1
00:00:01,520 --> 00:00:03,960
Ladies and gentlemen, welcome to
the Business Brew. 

2
00:00:04,000 --> 00:00:07,440
I am your host, Bill Brewster. 
This episode features Andrew 

3
00:00:07,440 --> 00:00:09,800
Friedman of Hedgeye Risk 
Management. 

4
00:00:10,440 --> 00:00:13,960
Andrew is an equity research 
analyst at Hedgeye. 

5
00:00:13,960 --> 00:00:18,040
He covers TMT and healthcare and
on the LinkedIn machine. 

6
00:00:18,040 --> 00:00:20,560
He describes himself as 
intellectually curious and 

7
00:00:20,560 --> 00:00:24,040
honest, with a passion for 
financial markets and a 

8
00:00:24,040 --> 00:00:27,760
competitive drive to outperform.
If you've listened to the show 

9
00:00:27,760 --> 00:00:30,280
since inception, you will know 
that Andrew's been on the show 

10
00:00:30,280 --> 00:00:34,520
before. 
We have in the past had nice 

11
00:00:34,520 --> 00:00:39,640
discussions about cable and I 
think we talked Matt at at a 

12
00:00:39,640 --> 00:00:44,520
time. 
Either way, Andrew was tweeting 

13
00:00:44,520 --> 00:00:50,680
a little bit about his 
perception of people not really 

14
00:00:50,680 --> 00:00:58,280
appreciating what AI was going 
to do to either the world or 

15
00:00:58,280 --> 00:01:02,000
markets or whatever. 
And I saw a tweet that he sent 

16
00:01:02,000 --> 00:01:04,080
and I said, man, you want to 
come on and have a chat. 

17
00:01:04,640 --> 00:01:07,120
So I quite like this 
conversation. 

18
00:01:07,120 --> 00:01:12,080
I would say that this is more in
the spirit of kind of how the 

19
00:01:12,080 --> 00:01:15,640
show was born. 
And it's a it's a more laid back

20
00:01:15,640 --> 00:01:18,000
discussion. 
And by discussion, I mean Andrew

21
00:01:18,320 --> 00:01:21,920
teaches me a lot in this episode
and I do a lot of listening, but

22
00:01:22,160 --> 00:01:25,120
he talks about how he's 
integrated AI in his life. 

23
00:01:25,160 --> 00:01:29,800
And I have a lot, a lot to 
improve on in the way that I use

24
00:01:29,800 --> 00:01:31,520
it. 
But it's certainly become a real

25
00:01:31,520 --> 00:01:35,920
part of my life. 
And Mary Meeker has a deck that 

26
00:01:35,920 --> 00:01:38,240
I will link to the in the show 
notes. 

27
00:01:38,800 --> 00:01:40,960
I encourage you all to flip 
through it. 

28
00:01:41,040 --> 00:01:44,480
And I suspect if you're a 
listener of this show, it's you 

29
00:01:44,480 --> 00:01:47,760
probably are playing around with
some of these LLMS at a minimum.

30
00:01:47,760 --> 00:01:51,400
So I hope you all enjoy the 
show. 

31
00:01:51,920 --> 00:01:54,960
I enjoyed talking to Andrew 
again and he may come back and 

32
00:01:54,960 --> 00:01:59,200
do more of a single name 
discussion in the not too 

33
00:01:59,200 --> 00:02:02,440
distant future, but this one is 
more one for you all to chew on 

34
00:02:02,440 --> 00:02:05,240
and think about. 
I, I kind of think it's Andrew 

35
00:02:05,600 --> 00:02:08,080
Sharon best practices. 
So I hope you all enjoy. 

36
00:02:08,080 --> 00:02:35,080
I enjoyed recording it. 
Ladies and gentlemen, thrilled 

37
00:02:35,080 --> 00:02:39,400
to be joined by Andrew Friedman 
for I think Part 3 here, also 

38
00:02:39,400 --> 00:02:41,680
known as Free Bird. 
How's everything going, man? 

39
00:02:42,040 --> 00:02:43,600
It's going well. 
Yeah. 

40
00:02:44,080 --> 00:02:44,960
Yeah. 
This is great. 

41
00:02:45,000 --> 00:02:47,200
It's been a, it's been a little 
while, but you know, it's good 

42
00:02:47,200 --> 00:02:51,400
to be back. 
It has been a while I I prepped 

43
00:02:51,400 --> 00:02:54,480
for this interview, or 
discussion, should I say, by 

44
00:02:54,480 --> 00:02:56,080
using Grok. 
There you go. 

45
00:02:56,120 --> 00:02:59,920
So So Grok has summarized some 
of your AI thoughts. 

46
00:03:00,320 --> 00:03:03,760
Boy, let's hear it. 
Well, hang on, I got to bring it

47
00:03:03,760 --> 00:03:06,160
up. 
But the first, so first of all, 

48
00:03:06,160 --> 00:03:09,680
to give people some context, 
Andrew has been tweeting a 

49
00:03:09,680 --> 00:03:14,640
little bit about people 
underestimating what AI is going

50
00:03:14,640 --> 00:03:16,920
to do. 
Is that a fair characterization,

51
00:03:16,920 --> 00:03:18,520
Andrew? 
I would say so. 

52
00:03:19,760 --> 00:03:21,680
All right. 
And one of the things that Grok 

53
00:03:21,680 --> 00:03:25,520
has highlighted is your 
prediction of unemployment. 

54
00:03:26,040 --> 00:03:30,040
So I guess, I guess before we 
really get into it, you know, 

55
00:03:30,040 --> 00:03:34,040
just kind of where's your head 
at and and why is this so much 

56
00:03:34,040 --> 00:03:36,000
on your on your mind right now? 
Yeah. 

57
00:03:36,000 --> 00:03:39,400
So I guess I actually that's 
really interesting and we can 

58
00:03:39,400 --> 00:03:42,120
get into a little bit later. 
But the fact that Grok picked 

59
00:03:42,120 --> 00:03:44,800
that up is also kind of also 
addresses a little bit of the 

60
00:03:44,800 --> 00:03:47,720
risks associated with using 
certain AI models, right? 

61
00:03:48,120 --> 00:03:51,960
Because you know, it will pick 
up on the content that it's 

62
00:03:51,960 --> 00:03:53,640
trained on, whether it has 
access to, right? 

63
00:03:53,640 --> 00:03:57,720
Which in this case is Twitter or
or X, which is a great source of

64
00:03:57,720 --> 00:04:01,400
information, of course, real 
time information, but it's also 

65
00:04:01,400 --> 00:04:04,360
social media. 
And there's a risk because some 

66
00:04:04,360 --> 00:04:06,200
of those things can be taken out
of context. 

67
00:04:07,600 --> 00:04:09,240
Oh, well, unemployment is a 
risk. 

68
00:04:09,240 --> 00:04:12,520
And I put out this kind of 
dystopian tweet the other day 

69
00:04:12,520 --> 00:04:15,120
around this future state of the 
world and what it could look 

70
00:04:15,120 --> 00:04:17,560
like by 20-30. 
And of course, anything could 

71
00:04:17,560 --> 00:04:24,200
happen, you know, if it's 
sarcasm or any type of I guess 

72
00:04:24,840 --> 00:04:29,520
tivity can be misconstrued as 
fact, but so just kind of say 

73
00:04:29,520 --> 00:04:32,280
that. 
But look, I, I would say I have 

74
00:04:32,320 --> 00:04:34,360
spent. 
Let me, let me start over. 

75
00:04:34,360 --> 00:04:39,080
I told myself, right, That I 
made a promise to myself that 

76
00:04:39,400 --> 00:04:43,200
when there is a obvious 
technology, technological 

77
00:04:43,200 --> 00:04:45,920
disruption, new technology that 
comes out, that I wasn't going 

78
00:04:45,920 --> 00:04:51,080
to just sit idly, buy and be a 
passive observer, right? 

79
00:04:51,600 --> 00:04:54,160
Yes. 
And obviously everyone likes to 

80
00:04:54,520 --> 00:04:57,600
financial podcast, right? 
We all want to try to make money

81
00:04:57,600 --> 00:05:03,640
with what we do in our time. 
But you know, I, I was young 

82
00:05:03,640 --> 00:05:05,560
when the Internet first came 
out, right? 

83
00:05:05,640 --> 00:05:11,360
I was probably 10 or 12 when 
things kind of really started to

84
00:05:11,360 --> 00:05:13,800
took off and I was like building
websites then and doing all 

85
00:05:13,800 --> 00:05:17,520
these different things, but 
obviously not in kind of like an

86
00:05:17,520 --> 00:05:20,560
age or place where you can 
actually really, you know, do 

87
00:05:20,560 --> 00:05:22,240
anything with that and 
capitalize on the way. 

88
00:05:23,000 --> 00:05:26,000
You know, we're kind of at a 
similar crossroads today. 

89
00:05:26,120 --> 00:05:29,120
And you know me, Bill, like I'm 
kind of cynical, right? 

90
00:05:29,320 --> 00:05:32,080
I'm not like if you put me in 
the cynical camp versus the 

91
00:05:32,080 --> 00:05:34,880
Galaxy brain camp, like I'm 
definitely not in the Galaxy 

92
00:05:34,880 --> 00:05:37,600
brain camp. 
I think of you as data-driven. 

93
00:05:37,600 --> 00:05:39,920
I don't think of you as cynical.
I appreciate that. 

94
00:05:39,920 --> 00:05:42,520
I appreciate that. 
But anyway, the point is that 

95
00:05:43,520 --> 00:05:48,040
this is really an important part
time in our history and I've 

96
00:05:48,040 --> 00:05:51,280
spent the last year and a half 
really just going down the 

97
00:05:51,280 --> 00:05:54,080
rabbit hole in a really big way 
on this and trying to understand

98
00:05:54,080 --> 00:05:55,720
how it works. 
Partly because, you know, I 

99
00:05:55,720 --> 00:06:00,320
cover Google, Meta software, all
these, you know, very large mega

100
00:06:00,320 --> 00:06:03,760
cap companies that everyone 
cares a lot about and a is and 

101
00:06:04,000 --> 00:06:06,760
the main topic. 
But what is AI, right? 

102
00:06:06,760 --> 00:06:10,280
I mean, from a analyst 
perspective, like I felt kind of

103
00:06:10,280 --> 00:06:14,920
like a fraud, like sitting there
talking about llama this and 

104
00:06:14,920 --> 00:06:18,880
grok that it without actually 
knowing fundamentally what these

105
00:06:18,880 --> 00:06:22,000
things are. 
So I went out and I bought a 

106
00:06:22,000 --> 00:06:27,600
computer and I built a computer 
with two NVIDIA GPUs ultimately 

107
00:06:27,600 --> 00:06:32,840
like spent the equivalent of 
like a Toyota Camry really on 

108
00:06:32,840 --> 00:06:34,400
hardware. 
Yeah, yeah, yeah. 

109
00:06:34,400 --> 00:06:38,200
Wow And you know, kind of just 
trying to figure this out behind

110
00:06:38,200 --> 00:06:40,920
the scenes and not that you have
to go out and spend like 30 

111
00:06:40,920 --> 00:06:44,920
grand on a rig, right to to 
figure to do this stuff like, I 

112
00:06:44,920 --> 00:06:48,040
mean, there's cloud solutions, 
all these things, but you know, 

113
00:06:48,160 --> 00:06:51,160
I, I wanted to understand how 
the hardware worked, right? 

114
00:06:51,160 --> 00:06:54,200
I wanted to go through all the 
pain points and the problems of 

115
00:06:54,200 --> 00:06:58,000
trying to host llama locally on 
your computer. 

116
00:06:58,040 --> 00:07:00,800
What limitations do you hit 
trying to train your own models,

117
00:07:00,800 --> 00:07:04,560
fine tune your own models? 
What type of limitations do you 

118
00:07:04,560 --> 00:07:08,480
experience and then hit those 
problems and then solve for 

119
00:07:08,480 --> 00:07:10,320
those problems? 
Because that's for at least me, 

120
00:07:10,320 --> 00:07:12,600
that's been the best way to 
learn is actually by doing. 

121
00:07:12,600 --> 00:07:16,320
And that way I felt that I could
actually understand what an LLM 

122
00:07:16,320 --> 00:07:19,080
is, right? 
How does it work functionally? 

123
00:07:19,200 --> 00:07:22,040
And that's been a really 
fascinating learning experience.

124
00:07:22,040 --> 00:07:26,200
And, you know, bringing it to 
kind of today and all those 

125
00:07:26,200 --> 00:07:29,880
learnings, I think that it just 
kind of comes down to a lot of a

126
00:07:29,880 --> 00:07:31,640
lot of it comes down to like 
psychology, right? 

