI and 95% did.
• 51:30
  SPEAKER_00: I look forward to that.
• 51:30
  SPEAKER_00: I think at this point,
• 51:31
  SPEAKER_00: we could do another episode on MIT reports,
• 51:35
  because there's one more that comes up about how ChatGPT kind
• 51:38
  of alters our brain
• 51:39
  SPEAKER_00: functions.
• 51:40
  SPEAKER_00: Could be another one worth talking about.
• 51:43
  SPEAKER_01: That's a good one.
• 51:44
  SPEAKER_01: AI psychosis.
• 51:46
  SPEAKER_00: Yeah, I mean, it could be a series, actually.
• 51:48
  SPEAKER_00: Think about it.
• 51:50
  SPEAKER_01: We'll put that on the notes for what's our next rant going to be about.
• 51:55
  SPEAKER_00: Yeah, I think that's quite a bit of... We have a list now at this point.
• 51:59
  SPEAKER_00: There's one more from each series that we haven't got round to,
• 52:01
  SPEAKER_00: but we will eventually get to it.
• 52:04
  Coming back to this,
• 52:05
  which means that this might actually be the best time to build
• 52:07
  in AI.
• 52:08
  SPEAKER_00: Less noise,
• 52:08
  SPEAKER_00: clearer signal,
• 52:09
  SPEAKER_00: more focus on fundamentals,
• 52:11
  SPEAKER_00: just like the stock market,
• 52:12
  SPEAKER_00: or at least the equity stock market.
• 52:14
  SPEAKER_01: Exactly.
• 52:15
  SPEAKER_01: Exactly.
• 52:16
  SPEAKER_01: Easy money is gone, but the real money is just getting started.
• 52:19
  SPEAKER_00: And that's a wrap on this episode of the Sea of Startups.
• 52:22
  SPEAKER_00: As always, we want to hear from you.
• 52:25
  SPEAKER_00: Are you seeing the Gen I divide in your organization?
• 52:30
  Or are you part of the 5% success stories or still stuck in
• 52:34
  that pilot purgatory?
• 52:35
  SPEAKER_00: Yeah.
• 52:39
  SPEAKER_01: All right.
• 52:39
  SPEAKER_01: So please do hit us up on social media or email us.
• 52:42
  SPEAKER_01: And if you're building something in the AI space,
• 52:45
  SPEAKER_01: we'd love to hear about it,
• 52:46
  especially if you think you cracked the code on the learning
• 52:50
  gap or any of the
• 52:51
  SPEAKER_01: problems that we've been talking about in today's episode.
• 52:55
  SPEAKER_00: There's so many gaps.
• 52:56
  SPEAKER_00: I hope at least you cracked one of them.
• 52:57
  At least one solved rather than the big enterprise
• 53:00
  transformation kind of thing.
• 53:02
  SPEAKER_00: Anyway, subscribe wherever you get your podcasts.
• 53:04
  SPEAKER_00: Leave us a review if this was helpful.
• 53:07
  SPEAKER_00: And we'll see you next when we dive into the next episode.
• 53:13
  SPEAKER_00: And again, something worth putting out there.
• 53:17
  SPEAKER_00: All these are just strictly our opinions and nothing else.
• 53:20
  SPEAKER_00: Just putting out so that people,
• 53:23
  SPEAKER_00: like the companies that we mentioned,
• 53:24
  SPEAKER_00: don't come at us just for safety reasons for us.
• 53:28
  SPEAKER_00: Yeah.
• 53:29
  SPEAKER_01: Yeah, this is all commentary.
• 53:32
  SPEAKER_01: No, no, no, no intention of causing harm.
• 53:36
  SPEAKER_01: Anyways,
• 53:37
  SPEAKER_01: until next time,
• 53:38
  SPEAKER_01: keep building,
• 53:38
  SPEAKER_01: keep learning,
• 53:39
  SPEAKER_01: and maybe,
• 53:40
  SPEAKER_01: just maybe,
• 53:41
  keep your AI experiments away from LinkedIn until they're
• 53:43
  actually working.
• 53:45
  SPEAKER_00: Yeah, savage, but fair.
• 53:48
  SPEAKER_00: See you next time, guys.
• 53:49
  SPEAKER_00: Peace out.