127
00:07:31,640 --> 00:07:34,800
And I know you're always like AI
always enjoy listening to your 

128
00:07:34,800 --> 00:07:37,520
podcast because you often get 
into like philosophical 

129
00:07:37,520 --> 00:07:41,160
conversations and very like 
introspective in terms of human 

130
00:07:41,160 --> 00:07:44,400
behavior and how the mind works 
and all the biases embedded 

131
00:07:44,400 --> 00:07:47,160
there in. 
But, you know, I think it come 

132
00:07:47,160 --> 00:07:50,600
kind of like this whole concept 
of like people not expecting or 

133
00:07:50,600 --> 00:07:52,520
knowing a very smart people 
rather. 

134
00:07:52,520 --> 00:07:55,080
Was the tweet that you reached 
out on very smart people don't 

135
00:07:55,080 --> 00:07:57,520
really understand the 
implications of AI? 

136
00:07:57,520 --> 00:08:01,600
And I think it comes down to 
just the two types of people, 

137
00:08:01,760 --> 00:08:03,240
right? 
You have people that come at 

138
00:08:03,240 --> 00:08:06,080
things from a scarcity mindset 
versus an abundance mindset. 

139
00:08:07,080 --> 00:08:10,160
Yeah. 
You know, initially with any 

140
00:08:10,160 --> 00:08:15,120
technology transformation, 
people look at new tools as kind

141
00:08:15,120 --> 00:08:17,680
of replacements for existing 
tasks, right? 

142
00:08:18,680 --> 00:08:22,880
But actually what AI can 
accomplish and what I've been 

143
00:08:22,880 --> 00:08:26,640
trying to do and, and mind you, 
I actually came at this my 

144
00:08:26,640 --> 00:08:28,880
scarcity mindset and then it 
turned into an abundance 

145
00:08:28,880 --> 00:08:30,440
mindset. 
Interesting. 

146
00:08:30,640 --> 00:08:33,400
You know, you can, these tools 
are enable, like can enable 

147
00:08:33,400 --> 00:08:37,240
things that you previously 
thought you couldn't do and 

148
00:08:37,720 --> 00:08:41,960
really creating value above and 
beyond just core, you know, 

149
00:08:41,960 --> 00:08:44,480
basically replacing your, your 
daily tasks. 

150
00:08:44,480 --> 00:08:48,400
So there's a lot to talk about. 
It's, it's evolving like very 

151
00:08:48,400 --> 00:08:49,680
quickly. 
It's amazing how fast the 

152
00:08:49,680 --> 00:08:51,960
technology is moving. 
But I'll just, I'll just pause 

153
00:08:51,960 --> 00:08:55,520
there, I guess. 
So what since you were using 

154
00:08:55,520 --> 00:08:57,880
Llama, I'm curious for your 
perspective on this. 

155
00:08:58,160 --> 00:09:02,440
What happened to Llama? 
Because it was my default engine

156
00:09:02,440 --> 00:09:05,480
in perplexity. 
And now it has just been removed

157
00:09:05,480 --> 00:09:10,080
from the platform And like, I 
can't even use it if I mean, I 

158
00:09:10,080 --> 00:09:12,200
could, right? 
If I if I did it but I I go 

159
00:09:12,200 --> 00:09:16,760
through my primary I, I use 
Gemini and perplexity for the 

160
00:09:16,760 --> 00:09:19,960
most part. 
Yeah, so I I can't speak to 

161
00:09:19,960 --> 00:09:22,800
specifically what happened with 
Llamas availability and 

162
00:09:22,800 --> 00:09:25,520
Perplexity itself. 
So Perplexity is a rapper, 

163
00:09:25,640 --> 00:09:26,120
right? 
And. 

164
00:09:26,120 --> 00:09:28,680
Yeah, but it kind of disappeared
a little bit, right? 

165
00:09:28,680 --> 00:09:33,040
I mean, it's an open source code
and it was getting adoption and 

166
00:09:33,040 --> 00:09:35,400
then I've heard less about it 
recently. 

167
00:09:35,840 --> 00:09:38,880
Yeah. 
I mean, look, I think as I 

168
00:09:38,880 --> 00:09:42,800
mentioned before, like the rate 
of innovation is happening 

169
00:09:42,800 --> 00:09:46,920
pretty quickly, right? 
And you know, really things only

170
00:09:46,920 --> 00:09:50,800
took off a year ago and the open
source models had, you know, 

171
00:09:50,800 --> 00:09:53,160
the, the open source models will
always have a place. 

172
00:09:53,200 --> 00:09:57,320
But I think if you look at just 
Llama recently and the 

173
00:09:57,320 --> 00:10:00,840
competition and the rise of the 
closed models with, you know, 

174
00:10:00,840 --> 00:10:06,040
Claude and Tropic open AI Gemini
2.5 is a great example, right? 

175
00:10:06,120 --> 00:10:08,640
They're just so much better, 
right? 

176
00:10:08,880 --> 00:10:13,360
And, and they're scoring on 
benchmarks, they're leading 

177
00:10:13,360 --> 00:10:16,240
benchmarks, not that you know, 
you can't game the benchmarks. 

178
00:10:16,240 --> 00:10:18,200
You can of course game the 
benchmarks a little bit. 

179
00:10:18,200 --> 00:10:20,680
And I'm sure there's some some 
of that going on. 

180
00:10:21,200 --> 00:10:24,480
But I can tell you this, much 
like I've hosted Llama, I've 

181
00:10:24,480 --> 00:10:27,240
hosted DeepSeek, I've fine-tuned
both models. 

182
00:10:27,240 --> 00:10:31,400
I currently have like the 
premium subscriptions to all the

183
00:10:31,400 --> 00:10:36,400
different chatbots, Grok Gemini 
Ultra clawed, all these things. 

184
00:10:36,400 --> 00:10:40,840
And there is like a meaningful 
difference between the models in

185
00:10:40,840 --> 00:10:43,080
terms of what they can do. 
Each model can. 

186
00:10:43,120 --> 00:10:45,120
I would say like some models are
better at coding, right? 

187
00:10:45,120 --> 00:10:47,400
So like Claude is amazing at 
cloak at coding. 

188
00:10:48,080 --> 00:10:51,440
Gemini 2.5 has a huge context 
window and it's also really good

189
00:10:51,440 --> 00:10:52,800
at coding. 
So there's a lot of things that 

190
00:10:52,800 --> 00:10:55,320
you can do with that that you 
can't do with with clod. 

191
00:10:56,640 --> 00:11:00,840
You know, ChatGPT with their O3 
model, their thinking model. 

192
00:11:01,360 --> 00:11:06,000
It's very good at creativity 
and, you know, brainstorming 

193
00:11:06,000 --> 00:11:09,120
activities. 
And so you can kind of use each 

194
00:11:09,120 --> 00:11:13,120
one of these models as different
tasks to accomplish certain 

195
00:11:13,120 --> 00:11:15,320
outcomes. 
You could also kind of hit them 

196
00:11:15,320 --> 00:11:19,080
against each other, which I've 
done numerous times to kind of 

197
00:11:19,280 --> 00:11:22,560
enhance the outcomes and, and 
also creating different agents. 

198
00:11:22,560 --> 00:11:26,720
But for Lama specifically, look,
I think it comes down to 

199
00:11:26,720 --> 00:11:30,560
execution, you know, and we've 
seen what's happened with Meta 

200
00:11:30,560 --> 00:11:34,960
and the turnover that they've 
experienced on their AI research

201
00:11:34,960 --> 00:11:37,640
team. 
That's a it's a huge deal, in my

202
00:11:37,640 --> 00:11:40,720
opinion, also like that culture 
of no bad news that's kind of 

203
00:11:40,720 --> 00:11:45,040
emerged recently. 
But I think ultimately the 

204
00:11:45,040 --> 00:11:48,840
question becomes, you know, what
is llama? 

205
00:11:48,840 --> 00:11:50,400
Is llama better than any other 
models? 

206
00:11:50,400 --> 00:11:53,680
And why should you use llama? 
And the answer is today it's 

207
00:11:53,680 --> 00:11:55,280
it's not. 
It's actually falling behind 

208
00:11:55,280 --> 00:11:59,320
quite substantially. 
And when you're thinking about 

209
00:11:59,480 --> 00:12:02,560
working with these models at 
scale, trying to do certain 

210
00:12:02,560 --> 00:12:07,160
tasks and not just, you know, 
you know, basic summary, right 

211
00:12:07,160 --> 00:12:10,480
or translation, like, I'm, I'm 
talking like not, not things 

212
00:12:10,480 --> 00:12:12,760
that your approach from a 
scarcity mindset, but from the 

213
00:12:12,760 --> 00:12:16,880
abundance mindset, like coding 
and creating, you know, like a 

214
00:12:17,280 --> 00:12:25,880
30% or even a 10% efficiency gap
is a really big deal. 

215
00:12:26,800 --> 00:12:28,600
Think about it like, if you're 
not, and This is why Google 

216
00:12:28,600 --> 00:12:35,280
search became, you know, that's 
a 90% share because it was the 

217
00:12:35,280 --> 00:12:37,920
best search index. 
They had the best algorithm, 

218
00:12:37,920 --> 00:12:40,200
they had the best data over 
time. 

219
00:12:40,200 --> 00:12:43,680
That matters a lot because when 
you're dealing with information,

220
00:12:43,680 --> 00:12:46,080
you want access to the best 
information, the most accurate 

221
00:12:46,080 --> 00:12:50,480
information, the fastest. 
And the risk of operating and 

222
00:12:50,480 --> 00:12:53,920
incorporating bad information or
wrong answers into your workflow

223
00:12:54,240 --> 00:12:58,280
is extremely high. 
And especially with AI, because 

224
00:12:58,280 --> 00:13:02,360
it's an iterative process of 
compounds and those risks can 

225
00:13:02,360 --> 00:13:05,040
become exponential and those 
costs can grow and be quite 

226
00:13:05,040 --> 00:13:08,120
significant when scaled out 
across multiple queries. 

227
00:13:08,160 --> 00:13:10,960
Million, you know, millions and 
millions of users on the 

228
00:13:10,960 --> 00:13:13,160
enterprise, thousands of 
employees, right? 

229
00:13:13,200 --> 00:13:16,880
And so I think that's where 
Llama has kind of started to 

230
00:13:16,880 --> 00:13:18,720
struggle. 
And you can see that like they 

231
00:13:18,720 --> 00:13:21,880
were initially took over. 
Like if you look at the 

232
00:13:22,720 --> 00:13:25,000
developer adoption, which is 
always like a really, really 

233
00:13:25,040 --> 00:13:28,560
there's like one I would say 
like key metric that you can 

234
00:13:28,560 --> 00:13:31,680
look at to track the success of 
models is what are the 

235
00:13:31,800 --> 00:13:33,720
developers using? 
Yeah. 

236
00:13:34,120 --> 00:13:38,000
And you know, last year Llama 
was including myself. 

237
00:13:38,000 --> 00:13:39,920
I'm not, I wouldn't classify 
myself as a developer. 

238
00:13:39,920 --> 00:13:41,480
I'm more of a vibe coder than 
anything else. 

239
00:13:41,480 --> 00:13:45,960
But I did spend a lot of time 
with it and it was because it 

240
00:13:45,960 --> 00:13:48,680
was the best at the tool, best 
model at the time. 

241
00:13:49,120 --> 00:13:52,240
And now it's kind of an 
afterthought. 

242
00:13:52,240 --> 00:13:55,560
And I would rather spend, you 
know, thousands of dollars on a 

243
00:13:55,560 --> 00:13:59,160
closed proprietary model like 
Claude or Jeff, you know, or 

244
00:13:59,200 --> 00:14:05,000
even 2.5, then spend the same 
amount of time but $0.00 on 

245
00:14:05,000 --> 00:14:07,600
Llama because it's just not that
good. 

246
00:14:08,280 --> 00:14:11,240
Yeah, I to your point on 
perplexity as a rapper, I was 

247
00:14:11,240 --> 00:14:13,120
looking up as you were 
describing it. 

248
00:14:13,400 --> 00:14:19,440
My man Evan Tyndale set me up on
4 mini in perplexity and now I'm

249
00:14:19,440 --> 00:14:22,480
on 4/4 point O Claude, 4 point O
Sonnet, right. 

250
00:14:22,480 --> 00:14:26,440
And I dude, it is crazy what 
these things can do. 

251
00:14:26,440 --> 00:14:28,440
Like it is absolutely 
phenomenal. 

252
00:14:28,920 --> 00:14:31,280
I don't I'm not a vibe coder. 
I wish I knew. 

253
00:14:31,320 --> 00:14:32,520
I need to learn. 
I mean. 

254
00:14:32,520 --> 00:14:35,560
You can though, you can't feel. 
Don't sell yourself short, man. 

255
00:14:35,560 --> 00:14:39,360
Like it's like you can do it. 
You just have to have the you 

256
00:14:39,360 --> 00:14:44,160
know, the time and the curiosity
and it's and honestly like 

257
00:14:45,000 --> 00:14:47,080
coming at it from the learning 
mentality kind of. 

258
00:14:48,120 --> 00:14:50,200
You always usually approach 
things like I know nothing. 

259
00:14:50,920 --> 00:14:53,760
Yeah. 
And sometimes people misconstrue

260
00:14:53,760 --> 00:14:56,480
that as me not having 
conviction, whatever, you know, 

261
00:14:56,480 --> 00:14:59,360
the reality is that it's 
important to stay humble, 

262
00:14:59,360 --> 00:15:00,880
especially with new 
technologies. 

263
00:15:00,880 --> 00:15:03,480
And there's a level of 
intellectual curiosity, humility

264
00:15:03,480 --> 00:15:06,560
that you have to have if you're 
going to actually take advantage

265
00:15:06,560 --> 00:15:08,400
of AI and what it can actually 
do. 

266
00:15:08,680 --> 00:15:11,840
Because you kind of, it's kind 
of like the death of the ego, 

267
00:15:12,120 --> 00:15:15,000
right? 
Like and, and people 

268
00:15:15,000 --> 00:15:17,200
instinctually find it extremely 
threatening. 

269
00:15:17,920 --> 00:15:21,160
And how can AI replace me? 
Just like every single pass 

270
00:15:21,160 --> 00:15:23,920
technology before, right? 
We've seen it with, you know, 

271
00:15:23,920 --> 00:15:28,320
Blockbuster, we've seen it with,
you know, going all the way back

272
00:15:28,320 --> 00:15:31,960
to the printing press, you know,
and I think that you just have 

273
00:15:31,960 --> 00:15:34,200
to have an open mind. 
And, you know, I'm pretty 

274
00:15:34,200 --> 00:15:37,240
confident, Billy, that you can 
do it, you know, and if you 

275
00:15:37,240 --> 00:15:39,480
don't know how to do it, you 
just ask AI to give you a 

276
00:15:39,480 --> 00:15:42,280
specific steps on how to 
accomplish your goal and then 

277
00:15:42,280 --> 00:15:44,640
just follow those steps and and 
learn along the way. 

278
00:15:45,080 --> 00:15:49,960
Yeah, Well, interestingly, I 
have been, I've been in a rut 

279
00:15:49,960 --> 00:15:52,320
lately. 
Not like a huge rut, but a rut. 

280
00:15:52,320 --> 00:15:56,080
And I, I've sort of narrowed it 
down through working through why

281
00:15:56,080 --> 00:15:59,840
I'm in a rut is I think that 
there is a divergent between 

282
00:15:59,840 --> 00:16:04,400
sort of like my alignment of 
principles and how I want to 

283
00:16:04,400 --> 00:16:06,280
live and how I'm currently 
living. 

284
00:16:06,280 --> 00:16:10,360
And one of the, one of the 
issues that I'm, I, I think has 

285
00:16:10,360 --> 00:16:14,240
spurred this thought is I read 
some of these research reports 

286
00:16:14,240 --> 00:16:19,320
that AI is putting it like, I 
mean, if I use notebook and I 

287
00:16:19,320 --> 00:16:22,880
use Gemini deep research, I look
at the output and I'm like, I'm 

288
00:16:22,880 --> 00:16:27,640
dead right? 
Like like I can't, I'm like, I 

289
00:16:27,640 --> 00:16:33,280
can't do much better than this. 
But but I think that that a more

290
00:16:33,280 --> 00:16:37,960
empowering question is OK. 
I just basically, I mean, I 

291
00:16:37,960 --> 00:16:40,480
understand that they make 
mistakes, but interns make 

292
00:16:40,480 --> 00:16:42,800
mistakes, employees make 
mistakes I make. 

293
00:16:42,920 --> 00:16:46,080
That's exactly right. 
That's exactly how I've been. 

294
00:16:46,080 --> 00:16:48,520
Yes, 100%. 
So it's like, OK, well, how now 

295
00:16:48,520 --> 00:16:53,400
that I have these tools at my 
disposal for 200 bucks a year, 

296
00:16:54,000 --> 00:16:58,560
how am I going to make myself 
more needed or how am I going to

297
00:16:58,560 --> 00:17:00,560
use them rather than being 
fearful of them? 

298
00:17:00,560 --> 00:17:02,240
And for a little while I've been
kind of fearful. 

299
00:17:02,240 --> 00:17:04,920
I've been like, what is my role 
in the world going to be? 

300
00:17:05,480 --> 00:17:08,920
Because I mean, I don't know my,
my hit rate. 

301
00:17:09,520 --> 00:17:13,079
I mean, I was, I was following 
the news fairly closely. 

302
00:17:13,079 --> 00:17:16,920
If I'm, I'm so glad I didn't put
out podcasts over the past month

303
00:17:16,920 --> 00:17:19,079
because they would have turned 
out to be so wrong. 

304
00:17:19,079 --> 00:17:21,560
The interviews that I would have
done right. 

305
00:17:21,560 --> 00:17:25,599
Like, I mean, I, I I would have 
classically chased some macro 

306
00:17:25,599 --> 00:17:27,119
stuff. 
It probably all would have 

307
00:17:27,119 --> 00:17:29,080
turned out wrong. 
And it's just got me like 

308
00:17:29,080 --> 00:17:31,600
thinking like man, I'm I really 
got to make sure that I don't 

309
00:17:31,600 --> 00:17:34,440
just disappear in this world. 
Thank because the computers are 

310
00:17:34,440 --> 00:17:37,800
probably better. 
There's a lot of thoughts in 

311
00:17:37,800 --> 00:17:41,280
there, and I realize some of 
it's disjointed, but no. 

312
00:17:41,360 --> 00:17:43,720
Hey, that's no, I think that's 
amazing. 

313
00:17:43,720 --> 00:17:45,320
I think what you just said is 
spot on. 

314
00:17:45,320 --> 00:17:47,600
And it's I, I was, you know, 
preparing for this. 

315
00:17:47,600 --> 00:17:50,680
I was taking notes and you kind 
of just hit on a key part of it,

316
00:17:50,680 --> 00:17:55,440
which is, you know, like the and
the the psychological stages of 

317
00:17:55,440 --> 00:17:57,200
this. 
All right, like, I mean, all 

318
00:17:57,200 --> 00:18:00,200
tech innovation is you kind of 
have like you go through these 

319
00:18:00,200 --> 00:18:02,160
different phases, right? 
Like, and it's kind of the 

320
00:18:02,160 --> 00:18:03,400
grief. 
It's like the mornings. 

321
00:18:03,400 --> 00:18:07,400
It is the grief stage. 
Smile it's like AI can't 

322
00:18:07,400 --> 00:18:09,920
possibly do the type of nuance 
work that I can do. 

323
00:18:09,920 --> 00:18:12,400
And then it's the anger phase, 
right? 

324
00:18:12,400 --> 00:18:16,360
When you see other people come 
on and creating all this content

325
00:18:16,360 --> 00:18:18,400
and you're like angry. 
It's like, this is just hype. 

326
00:18:18,440 --> 00:18:22,080
You know, real analysis requires
human judgement and then, you 

327
00:18:22,080 --> 00:18:24,280
know, bargaining. 
Yeah, Yeah, I can help with some

328
00:18:24,280 --> 00:18:27,760
basic stuff, the simple stuff. 
But the complex work, like all 

329
00:18:27,760 --> 00:18:30,320
that type of analysis, like how 
to call a quarter or like all 

330
00:18:30,320 --> 00:18:32,400
this like thoughtful strategic 
analysis. 

331
00:18:32,400 --> 00:18:35,280
I, it still needs me. 
And then kind of like what you 

332
00:18:35,280 --> 00:18:38,760
were just describing now, which 
is like the depression stage, 

333
00:18:38,760 --> 00:18:43,000
which is like everything I'm 
good at is becoming worthless. 

334
00:18:43,000 --> 00:18:45,000
Like, what is my value in this 
world? 

335
00:18:45,000 --> 00:18:48,920
And it's this existential crisis
that I think everybody is going 

336
00:18:48,920 --> 00:18:51,480
to go through, Right. 
And you know, like I said 

337
00:18:51,480 --> 00:18:54,440
before, I've actually, because 
I, I've been going through this,

338
00:18:54,440 --> 00:18:55,880
right? 
I mean, and part of the reason 

339
00:18:55,960 --> 00:18:59,040
and why I, I, I kind of went on 
this journey a year and a half 

340
00:18:59,040 --> 00:19:03,440
ago was because like, I've, I've
learned enough and I've studied 

341
00:19:03,440 --> 00:19:05,720
history enough to know how this 
all ends, right? 

342
00:19:05,720 --> 00:19:07,480
And I know how human behavior 
is. 

343
00:19:07,480 --> 00:19:10,160
And I know that a lots of people
are also not going to take the 

344
00:19:10,160 --> 00:19:12,960
time to necessarily study 
history and how history has 

345
00:19:12,960 --> 00:19:17,280
worked. 
I, you know, kind of just 

346
00:19:17,280 --> 00:19:20,400
basically like immediately 
jumped to all those and try to 

347
00:19:20,400 --> 00:19:23,880
just disregard all those 
feelings that I still have and 

348
00:19:23,880 --> 00:19:26,400
just move right to the, the 
final stage, which is 

349
00:19:26,400 --> 00:19:29,360
acceptance. 
And that's basically where you 

350
00:19:29,360 --> 00:19:32,120
just need to find new ways to 
create value in this world. 

351
00:19:32,120 --> 00:19:36,440
And I think that with that, 
there's never be going to be, 

352
00:19:36,440 --> 00:19:39,320
there's always like there are, 
there are grifters out there, 

353
00:19:39,560 --> 00:19:43,160
right, Lots of them there. 
You know, hopefully and 

354
00:19:43,160 --> 00:19:46,920
hopefully with like new 
technology in this, like that 

355
00:19:46,920 --> 00:19:49,760
long tail will get cut off. 
You know, I operate, we're 

356
00:19:49,760 --> 00:19:51,120
hedge. 
I work for an independent 

357
00:19:51,120 --> 00:19:53,080
research firm. 
I like to think I'm really good 

358
00:19:53,080 --> 00:19:55,360
at my job. 
I get my fares, I get stuff 

359
00:19:55,360 --> 00:19:57,640
wrong just like everybody else, 
But I think I'm, I'm pretty good

360
00:19:57,640 --> 00:20:00,120
at what I do. 
But I also know that the sell 

361
00:20:00,120 --> 00:20:05,760
side has has just kind of become
increasingly less valuable over 

362
00:20:05,760 --> 00:20:08,240
time. 
And there's a lot of shops out 

363
00:20:08,240 --> 00:20:11,640
there, but there's also a lot of
long tail analysts that just 

364
00:20:11,640 --> 00:20:15,480
wake up everyday. 
And they're kind of glorified 

365
00:20:15,480 --> 00:20:18,760
administrators just working on 
connecting investors with 

366
00:20:18,760 --> 00:20:22,320
management, you know, doing 
NDRS, those types of things. 

367
00:20:22,920 --> 00:20:26,240
You know, they're writing 
earnings recaps and preview 

368
00:20:26,240 --> 00:20:27,440
notes. 
They're just regurgitating 

369
00:20:27,440 --> 00:20:32,080
management very, very little, 
like low value add, not a ton of

370
00:20:32,080 --> 00:20:36,280
intellectual rigor and, and 
integrity in their work. 

371
00:20:37,000 --> 00:20:40,120
And like, that's an example 
where AI, as far as the written 

372
00:20:40,120 --> 00:20:42,680
research goes, like, you know, 
for those types of people, 

373
00:20:42,680 --> 00:20:44,280
they're in a lot of trouble, 
right? 

374
00:20:44,360 --> 00:20:47,960
Because AI can come in and 
through that kind of scarcity 

375
00:20:47,960 --> 00:20:50,400
mindset, just come in and 
replace what they are because 

376
00:20:50,400 --> 00:20:52,520
they're already operating from a
scarcity mindset, right? 

377
00:20:52,520 --> 00:20:56,680
They're just trying to keep 
what's around them, you know, 

378
00:20:56,680 --> 00:21:00,640
maintain the status quo. 
And I think it's those like, you

379
00:21:00,640 --> 00:21:02,920
know, those types of people. 
And I know that sounds 

380
00:21:02,920 --> 00:21:05,960
pejorative, but whatever. 
Well, look for like an earnings 

381
00:21:05,960 --> 00:21:10,520
recap or or an earnings preview.
To your point, I don't see why. 

382
00:21:10,960 --> 00:21:14,040
I don't see why AI can't write 
that every bit as well as he can

383
00:21:14,040 --> 00:21:15,640
and can. 
For a lot of it, yeah. 

384
00:21:15,640 --> 00:21:18,120
I mean, especially if it's, you 
know, like especially if it's 

385
00:21:18,120 --> 00:21:20,720
just kind of reading the 
transcript and going off the 

386
00:21:20,720 --> 00:21:23,200
highlights, right. 
I mean, there's certain nuances 

387
00:21:23,200 --> 00:21:26,720
related to being able to update 
the model, build your forward 

388
00:21:26,720 --> 00:21:30,320
projections, you know, dissect 
the quarter, look at things 

389
00:21:30,320 --> 00:21:32,600
critically in the sense of, 
well, this is what management 

390
00:21:32,600 --> 00:21:35,040
saying. 
What's my independent view of it

391
00:21:35,400 --> 00:21:38,360
relative to my thesis? 
Has something changed? 

392
00:21:38,600 --> 00:21:41,680
You know, management saying XYZ.
But I actually think that's, you

393
00:21:41,680 --> 00:21:43,720
know, their bias. 
I actually think it's, you know,

394
00:21:43,800 --> 00:21:45,800
ABC, you know, that type of 
thing. 

395
00:21:45,880 --> 00:21:48,200
But again, that kind of comes 
down to the more like critical 

396
00:21:48,200 --> 00:21:51,200
thinking components that AI 
today, you know, still struggles

397
00:21:51,200 --> 00:21:54,080
with because the models are 
still looking for context, 

398
00:21:54,120 --> 00:21:57,040
right? 
We don't have, you know, AGI, 

399
00:21:57,320 --> 00:21:58,960
right? 
We haven't kind of hit the 

400
00:21:58,960 --> 00:22:03,160
singularity in that sense where 
they can truly mimic human 

401
00:22:03,560 --> 00:22:07,040
thought and critical thinking, 
because the way that the LLMS 

402
00:22:07,040 --> 00:22:09,760
work is they're stringing, you 
know, text together, right? 

403
00:22:09,760 --> 00:22:12,120
They're identifying patterns and
sequential. 

404
00:22:12,480 --> 00:22:14,720
They're very good at what they 
do, right? 

405
00:22:14,720 --> 00:22:17,360
But at the end, but you know, 
like, I'll give you like, here's

406
00:22:17,360 --> 00:22:22,040
another example, right? 
So remember, what was what was 

407
00:22:22,040 --> 00:22:24,800
it? 
It was it was with Doge, right? 

408
00:22:24,800 --> 00:22:27,880
And earlier in the year and 
there was a bunch of articles 

409
00:22:27,880 --> 00:22:32,120
that were coming out about how 
the government was basically 

410
00:22:32,120 --> 00:22:35,080
paying for a bunch of like New 
York Times subscriptions and you

411
00:22:35,080 --> 00:22:38,080
know, X and everyone on Twitter 
was like just freaking out about

412
00:22:38,080 --> 00:22:40,360
it. 
And you know, all this big like 

413
00:22:40,360 --> 00:22:43,080
conspiracy theory because 
everything's polarized in this 

414
00:22:43,080 --> 00:22:45,840
world and I'm not right or left,
by the way, I'm kind of like 

415
00:22:45,840 --> 00:22:47,960
just data-driven, like you said,
like in the middle. 

416
00:22:47,960 --> 00:22:52,160
I I hate all polarization pretty
much, but the but you know, I 

417
00:22:52,160 --> 00:22:54,360
was like, I was looking into it 
because I was like, well, you 

418
00:22:54,360 --> 00:22:56,840
know, you can just everyone's 
looking at USA spending that Gov

419
00:22:57,200 --> 00:22:58,640
right. 
And just like I'm looking at the

420
00:22:58,920 --> 00:23:02,080
big increase in spending that 
was associated with these 

421
00:23:02,080 --> 00:23:05,880
subscriptions. 
And I spent a ton of time on USA

422
00:23:05,880 --> 00:23:08,120
spending that Gov like looking 
at government contracts with the

423
00:23:08,120 --> 00:23:11,200
software companies in the past. 
So, and I know how complex it 

424
00:23:11,200 --> 00:23:14,120
is, you can't like if you, if 
you're just looking at it for 

425
00:23:14,120 --> 00:23:16,920
the first time, there's a lot of
traps you can fall into. 

426
00:23:16,920 --> 00:23:19,680
And so I, I looked at it and I 
strung together a lot of the 

427
00:23:19,680 --> 00:23:23,880
different award IDs and contract
IDs and it turned out that it 

428
00:23:23,880 --> 00:23:26,440
was all related to some, you 
know, library. 

429
00:23:26,560 --> 00:23:28,880
A lot of it was really to the 
Library of Congress grants, 

430
00:23:28,920 --> 00:23:32,000
right to digitize information. 
So really nothing nefarious or 

431
00:23:32,000 --> 00:23:34,560
crazy, but of course, like he 
can spin it however you want. 

432
00:23:34,680 --> 00:23:38,920
And then I asked, what did I do?
I asked ChatGPT just after web 

433
00:23:38,920 --> 00:23:41,880
search became available, because
the reality is that a year ago, 

434
00:23:41,880 --> 00:23:43,840
a lot of these models, the 
amount of access to web 

435
00:23:43,840 --> 00:23:47,240
information, real information, 
and I asked it to kind of run 

436
00:23:47,240 --> 00:23:49,440
the analysis and what it 
thought. 

437
00:23:49,440 --> 00:23:51,800
And I gave it specific sources. 
I was like, Hey, look at 

438
00:23:51,800 --> 00:23:54,760
usaspending.gov. 
Look at all this stuff and you 

439
00:23:54,760 --> 00:23:57,120
can go through because it's 
basically an agent. 

440
00:23:57,120 --> 00:23:59,040
You can see it's thinking in 
real time, right? 

441
00:23:59,040 --> 00:24:00,440
The different steps that it's 
taking. 

442
00:24:00,440 --> 00:24:03,000
And even though we actually said
that it was going to 

443
00:24:03,000 --> 00:24:06,840
usaspending.gov and it was doing
all these things, it actually 

444
00:24:06,840 --> 00:24:11,800
its conclusion was still biased 
by an article that it got from 

445
00:24:12,040 --> 00:24:17,760
auk public publication that was 
just a very basic editorial that

446
00:24:17,760 --> 00:24:23,640
reference the tweet of the 
person who caused all the. 

447
00:24:23,960 --> 00:24:25,880
Perfuffle related to that? 
Interesting. 

448
00:24:25,880 --> 00:24:30,600
And they used that as context to
anchor the entire analysis, 

449
00:24:31,680 --> 00:24:35,720
right? 
And so in that sense, AI kind of

450
00:24:35,720 --> 00:24:38,560
failed, right? 
Because it wasn't objectively 

451
00:24:38,560 --> 00:24:41,640
analyzing the information. 
It was kind of this circular 

452
00:24:41,640 --> 00:24:46,040
reference that, you know, a lot 
of people like a lot of these 

453
00:24:46,040 --> 00:24:49,480
models, a lot of these search 
tools can get caught up in. 

454
00:24:49,480 --> 00:24:51,880
And I think the danger, you 
know, the danger with that 

455
00:24:51,880 --> 00:24:56,440
right, is in the long term, 
while there's a lot of healthy 

456
00:24:56,440 --> 00:25:00,960
amount of skepticism of AII 
think we are going to be in a 

457
00:25:00,960 --> 00:25:06,200
place where especially just 
given the this natural distrust 

458
00:25:06,560 --> 00:25:10,120
like that people have now in 
society with like government 

459
00:25:10,120 --> 00:25:12,840
figureheads, authority, all 
these things where they're 

460
00:25:12,840 --> 00:25:16,760
probably going to actually be 
more willing to trust an AI to 

461
00:25:16,760 --> 00:25:21,800
give them an answer. 
Well, that's a problem. 

462
00:25:21,920 --> 00:25:25,160
And so it's just kind of goes 
back to like the same problems 

463
00:25:25,160 --> 00:25:27,680
that we've always experienced, 
whether it was social media, but

464
00:25:27,680 --> 00:25:30,560
just in a, you know, in a in a 
different. 

465
00:25:30,840 --> 00:25:32,280
Yeah. 
And it's just in a in a 

466
00:25:32,280 --> 00:25:33,840
completely different lens. 
And. 

467
00:25:34,640 --> 00:25:37,160
I've noticed that I'll ask it 
for like company overviews and 

468
00:25:37,160 --> 00:25:39,320
then I'll click through the 
citations and it's like some 

469
00:25:39,320 --> 00:25:41,840
random guy's blog, which, I 
mean, there's a lot of good work

470
00:25:41,840 --> 00:25:43,680
out there, right? 
But it's like I don't actually. 

471
00:25:43,680 --> 00:25:46,480
Want that random guy's podcast? 
This guy building? 

472
00:25:46,520 --> 00:25:48,280
That's where you know this 
transcript. 

473
00:25:48,280 --> 00:25:50,600
Like who's this guy? 
Yeah, that's right. 

474
00:25:50,600 --> 00:25:52,240
Yeah. 
I, I need to know a little bit. 

475
00:25:52,240 --> 00:25:55,160
Let's limit this to like actual 
sources, right? 

476
00:25:55,320 --> 00:25:58,680
Yeah. 
But I mean, look to our point 

477
00:25:58,680 --> 00:26:01,560
earlier, that's no different 
than the Internet, and it's no 

478
00:26:01,560 --> 00:26:06,040
different than a human that 
comes at an analysis with a bias

479
00:26:06,040 --> 00:26:10,080
coming up with a flawed premise.
It's incumbent upon us as users 

480
00:26:10,080 --> 00:26:13,520
to make sure that we try to 
mitigate those responses. 

481
00:26:14,280 --> 00:26:18,760
Yeah, no, absolutely. 
And, you know, I think this 

482
00:26:18,760 --> 00:26:22,120
natural distrust too, I mean, 
and, and this all kind of comes 

483
00:26:22,120 --> 00:26:24,840
down to like the initial part of
like the whole premise of this, 

484
00:26:25,080 --> 00:26:28,200
this podcast and conversation, 
right, is around, you know, 

485
00:26:28,200 --> 00:26:31,360
limiting adoption, right? 
And just people not necessarily 

486
00:26:31,360 --> 00:26:33,560
appreciating what's coming. 
And I, and I think a lot of it 

487
00:26:33,560 --> 00:26:36,760
has to do with like the fear of 
uncertainty, the fear of giving 

488
00:26:36,760 --> 00:26:39,880
it access to information. 
Will it train on your data, 

489
00:26:39,880 --> 00:26:42,680
right? 
I think, you know, a lot of 

490
00:26:44,120 --> 00:26:46,720
investment professionals that 
are probably listening to this, 

491
00:26:48,120 --> 00:26:51,720
you know, they all work at firms
that have really strict, the 

492
00:26:51,720 --> 00:26:54,320
regulated, they have really 
strict compliance requirements, 

493
00:26:54,320 --> 00:26:57,400
right? 
And and that limits their 

494
00:26:57,400 --> 00:27:00,600
ability to like use a lot of 
these tools the way that they 

495
00:27:00,600 --> 00:27:03,680
can because they would either 
have to take it offline. 

496
00:27:04,680 --> 00:27:07,520
I mean, look, and to be fair, 
like, you know, I've been 

497
00:27:07,520 --> 00:27:10,280
pushing this internally a hedge.
I in a really big way. 

498
00:27:10,560 --> 00:27:15,280
Like earlier in the year, I 
basically told I had a, you 

499
00:27:15,280 --> 00:27:18,680
know, meeting and I was like, 
look like this is existential to

500
00:27:18,680 --> 00:27:21,280
what we do. 
You know, we have to figure this

501
00:27:21,280 --> 00:27:24,600
out and we have to be riding the
wave and we're small, nimble 

502
00:27:24,600 --> 00:27:27,000
organization and we have to 
innovate and we have to embrace 

503
00:27:27,000 --> 00:27:32,040
this because our larger 
competitors are going to be on a

504
00:27:32,040 --> 00:27:34,000
lag. 
And even our clients, like a lot

505
00:27:34,000 --> 00:27:38,160
of them through conversations, 
they are not using AI the way 

506
00:27:38,160 --> 00:27:41,080
that they as much as they should
because their compliance 

507
00:27:41,080 --> 00:27:44,000
departments court, which of 
course are risk reverse are 

508
00:27:44,000 --> 00:27:46,760
telling them that they can't or 
they're putting shackles on them

509
00:27:46,840 --> 00:27:49,080
in in many ways. 
But that's I guess that's the 

510
00:27:49,080 --> 00:27:52,160
advantage of, you know, people 
that are independent, that can 

511
00:27:52,160 --> 00:27:53,880
stay nimble, that can do all 
these things. 

512
00:27:53,880 --> 00:27:57,520
But at the same time, you know, 
I think it, it from a research 

513
00:27:57,520 --> 00:28:00,760
standpoint, I think it's makes 
it harder for people to actually

514
00:28:00,760 --> 00:28:03,040
understand like what the future 
state of the world is. 

515
00:28:03,400 --> 00:28:06,040
Because if you're not using 
these things like day-to-day, 

516
00:28:07,520 --> 00:28:11,440
you know, I basically been using
AI 3 to 4 hours a day for the 

517
00:28:11,440 --> 00:28:15,880
last year and a half. 
It's like, I almost like, like 

518
00:28:17,000 --> 00:28:18,800
they're like, they're, it's 
great. 

519
00:28:19,840 --> 00:28:21,760
And they're, it's like an 
analyst, right? 

520
00:28:21,760 --> 00:28:23,360
And yeah. 
And you have to learn how to use

521
00:28:23,360 --> 00:28:27,120
them the right way and 
understand their their nuances 

522
00:28:27,120 --> 00:28:31,080
and you know their strengths, 
their weaknesses, again, just 

523
00:28:31,080 --> 00:28:33,960
like you would with an intern or
an analyst. 

524
00:28:34,040 --> 00:28:38,200
But the whole information 
verification thing is it is 

525
00:28:38,200 --> 00:28:41,880
really critical. 
But I don't really see it being 

526
00:28:41,880 --> 00:28:44,800
much different than kind of when
the Internet came out, right? 

527
00:28:46,760 --> 00:28:48,440
You probably remember this as 
well. 

528
00:28:49,160 --> 00:28:51,040
I mean, you're a little bit 
older than me, but we're not 

529
00:28:51,040 --> 00:28:53,880
that far apart. 
Like in the, I remember growing 

530
00:28:53,880 --> 00:28:56,000
up in elementary school and 
grade school, you know, when the

531
00:28:56,000 --> 00:28:58,680
computer labs first came out and
they were teaching how to type 

532
00:28:58,680 --> 00:29:00,240
and all the stuff, but the 
Internet was still pretty 

533
00:29:00,240 --> 00:29:01,880
nascent. 
But they would still teach you 

534
00:29:01,880 --> 00:29:05,560
how to run research by going to 
the library, the Dewey Decimal 

535
00:29:05,560 --> 00:29:08,200
system, getting the encyclopedia
out, doing all these things. 

536
00:29:08,200 --> 00:29:10,480
And you were told to be very 
wary of the Internet. 

537
00:29:10,480 --> 00:29:12,360
Don't trust Wikipedia, You have 
to verify. 

538
00:29:12,360 --> 00:29:14,800
Remember microfiche? 
That was crazy. 

539
00:29:14,880 --> 00:29:17,800
What a different time. 
Yeah, you know, verify this, 

540
00:29:17,800 --> 00:29:20,680
verify that with actual hard 
copy. 

541
00:29:22,000 --> 00:29:26,320
And that was a form of like 
parallelism where you were doing

542
00:29:26,320 --> 00:29:31,520
both for a period of time. 
And now Fast forward to 2025, 

543
00:29:31,520 --> 00:29:34,560
and it's very similar. 
It's like, OK, use AI for 

544
00:29:34,560 --> 00:29:37,240
research, check your sources, 
but instead of using 

545
00:29:37,240 --> 00:29:40,560
encyclopedia, you're actually 
using the Internet like Google, 

546
00:29:41,040 --> 00:29:44,680
you know, to, to, to anchor 
your, your truth and verify to 

547
00:29:44,680 --> 00:29:46,320
make sure that the model is not 
hallucinating. 

548
00:29:46,320 --> 00:29:50,680
So it's just very interesting 
how, you know, every generation 

549
00:29:50,680 --> 00:29:54,080
thinks they've kind of found the
real source of truth, but only 

550
00:29:54,080 --> 00:29:56,280
to inevitably abandon it for 
something better. 

551
00:29:56,280 --> 00:29:59,200
And, you know, we're not really 
getting closer to the truth. 

552
00:29:59,200 --> 00:30:02,440
We're just getting, you know, 
more efficient at feeling 

553
00:30:02,720 --> 00:30:04,320
confident about uncertainty, 
right? 

554
00:30:04,320 --> 00:30:06,280
Like that's ultimately what it 
comes down to. 

555
00:30:06,280 --> 00:30:09,320
And you know, kind of one of the
things that I've thought about, 

556
00:30:09,480 --> 00:30:12,640
and I'll pause after this, is 
just like you look at human 

557
00:30:12,640 --> 00:30:15,560
history and technology and just 
behavior. 

558
00:30:15,560 --> 00:30:18,240
And I was almost, I almost meant
to become a philosophy major. 

559
00:30:18,320 --> 00:30:20,600
Thank God I didn't. 
But it's super interesting. 

560
00:30:20,840 --> 00:30:24,240
But, you know, the only concept 
in human history is I think, you

561
00:30:24,240 --> 00:30:28,040
know, is using the thing we just
learned to trust to verify the 

562
00:30:28,040 --> 00:30:29,960
thing we're learning to trust, 
right? 

563
00:30:30,000 --> 00:30:35,560
And it's that constant cycle. 
And if you understand that it 

564
00:30:35,560 --> 00:30:39,320
actually, and you study the 
different phases from a 

565
00:30:39,320 --> 00:30:43,520
psychological perspective and 
behavioral perspective, like the

566
00:30:43,520 --> 00:30:45,240
world actually becomes a lot 
simpler. 

567
00:30:45,240 --> 00:30:48,920
Decisions become a lot easier to
make the higher degree of 

568
00:30:48,920 --> 00:30:53,960
confidence because it's all 
pattern recognition, you know? 

569
00:30:53,960 --> 00:30:57,360
Yeah, we're all individuals and 
we all make individual decisions

570
00:30:57,360 --> 00:31:01,440
and like to think we're special.
But at least on kind of a macro 

571
00:31:01,440 --> 00:31:05,000
level, you can really identify 
like very similar behavior 

572
00:31:05,000 --> 00:31:08,160
patterns. 
And because we live in the 

573
00:31:08,160 --> 00:31:11,240
Internet, the age of information
and Internet with social media, 

574
00:31:11,600 --> 00:31:14,800
a lot of those patterns become a
lot easier to recognize because 

575
00:31:14,800 --> 00:31:17,320
they all kind of surface 
themselves through trending 

576
00:31:17,320 --> 00:31:19,720
topics. 
You know, what you see on X, 

577
00:31:19,720 --> 00:31:22,360
while people will say that's not
the real world, that actually is

578
00:31:22,360 --> 00:31:26,240
kind of, you know, much closer 
to the real world than, you 

579
00:31:26,240 --> 00:31:29,200
know, psych and and people's 
thoughts and psychology that I 

580
00:31:29,200 --> 00:31:30,720
think what people actually like 
to admit. 

581
00:31:31,600 --> 00:31:33,720
Yeah, no doubt. 
So anyway, sorry I just went on 

582
00:31:33,760 --> 00:31:36,920
like no a philosophical rant, 
but that's it's, it's all I 

583
00:31:36,920 --> 00:31:39,440
think it's all really important 
in the long term. 

584
00:31:40,200 --> 00:31:42,480
Well, the thing that when you 
were talking about when the 

585
00:31:42,480 --> 00:31:45,800
Internet first came out, one of 
the I, I've been looking at Mary

586
00:31:45,800 --> 00:31:51,000
Meeker's recent AI adoption, you
know, slide deck or whatever. 

587
00:31:51,440 --> 00:31:55,000
And the thing I, I may misquote 
this, so Fact Check it for 

588
00:31:55,000 --> 00:32:00,880
yourself, but I think she said 
it took 12 years for the desktop

589
00:32:00,880 --> 00:32:05,680
Internet to get to 50% adoption 
and the projection is 3 years 

590
00:32:05,680 --> 00:32:08,320
for AI agents to get to 50% 
adoption. 

591
00:32:09,040 --> 00:32:11,280
And like when you think about 
the difference in the rate of 

592
00:32:11,280 --> 00:32:14,560
change there, I mean, that is 
crazy. 

593
00:32:15,640 --> 00:32:17,280
Now, I don't know. 
I don't know what the definition

594
00:32:17,280 --> 00:32:19,240
of adoption is, right? 
I mean, I don't know if that's 

595
00:32:19,240 --> 00:32:23,960
just people playing around with 
it or, or like it's truly being 

596
00:32:23,960 --> 00:32:27,560
used in their, you know, 
integrated in their life. 

597
00:32:27,560 --> 00:32:30,120
But yeah, it's super 
interesting, man. 

598
00:32:30,120 --> 00:32:32,920
I have not. 
I mean I I don't want to over 

599
00:32:32,920 --> 00:32:37,560
exaggerate but I have dropped my
my use of Google search. 

600
00:32:38,560 --> 00:32:41,960
Now, like I said, I use Gemini, 
so I have, I'm not off of Google

601
00:32:41,960 --> 00:32:44,840
properties, but but the 
traditional search is like 

602
00:32:44,840 --> 00:32:48,080
really really more or less gone 
out of my life unless I'm 

603
00:32:48,080 --> 00:32:51,720
looking for something specific, 
which is crazy. 

604
00:32:52,360 --> 00:32:55,280
Yeah, I mean, I think that's 
yeah. 

605
00:32:55,280 --> 00:32:58,320
So, yeah. 
So I mean, I, it's and I, and 

606
00:32:58,320 --> 00:33:00,760
just for the record, like I'm, 
you know, I'm, I'm pretty 

607
00:33:00,760 --> 00:33:05,520
bullish on Google long term. 
And that's a lot of the cut 

608
00:33:05,520 --> 00:33:08,920
like, and I've just like I've 
been on the road speaking with 

609
00:33:09,000 --> 00:33:11,440
investors and it's always like a
great, it's always just an 

610
00:33:11,440 --> 00:33:13,680
awesome experience, right? 
Because the long as you do this,

611
00:33:14,080 --> 00:33:17,800
like you just have really like 
thought provoking conversations 

612
00:33:17,800 --> 00:33:21,160
with folks and you know, these 
are people managing billions of 

613
00:33:21,160 --> 00:33:23,440
dollars of capital. 
And you know, this has been a 

614
00:33:23,440 --> 00:33:27,640
huge topic conversation. 
I guess like embedded in your 

615
00:33:27,640 --> 00:33:32,640
comment is a question about 
Google's business model, the 

616
00:33:32,640 --> 00:33:35,440
sustainability of search. 
What does it look like what 

617
00:33:35,440 --> 00:33:38,040
happens from margin perspective,
all the all these other things, 

618
00:33:38,040 --> 00:33:41,160
right? 
Yeah. 

619
00:33:41,680 --> 00:33:44,880
I think again to kind of look at
this from a historical 

620
00:33:44,880 --> 00:33:50,800
perspective and framing this 
between a, you know, like 

621
00:33:50,800 --> 00:33:53,800
framing it between like that 
using that scarcity and 

622
00:33:53,800 --> 00:34:00,320
abundance framework, right? 
Is that from a scarcity 

623
00:34:00,320 --> 00:34:03,400
perspective? 
You look at what you just 

624
00:34:03,400 --> 00:34:07,200
described and you extrapolate 
that to Google search as a whole

625
00:34:07,640 --> 00:34:12,120
and you say you would come to 
the conclusion that Google's a 0

626
00:34:13,000 --> 00:34:14,600
or you know, like search. 
Is search. 

627
00:34:14,719 --> 00:34:16,159
Search is definitely going to 
have it. 

628
00:34:16,320 --> 00:34:19,120
And I want to say Google is 0 
for those of you listening. 

629
00:34:19,440 --> 00:34:20,840
It's like a thing I do on 
Twitter. 

630
00:34:20,840 --> 00:34:23,719
It's related to Fuba. 
I just like it's, it's just 

631
00:34:23,719 --> 00:34:25,000
funny. 
I think it's funny. 

632
00:34:25,000 --> 00:34:28,040
But anyway, sorry, but the I 
don't obviously think Google is 

633
00:34:28,040 --> 00:34:29,800
a 0. 
Even if search goes to 0, Google

634
00:34:29,800 --> 00:34:31,760
is not a zero. 
But the point is it's worth 

635
00:34:31,760 --> 00:34:33,000
something. 
I'm sorry, I just got myself 

636
00:34:33,000 --> 00:34:34,480
totally. 
No, you know what I think is 

637
00:34:34,480 --> 00:34:37,080
interesting because I think 
about this, where I go with it 

638
00:34:37,080 --> 00:34:40,679
is I look at the value that 
Gemini creates and I and I 

639
00:34:40,679 --> 00:34:43,000
think, OK, well theoretically 
what could they charge me for 

640
00:34:43,000 --> 00:34:47,120
this? 
And I and I think subscription 

641
00:34:47,120 --> 00:34:50,040
revenue has a massive upside 
from this. 

642
00:34:50,239 --> 00:34:53,400
Totally. 
There's so I guess at its core. 

643
00:34:54,199 --> 00:34:57,120
I don't own it for the record, 
but I just, you know, I spend 

644
00:34:57,120 --> 00:34:59,760
time thinking about it and I, 
and I and I look at my search 

645
00:34:59,760 --> 00:35:02,520
usage and then I open up the 
financials and I look at the 

646
00:35:02,520 --> 00:35:06,920
growth in search and other and 
I'm like, how in the heck does 

647
00:35:06,920 --> 00:35:08,440
what I think the world looks 
like? 

648
00:35:08,440 --> 00:35:11,960
Like this is one of those Adam 
Robinson's things that make no 

649
00:35:11,960 --> 00:35:13,840
sense. 
He he likes to say that you can 

650
00:35:13,840 --> 00:35:17,040
make a lot of money and things 
that are very obvious and things

651
00:35:17,040 --> 00:35:20,560
that make no sense. 
Google's search results make no 

652
00:35:20,560 --> 00:35:24,520
sense to me given what I think 
is going on, but that probably 

653
00:35:24,520 --> 00:35:26,680
means that I don't have the 
right model of the world, right?

654
00:35:27,080 --> 00:35:29,120
It turns out. 
It turns out that I am not the 

655
00:35:29,120 --> 00:35:30,920
world, which is sad, but. 
True. 

656
00:35:31,000 --> 00:35:32,640
No, but it's true. 
I mean, look, I think that's a 

657
00:35:32,640 --> 00:35:34,960
very important point. 
There's everything. 

658
00:35:34,960 --> 00:35:36,520
There's always a Wall Street 
bias, right? 

659
00:35:37,640 --> 00:35:41,400
You know, if you go and ask, you
know, 90% of people, and we did 

660
00:35:41,640 --> 00:35:46,280
a, a survey on this too. 
You know, most people are still 

661
00:35:46,480 --> 00:35:48,800
using Google search and they 
still plan on using Google 

662
00:35:48,800 --> 00:35:52,320
search, right? 
Your use case, my use case, 

663
00:35:52,320 --> 00:35:54,160
really, it's a search, 
especially for informational 

664
00:35:54,160 --> 00:35:56,880
queries and research projects, 
is much more advanced and 

665
00:35:56,880 --> 00:36:00,480
nuanced than the use case for a 
lot of people. 

666
00:36:01,760 --> 00:36:05,440
If you just, if you don't 
believe me on that, just go to, 

667
00:36:05,440 --> 00:36:07,200
I'm not saying you, I'm just 
saying I'm listening. 

668
00:36:07,200 --> 00:36:09,640
Like go to Google's trending 
results page, like look at the 

669
00:36:09,640 --> 00:36:12,680
most trending topics. 
Like, it's not anything that you

670
00:36:12,680 --> 00:36:16,000
would probably be, you know, I'm
sure it's not Bill Brewster's. 

671
00:36:17,440 --> 00:36:19,360
Google. 
Search history is coming up as 

672
00:36:19,360 --> 00:36:20,720
most trending, let's just put it
that way. 

673
00:36:20,720 --> 00:36:24,440
Or or mine either. 
I think that if you look at it 

674
00:36:24,440 --> 00:36:26,280
from the abundance mindset, 
right? 

675
00:36:26,280 --> 00:36:29,880
And look at through history, it 
kind of comes down to a 

676
00:36:29,880 --> 00:36:33,000
company's willingness to disrupt
themselves, right? 

677
00:36:33,000 --> 00:36:37,720
And looking at their core 
technology and what they have in

678
00:36:37,720 --> 00:36:40,640
terms of assets to help them 
make that transition. 

679
00:36:40,640 --> 00:36:44,200
So, you know, tying it back to 
the innovators dilemma, it's 

680
00:36:44,200 --> 00:36:47,880
like, OK, you know, even if you 
do have the assets, what's your 

681
00:36:47,880 --> 00:36:50,360
actually willingness to to 
disrupt yourself? 

682
00:36:50,400 --> 00:36:52,960
And if that's the case, at what 
speed are you willing to do 

683
00:36:52,960 --> 00:36:54,280
that? 
And what does it look like on 

684
00:36:54,280 --> 00:36:57,480
the other side? 
You know, in the case of looking

685
00:36:57,480 --> 00:37:01,560
at things like, you know, when 
the when the mobile phone, when 

686
00:37:01,560 --> 00:37:05,480
the when cell phones first came 
out, right, it was a yeah, when 

687
00:37:05,480 --> 00:37:07,200
cell phones came out, I was 
like, OK, this is going to be a 

688
00:37:07,200 --> 00:37:09,520
replacement for the landline, 
right? 

689
00:37:09,560 --> 00:37:13,880
Fast forward 1520 years and it's
like, OK, this is actually, you 

690
00:37:13,880 --> 00:37:18,240
know, an entire computing 
platform at on your hand, like 

691
00:37:18,240 --> 00:37:21,280
in your hand, right? 
And then and then the ecosystem 

692
00:37:21,280 --> 00:37:25,120
develops off that through the 
App Store that creates billions 

693
00:37:25,120 --> 00:37:27,520
and billions and billions of 
dollars of worth of value. 

694
00:37:27,520 --> 00:37:30,960
Where if you just looked at it 
from when the phone first came 

695
00:37:30,960 --> 00:37:34,200
out and you just looked at it 
from a OK, this is going to 

696
00:37:34,200 --> 00:37:36,960
replace landlines. 
Like, yeah, like landlines ended

697
00:37:36,960 --> 00:37:41,600
up getting like going away, 
sure, but it ended up being. 

698
00:37:41,960 --> 00:37:44,560
Just more. 
You can go back to look at 

699
00:37:44,640 --> 00:37:46,560
Kodak, right, is another 
example. 

700
00:37:46,560 --> 00:37:49,200
Like they actually kind of they 
invented like a digital camera 

701
00:37:49,200 --> 00:37:52,800
and then they were like, OK, 
well, we have to preserve film. 

702
00:37:53,160 --> 00:37:55,600
So they buried it, right, 
because they didn't want to self

703
00:37:55,600 --> 00:37:59,280
their self themselves. 
So there's just so many examples

704
00:37:59,280 --> 00:38:01,720
of that in history. 
I can keep going on. 

705
00:38:02,560 --> 00:38:06,520
But in the case of Google, I 
think that look, search as we 

706
00:38:06,520 --> 00:38:11,120
know knew it is gone like that 
is I agree with that 100%. 

707
00:38:11,760 --> 00:38:16,040
Now Google is disrupting itself 
currently for the use of AI 

708
00:38:16,040 --> 00:38:19,640
overuse rolling that out, right?
They're they're currently going 

709
00:38:19,640 --> 00:38:22,840
to the run phase and now they're
also rolling out AI mode. 

710
00:38:23,080 --> 00:38:26,960
So the all the reason why the 
only reason why Bill that every 

711
00:38:26,960 --> 00:38:30,560
you and everybody else and 
myself initially was so 

712
00:38:30,560 --> 00:38:34,880
attracted to ChatGPT and all 
these other bots that are giving

713
00:38:34,880 --> 00:38:38,160
all this information is because 
they provided users with a new 

714
00:38:38,360 --> 00:38:41,800
discovery factor, right to that 
replaced search. 

715
00:38:41,840 --> 00:38:44,960
And otherwise, if you look at 
Google, given their billions of 

716
00:38:44,960 --> 00:38:47,880
users, if they had just 
disrupted themselves sooner, 

717
00:38:48,200 --> 00:38:49,480
right? 
It would have made it so much 

718
00:38:49,480 --> 00:38:53,200
more difficult for an open AI 
and ChatGPT to actually come 

719
00:38:53,200 --> 00:38:56,480
into the market. 
Because instead, when ChatGPT 

720
00:38:56,480 --> 00:38:59,040
came out and it would have been 
like, oh, what is this like, Oh 

721
00:38:59,040 --> 00:39:00,880
yeah, this is just Google. 
Google already did this. 

722
00:39:01,240 --> 00:39:05,000
Now the question becomes, is it 
too late for Google? 

723
00:39:07,400 --> 00:39:13,520
And I think a year ago that was 
in my mind, especially in the 

724
00:39:13,520 --> 00:39:18,880
beginning of 2024 and a year 
ago, it was more of a 

725
00:39:18,880 --> 00:39:21,120
possibility, right? 
Like I had serious questions 

726
00:39:21,120 --> 00:39:26,080
about management's ability to 
execute Gemini in terms of the, 

727
00:39:26,240 --> 00:39:28,560
you know, 1/5. 
It wasn't in a good model. 

728
00:39:28,560 --> 00:39:30,680
I wasn't using Gemini. 
I tried using Gemini. 

729
00:39:30,960 --> 00:39:35,600
So I've, you know, been doing 
all this coding and, and, and 

730
00:39:35,600 --> 00:39:39,240
building applications and all 
these things that I naturally 

731
00:39:39,240 --> 00:39:42,720
just came back to Claude because
I tried everything else and it 

732
00:39:42,720 --> 00:39:45,240
was giving me bad code and it 
wasn't working and it was 

733
00:39:45,240 --> 00:39:47,120
breaking. 
And I was just like, well, This 

734
00:39:47,120 --> 00:39:50,000
is why, like, why am I ever 
going to use this if it's giving

735
00:39:50,000 --> 00:39:52,680
me bad results? 
And so I went and went back to 

736
00:39:52,680 --> 00:39:58,320
Anthropic, which owns Claude. 
And then that shifted earlier in

737
00:39:58,320 --> 00:40:01,120
the year when 2.5 came out and 
everything that I thought about 

738
00:40:01,120 --> 00:40:04,560
Google and their infrastructure 
advantage related to a six TP 

739
00:40:04,560 --> 00:40:08,120
use, having all the data, the 
tools to be able to give like 

740
00:40:08,360 --> 00:40:10,920
have the best model out there 
were 2.5 pro. 

741
00:40:10,960 --> 00:40:15,400
That was a huge unlock for them.
Like massive unlock. 

742
00:40:15,400 --> 00:40:18,800
You cannot be bare. 
You cannot be bullish on on 

743
00:40:18,800 --> 00:40:21,040
Google. 
If they were not, they didn't 

744
00:40:21,040 --> 00:40:24,200
have the best models. 
Now they're leading the frontier

745
00:40:24,560 --> 00:40:27,320
in terms of. 
Lowest token cost, highest 

746
00:40:27,320 --> 00:40:30,360
quality and as you know and I 
know you're like. 

747
00:40:30,360 --> 00:40:32,800
That then you get the developers
using it, yeah. 

748
00:40:33,040 --> 00:40:34,960
Absolutely. 
And by the way, like you know 

749
00:40:34,960 --> 00:40:37,960
how this works too, because you 
studied like network effects and

750
00:40:37,960 --> 00:40:41,920
scale networks and and even kind
of like can come back down to a 

751
00:40:41,920 --> 00:40:44,760
cable like or like a fiber like 
I'm not. 

752
00:40:45,440 --> 00:40:48,440
Why are you trying to bring up 
like old PTSD man? 

753
00:40:49,320 --> 00:40:53,120
No, you know what I mean, Like 
or, or or DSL like the whole, 

754
00:40:53,120 --> 00:40:55,000
the whole like, I think it's 
actually like a really 

755
00:40:55,000 --> 00:40:59,840
fascinating parallel because you
know, the transition from DSL to

756
00:40:59,840 --> 00:41:04,080
cable to fiber, right? 
The whole premise of that is 

757
00:41:04,080 --> 00:41:08,400
around new technologies coming 
in and being able to provide the

758
00:41:08,400 --> 00:41:10,560
fastest service at the lowest 
cost. 

759
00:41:11,120 --> 00:41:14,200
Yeah. 
And there's really no room for 

760
00:41:14,200 --> 00:41:16,960
any other alternative. 
Like if you can't, if you don't 

761
00:41:17,000 --> 00:41:20,640
have that structural cost 
advantage, right, then you're 

762
00:41:20,640 --> 00:41:23,240
always going to be behind. 
So in the case of, you know, 

763
00:41:23,240 --> 00:41:26,840
fiber versus cable, they can, 
cable can do whatever they want 

764
00:41:26,840 --> 00:41:29,680
with DOCSIS, right? 
But the reality is that because 

765
00:41:29,680 --> 00:41:32,800
fiber is just structurally 
faster and it's a soft and it's 

766
00:41:32,960 --> 00:41:36,200
largely software upgrade. 
As soon as cable gets to 2 gig 

767
00:41:36,200 --> 00:41:39,840
symmetrical, they can come out 
and you know, go to 10 gig 

768
00:41:39,840 --> 00:41:44,560
symmetrical, 15 gig symmetrical 
because the marginal cost right 

769
00:41:44,560 --> 00:41:47,320
to deliver that extra bit for 
that is like 0. 

770
00:41:47,800 --> 00:41:50,760
Yeah, right. 
And while the marginal cost to 

771
00:41:50,760 --> 00:41:54,120
deliver like the extra level of 
compute or inference capacity 

772
00:41:54,120 --> 00:41:58,720
for an LLM isn't 0. 
But if you're going to actually 

773
00:41:58,720 --> 00:42:02,480
develop something to and, and 
make this a mass market product 

774
00:42:02,480 --> 00:42:07,760
at scale and incorporate your AI
across billions of users to 

775
00:42:07,760 --> 00:42:10,560
drive adoption like Google has 
across all their applications, 

776
00:42:10,960 --> 00:42:14,000
you better have a massive cost 
advantage. 

777
00:42:14,000 --> 00:42:17,480
And they have that with TPU. 
So the inference capacity, 

778
00:42:17,880 --> 00:42:20,640
right? 
And, and inference is basically 

779
00:42:21,080 --> 00:42:23,800
the every single time you run a 
query, right? 

780
00:42:23,960 --> 00:42:27,960
It fires off it's tokens, text 
get translated into series of 

781
00:42:27,960 --> 00:42:30,640
tokens. 
And then once you submit the 

782
00:42:30,640 --> 00:42:35,520
query, the GPU fires and it runs
the compute and gives you back 

783
00:42:35,520 --> 00:42:38,680
the results, right? 
It's like the tokenizing like I 

784
00:42:38,680 --> 00:42:41,480
when I was hosting Llama or 
whenever, whenever you do it, 

785
00:42:41,480 --> 00:42:45,040
like you hear the GPU fire, it's
like it's like an engine like 

786
00:42:45,040 --> 00:42:46,320
it's just like. 
Interesting. 

787
00:42:46,320 --> 00:42:48,080
Yeah, it's really cool. 
And then it goes up and it goes 

788
00:42:48,080 --> 00:42:49,760
down so you can hear it actually
happening. 

789
00:42:49,760 --> 00:42:53,040
It is pretty cool. 
But but you know, like, like the

790
00:42:53,040 --> 00:42:55,560
point is like you need for 
Google, you need to have that, 

791
00:42:55,600 --> 00:42:57,480
that scale advantage is huge, 
right? 

792
00:42:57,480 --> 00:43:00,040
Because that's how you can, they
can disrupt themselves. 

793
00:43:00,040 --> 00:43:02,080
And while there's probably going
to be margin implications in the

794
00:43:02,080 --> 00:43:06,280
short term, that's how you get 
AI adoption to scale. 

795
00:43:06,640 --> 00:43:11,080
And so now what you're, if you 
use Google, you have AI mode, 

796
00:43:11,120 --> 00:43:13,920
like they're embedding AI mode, 
which is based, you know, kind 

797
00:43:13,920 --> 00:43:18,640
of deep search across the board 
right now, deep research if you 

798
00:43:18,640 --> 00:43:21,640
want to use that, which is 
really cool by the way, you 

799
00:43:21,640 --> 00:43:23,280
know. 
Yeah, deep research is awesome, 

800
00:43:23,280 --> 00:43:25,640
so. 
Cool, like I and but you need a 

801
00:43:25,640 --> 00:43:27,360
paid subscription for that, 
right? 

802
00:43:27,920 --> 00:43:29,720
Yes, and I would pay more for 
the record. 

803
00:43:30,040 --> 00:43:31,880
Yeah. 
And it's why is that? 

804
00:43:31,880 --> 00:43:35,920
Because it's expensive now. 
Yeah, to to run it up. 

805
00:43:35,920 --> 00:43:38,160
Yeah, that makes sense. 
And now Google can just is 

806
00:43:38,160 --> 00:43:41,240
incorporate can incorporate AI 
mode with deep search. 

807
00:43:41,600 --> 00:43:44,640
And because they have by the 
way, like the best search index 

808
00:43:44,640 --> 00:43:47,080
and all the history in the 
history of the like history of 

809
00:43:47,080 --> 00:43:49,080
the world, they have the best 
data, right? 

810
00:43:49,080 --> 00:43:51,520
They can they already know what 
users are looking for. 

811
00:43:51,520 --> 00:43:53,960
So they can pre train a lot of 
the models they can index, 

812
00:43:53,960 --> 00:43:57,280
create multiple indexes to 
better surface information 

813
00:43:57,280 --> 00:43:59,320
faster. 
And then because they have that 

814
00:43:59,320 --> 00:44:02,520
structural cost advantage with 
TP US and ASICS, which are like 

815
00:44:02,920 --> 00:44:06,800
probably anywhere from the 
research says they're like 5 to 

816
00:44:07,000 --> 00:44:12,680
almost, I've seen as low as 5 to
100 times more efficient than 

817
00:44:12,680 --> 00:44:15,240
the GPUs, right, when it comes 
to inference. 

818
00:44:15,240 --> 00:44:18,880
And even if it's like not 100 
times, but it's 10 times or if 

819
00:44:18,880 --> 00:44:20,160
it's five times, it's still 
huge. 

820
00:44:20,160 --> 00:44:22,600
It's still a huge cost, cost 
advantage that they have. 

821
00:44:23,240 --> 00:44:26,880
Then it's, it's they can embed 
this technology across all their

822
00:44:26,880 --> 00:44:30,000
applications, drive usage and do
things that the competitors 

823
00:44:30,000 --> 00:44:31,840
simply can't for a much lower 
cost. 

824
00:44:31,840 --> 00:44:36,120
And so my kind of hypothesis is 
that it's not too late for 

825
00:44:36,120 --> 00:44:40,360
Google that they are going to 
roll all this AI out across 

826
00:44:40,360 --> 00:44:42,840
every single part of the 
surface, try further adoption of

827
00:44:42,840 --> 00:44:45,320
it. 
And while, and that in a few 

828
00:44:45,320 --> 00:44:48,320
years from now, and hopefully 
sooner, if we're having a follow

829
00:44:48,320 --> 00:44:51,560
up conversation, Bill, you're 
saying that you cancelled your 

830
00:44:51,560 --> 00:44:54,000
perplexity subscription, some of
these other things because 

831
00:44:54,000 --> 00:44:57,160
Google, you know, just made all 
these things available through 

832
00:44:57,160 --> 00:44:59,000
it's, you know, a on mode 
interface. 

833
00:44:59,000 --> 00:45:02,080
And that there is really no 
longer traditional search as you

834
00:45:02,080 --> 00:45:03,880
know it. 
Instead, it's going to be, you 

835
00:45:03,880 --> 00:45:06,600
know, a chat bot, right? 
But it's going to be free and 

836
00:45:06,600 --> 00:45:08,240
they're going to have the best 
information. 

837
00:45:09,680 --> 00:45:11,720
And I think that's, that's my 
bet. 

838
00:45:11,720 --> 00:45:13,560
That's what I think is going to 
happen in the long run. 

839
00:45:13,560 --> 00:45:17,120
Now obviously in the short term,
you know, 6 to 9 months, there's

840
00:45:17,120 --> 00:45:18,600
going to be some disruption 
related to that. 

841
00:45:18,600 --> 00:45:23,000
But in the long term, they had 
every single asset for them to 

842
00:45:23,080 --> 00:45:25,320
succeed and they've been 
shipping product like crazies. 

843
00:45:25,320 --> 00:45:30,040
Anyway, that's that's kind of my
quick or not so quick thoughts 

844
00:45:30,040 --> 00:45:31,600
on it. 
No, that's interesting. 

845
00:45:31,600 --> 00:45:33,720
And the other, the other thing 
that you got me thinking about 

846
00:45:33,720 --> 00:45:38,760
is I remember talking to people 
about AWS when it was starting 

847
00:45:38,760 --> 00:45:41,960
and someone much smarter than me
said, you're not thinking enough

848
00:45:41,960 --> 00:45:46,920
about the, the cost going down 
and, and what that's going to 

849
00:45:46,920 --> 00:45:50,280
look like in five years. 
And to the extent that, that the

850
00:45:50,280 --> 00:45:55,120
costs decline like every other 
technological revolution has, if

851
00:45:55,120 --> 00:45:58,200
they, if they do keep me paying 
whatever I'm paying, they don't,

852
00:45:58,200 --> 00:46:00,680
they don't need to worry about 
necessarily raising the price 

853
00:46:00,680 --> 00:46:03,120
that I'm paying, right? 
They could just squeeze margin 

854
00:46:03,120 --> 00:46:07,520
out of efficiency gains, which 
is maybe what I should be 

855
00:46:07,520 --> 00:46:08,880
thinking more along the lines 
of. 

856
00:46:09,440 --> 00:46:12,000
Yeah and yeah. 
No, absolutely. 

857
00:46:12,160 --> 00:46:14,920
I'll tell you who would scare me
if I was an exec of a 

858
00:46:14,920 --> 00:46:17,400
traditional media company. 
You know, I think. 

859
00:46:17,640 --> 00:46:20,920
The amount of content that 
anyone can make now is. 

860
00:46:21,000 --> 00:46:24,320
Yeah, it is wild. 
I don't know what's more scary 

861
00:46:24,320 --> 00:46:29,400
to be as a exec at a traditional
media company because, I mean, 

862
00:46:29,400 --> 00:46:32,080
that's just been scary forever. 
Yeah, that's true. 

863
00:46:32,080 --> 00:46:34,640
It's just been awful. 
And actually I can make an 

864
00:46:34,640 --> 00:46:39,040
argument, I could make an 
argument that AI actually might 

865
00:46:39,040 --> 00:46:42,400
be better for them, like could 
help to them if, if there is 

866
00:46:42,400 --> 00:46:45,240
that structural, if it lowers 
the cost to producing content 

867
00:46:46,160 --> 00:46:49,280
and they can produce more 
content faster while maintaining

868
00:46:49,280 --> 00:46:53,400
quality or improving quality. 
And they have all the IP that 

869
00:46:53,400 --> 00:46:56,000
they already know is pretty 
bankable, right? 

870
00:46:56,000 --> 00:46:58,320
In terms of what people care 
about. 

871
00:46:58,880 --> 00:47:02,960
Then in theory, you could 
actually maybe see kind of that 

872
00:47:03,040 --> 00:47:07,080
value capture shift maybe back 
more directly towards the 

873
00:47:07,360 --> 00:47:09,320
studios. 
Yeah. 

874
00:47:09,360 --> 00:47:12,560
Where historically in the last 
five years, the the value ship 

875
00:47:12,560 --> 00:47:15,120
capture has been going more 
towards like talent and the 

876
00:47:15,120 --> 00:47:18,320
consumer. 
But but you're right, like the 

877
00:47:18,320 --> 00:47:21,680
proliferation of content is 
going to just be massive. 

878
00:47:22,560 --> 00:47:25,840
And so competition for eyeballs 
is only going to go up. 

879
00:47:26,200 --> 00:47:29,880
And on the flip side to your, I 
think to your, to your, you 

880
00:47:29,880 --> 00:47:34,760
know, the point you're you're 
alluding to before is that, you 

881
00:47:34,760 --> 00:47:38,800
know, if I can create cinematic 
quality or like very high 

882
00:47:38,800 --> 00:47:43,640
production value content at very
low costs, then the barriers to 

883
00:47:43,640 --> 00:47:47,680
entry get reduced. 
Then how can Paramount or Warner

884
00:47:47,680 --> 00:47:50,040
Brothers and all these studios 
like compete effectively? 

885
00:47:50,040 --> 00:47:53,400
And that's, that's, that's also 
true. 

886
00:47:53,400 --> 00:47:56,280
I mean, I mean, all of it's 
bullish for YouTube, right? 

887
00:47:56,280 --> 00:47:57,880
I mean, I think. 
That's well, that that's kind of

888
00:47:57,880 --> 00:48:01,360
where I get with it, right? 
That and and Instagram right. 

889
00:48:01,360 --> 00:48:03,440
To the extent that people are 
looking at reels. 

890
00:48:04,080 --> 00:48:06,800
I just I don't know. 
It's dude. 

891
00:48:06,800 --> 00:48:10,360
It's an interesting world, man. 
When one thing that I do worry 

892
00:48:10,360 --> 00:48:14,120
about is the amount of power 
that this is all requires. 

893
00:48:14,120 --> 00:48:16,440
It sort of seems to me that 
we've thrown the planet on the 

894
00:48:16,440 --> 00:48:19,880
back burner in this race. 
But but maybe I'm wrong. 

895
00:48:20,200 --> 00:48:21,120
I hope I'm wrong. 
Yeah. 

896
00:48:21,160 --> 00:48:24,880
I mean, there's been our energy 
analyst has done a lot of work 

897
00:48:24,880 --> 00:48:27,880
on this and a lot of the 
analysis out there says that, 

898
00:48:28,320 --> 00:48:32,000
you know, we just don't have the
amount of, you know, at the the 

899
00:48:32,000 --> 00:48:33,360
grid just can't handle it, 
right. 

900
00:48:33,360 --> 00:48:36,120
And so that's why we have like 
Meta and Google building nuclear

901
00:48:36,120 --> 00:48:37,840
power plants, which is also 
terrifying. 

902
00:48:38,280 --> 00:48:41,320
Like, like we're going to have 
pictures of like Mark Zuckerberg

903
00:48:41,320 --> 00:48:45,720
sitting there, like growing out 
in front of like, you know, 

904
00:48:46,440 --> 00:48:48,960
nuclear power plant. 
I don't know, it's, it's going 

905
00:48:49,840 --> 00:48:51,400
to be it's going to be 
interesting. 

906
00:48:51,400 --> 00:48:53,920
So, yeah, I mean there's, 
there's all these implications 

907
00:48:53,920 --> 00:48:57,680
and and that's also why like for
Meta, you know, they're they're 

908
00:48:57,680 --> 00:49:00,320
front loading a lot of the CapEx
spend related to data center 

909
00:49:00,320 --> 00:49:01,880
build outs, right. 
And the infrastructure piece, 

910
00:49:01,880 --> 00:49:07,480
because they know that the lead 
time to get a data center up is 

911
00:49:07,480 --> 00:49:10,880
like five years, five to seven 
years permitting, getting the 

912
00:49:10,880 --> 00:49:13,840
power set up, all these things. 
It's like they can they can get 

913
00:49:13,840 --> 00:49:17,880
GP us a lot faster then they can
actually build out the data 

914
00:49:17,880 --> 00:49:22,360
centers and NVIDIA won't deliver
GP us to their customers. 

915
00:49:22,360 --> 00:49:24,880
They don't have the ability to 
receive them. 

916
00:49:24,880 --> 00:49:26,600
And if you don't have the data 
center, that makes sense, you 

917
00:49:26,600 --> 00:49:27,880
don't have the ability to 
receive them. 

918
00:49:27,880 --> 00:49:31,840
So that's kind of I think a big 
part of where we are on that 

919
00:49:31,840 --> 00:49:33,520
front. 
It's it's just crazy. 

920
00:49:34,080 --> 00:49:35,960
I mean, I don't know what's were
like to your point before, like 

921
00:49:35,960 --> 00:49:39,000
I don't know what's scarier, who
should be more scared like a 

922
00:49:39,000 --> 00:49:42,920
traditional media company or a a
traditional software? 

923
00:49:43,120 --> 00:49:44,240
Traditional software is 
something, right? 

924
00:49:44,240 --> 00:49:47,280
We're like a SAS company, like 
a, a large incumbent, like a 

925
00:49:47,280 --> 00:49:54,200
sales force, if you think about 
it, because the whole barriers 

926
00:49:54,200 --> 00:49:56,400
to entry, barriers to 
development, lower cost argument

927
00:49:56,400 --> 00:50:01,960
is, is even more true, I think 
on the software side, because 

928
00:50:01,960 --> 00:50:06,120
with video content generation, 
even with Google's VO3, you can 

929
00:50:06,120 --> 00:50:09,080
still only like stitch together 
8 second clips, right? 

930
00:50:09,080 --> 00:50:11,520
And it's actually way more 
expensive to produce that type 

931
00:50:11,520 --> 00:50:13,800
of content than it's to reduce 
the lines of code. 

932
00:50:14,360 --> 00:50:16,800
And so for example, like I've 
been building software 

933
00:50:16,800 --> 00:50:21,800
applications just as a hobby, 
nights and weekends, just part 

934
00:50:21,800 --> 00:50:23,960
of the learning process. 
Part of it's just, you know, 

935
00:50:23,960 --> 00:50:28,000
like my family has a small 
vacation rental business out on 

936
00:50:28,000 --> 00:50:31,320
The Cave, right? 
And you know, one of the huge 

937
00:50:31,320 --> 00:50:34,760
issues we've been always had is 
around creating kind of a 

938
00:50:35,240 --> 00:50:39,960
customer database, right, to 
track everything, but also a 

939
00:50:39,960 --> 00:50:42,760
booking system. 
That is what we need, right? 

940
00:50:42,760 --> 00:50:44,920
We don't need anything crazy. 
It can be simple because we only

941
00:50:44,920 --> 00:50:47,920
have like, you know, it's like 3
properties. 

942
00:50:47,920 --> 00:50:51,800
So it's not like it's massive. 
So basically, you know, coded 

943
00:50:51,800 --> 00:50:56,520
out like our own little Airbnb 
app, right? 

944
00:50:56,600 --> 00:51:00,120
And it's all hosted front end 
develops. 

945
00:51:00,120 --> 00:51:04,400
It's a next JS application. 
It's all back ended on Postgres 

946
00:51:04,440 --> 00:51:08,040
with the database and you know, 
using Stripe for payment 

947
00:51:08,040 --> 00:51:10,920
integration. 
We actually, I also like 

948
00:51:10,920 --> 00:51:15,560
developed a rag off of it. 
So you can, we were able to take

949
00:51:15,560 --> 00:51:18,480
all of like all the paper and my
dad's old school, right? 

950
00:51:18,480 --> 00:51:21,720
He's like, like he was still 
using AOL press and like from 

951
00:51:21,720 --> 00:51:24,040
late 90s to do his website. 
Like they deprecated that thing 

952
00:51:24,040 --> 00:51:25,920
15 years ago. 
Like I, I think he's like the 

953
00:51:25,920 --> 00:51:29,240
last customer on it. 
And, but we basically digitized 

954
00:51:29,240 --> 00:51:32,800
everything, put it was able to 
using AI, was able to get it 

955
00:51:32,800 --> 00:51:37,320
into a database. 
And then we created a, a, a rag 

956
00:51:37,400 --> 00:51:40,560
with a vector database, which 
basically allows us to run 

957
00:51:40,560 --> 00:51:44,400
natural language queries against
all the data, which is like what

958
00:51:45,360 --> 00:51:48,960
it's like, OK, we have this 
availability for the third week 

959
00:51:48,960 --> 00:51:50,840
in June. 
What rate should we charge or 

960
00:51:50,840 --> 00:51:52,560
who should we? 
Can you give us a list of people

961
00:51:52,560 --> 00:51:56,480
that we should reach out to that
are the most likely to, you 

962
00:51:56,480 --> 00:51:58,560
know, book that week? 
Interesting. 

963
00:51:58,560 --> 00:52:02,000
And it goes back through 30 
years of information and returns

964
00:52:02,000 --> 00:52:06,920
us a source document with going 
back to the actual data, right, 

965
00:52:06,920 --> 00:52:09,480
that we have in the database of 
who we should reach out to and 

966
00:52:09,480 --> 00:52:11,680
why. 
And it's, you know, like that 

967
00:52:11,680 --> 00:52:17,240
took that was probably took me 
like 50 hours of like just 

968
00:52:17,240 --> 00:52:18,240
really. 
Yeah. 

969
00:52:18,280 --> 00:52:19,800
And. 
That's pretty quick. 

970
00:52:20,240 --> 00:52:23,280
Yeah, it actually is right. 
Like, yeah, for like, and that's

971
00:52:23,280 --> 00:52:26,880
me. 
Like not having and then by in 

972
00:52:26,880 --> 00:52:30,080
Full disclosure, like I didn't 
know how to code like up until 

973
00:52:30,080 --> 00:52:31,640
like Python. 
Any I know how to do any of 

974
00:52:31,640 --> 00:52:32,760
this. 
Like, I mean, I was always even 

975
00:52:32,760 --> 00:52:38,160
technically savvy, but and, and 
curious, but I'd actually like 

976
00:52:38,200 --> 00:52:43,440
up until like the very beginning
part of 2024, I had literally 

977
00:52:43,440 --> 00:52:45,760
just given up on coding. 
Like it's something I've always 

978
00:52:45,760 --> 00:52:48,520
wanted to do. 
I everyone's like, do you do all

979
00:52:48,520 --> 00:52:50,400
this stuff with data? 
Like you got to learn Python, 

980
00:52:50,920 --> 00:52:53,440
right? 
And I was just like, look, I'm, 

981
00:52:53,720 --> 00:52:57,280
I'm not old by any means, but I 
got two kids. 

982
00:52:57,280 --> 00:53:00,160
I'm kind of like in the middle 
enter like that middle phase of 

983
00:53:00,160 --> 00:53:03,840
my career. 
Like I don't have the like, it 

984
00:53:03,840 --> 00:53:08,080
takes too much time to like, you
know, to actually learn how to 

985
00:53:08,080 --> 00:53:10,720
code, but not just the basics, 
like to learn how to code to 

986
00:53:10,720 --> 00:53:13,600
actually be good enough to 
actually have it have a positive

987
00:53:13,600 --> 00:53:16,280
incremental impact from where 
I'm at. 

988
00:53:16,560 --> 00:53:18,080
So I kind of just like gave up 
on it. 

989
00:53:18,080 --> 00:53:19,920
I was like, yeah, I just 
accepted the fact that I'm never

990
00:53:19,920 --> 00:53:23,000
going to be able to do this. 
And then AI came along and holy 

991
00:53:23,000 --> 00:53:24,640
crap, it's just been life 
changing. 

992
00:53:24,640 --> 00:53:27,960
And so you know, yeah. 
So like 50 hours. 

993
00:53:27,960 --> 00:53:31,360
And is it, could it be, is it is
what I built like technically 

994
00:53:31,720 --> 00:53:34,960
like production ready, secure 
and scalable to like, you know, 

995
00:53:35,040 --> 00:53:36,160
thousands and millions of 
people? 

996
00:53:36,160 --> 00:53:38,960
Of course not, right. 
Like you still need, like if 

997
00:53:38,960 --> 00:53:41,200
that were ever going to be the 
case, you need a senior 

998
00:53:41,200 --> 00:53:43,160
engineer, you need to do all 
those things. 

999
00:53:43,320 --> 00:53:45,840
But you know, for our purpose, 
it, it works just fine. 

1000
00:53:46,480 --> 00:53:49,800
I think, you know, the next step
of it is we're working on 

1001
00:53:49,800 --> 00:53:53,160
creating a true agent, not just 
a workflow, but like an actual 

1002
00:53:53,160 --> 00:54:00,480
agent that can help source 
bookings and actually book for 

1003
00:54:00,480 --> 00:54:01,640
us. 
So like we don't. 

1004
00:54:02,280 --> 00:54:04,520
So, you know, they don't have to
necessarily call and speak to 

1005
00:54:04,520 --> 00:54:08,240
somebody, I mean, me or my 
sister or my dad, we can just, 

1006
00:54:08,640 --> 00:54:11,320
you know, they can just use an 
agent and they can handle a lot 

1007
00:54:11,320 --> 00:54:13,920
of the complexities associated 
with the booking process. 

1008
00:54:14,320 --> 00:54:16,640
And it can just be more of a 
stand alone system. 

1009
00:54:16,640 --> 00:54:20,200
And it can, it can handle all of
the manual processes like 

1010
00:54:20,200 --> 00:54:24,040
getting, you know, leases 
signed, coordinating with 

1011
00:54:24,600 --> 00:54:27,560
cleaners to have them come in, 
all these things. 

1012
00:54:27,560 --> 00:54:32,040
And so, you know, it, it, it's 
just, it's amazing in terms of 

1013
00:54:32,080 --> 00:54:33,600
like what a lot of this stuff 
can do. 

1014
00:54:33,600 --> 00:54:37,760
And yeah, I think the biggest. 
And so, so, so OK, so why did I 

1015
00:54:37,760 --> 00:54:41,120
say all that and said it because
I think if you're, you know, an 

1016
00:54:41,120 --> 00:54:43,520
enterprise software company, 
right? 

1017
00:54:43,840 --> 00:54:47,040
Yeah, like, you know, like like 
a lot of these things are 

1018
00:54:47,040 --> 00:54:51,120
enterprise grade yet, but the 
lower cost to entry to develop 

1019
00:54:51,120 --> 00:54:53,600
software is a really big deal. 
And it's going to matter over 

1020
00:54:53,600 --> 00:54:56,240
time because you take something 
like a monday.com, which is a 

1021
00:54:56,240 --> 00:54:58,880
task at its core is a basic task
management solution. 

1022
00:54:58,920 --> 00:55:02,440
I could, I could just I could 
code out something with probably

1023
00:55:02,440 --> 00:55:04,720
80% functionality of that in a 
weekend. 

1024
00:55:05,360 --> 00:55:08,240
But literally now I can't get it
to scale because, you know, 

1025
00:55:08,240 --> 00:55:10,840
because I don't have the sales 
team and all this stuff, but the

1026
00:55:10,840 --> 00:55:14,280
concept is kind of the same. 
So on the one hand, the barriers

1027
00:55:14,280 --> 00:55:17,320
to scale are still there just 
like everything else, right? 

1028
00:55:17,400 --> 00:55:22,440
But on the other hand, there 
will be a next wave of billion 

1029
00:55:22,440 --> 00:55:25,280
dollar companies, multi billion 
dollar companies that have been 

1030
00:55:25,280 --> 00:55:30,440
built using AI, you know, 110th 
of the budget that a lot of 

1031
00:55:30,440 --> 00:55:33,080
these predecessor software 
companies have. 

1032
00:55:33,080 --> 00:55:36,600
And so just like we've seen with
disruption before, right, scale,

1033
00:55:36,600 --> 00:55:39,120
lower cost advantage, they're 
going to translate that 

1034
00:55:39,480 --> 00:55:42,120
structurally lower cost 
advantage into lower pricing 

1035
00:55:42,120 --> 00:55:45,400
because they can and the 
existing cut, you know, the 

1036
00:55:45,400 --> 00:55:48,000
incumbents are not going to be 
able to compete on that, right? 

1037
00:55:48,120 --> 00:55:50,240
They're not a they're not going 
to be flexible enough, most 

1038
00:55:50,240 --> 00:55:53,320
likely to self disrupt and move 
faster and beat the unit 

1039
00:55:53,320 --> 00:55:56,440
economics. 
You know, like basically 

1040
00:55:56,440 --> 00:55:59,120
they're, you know, the cost of 
capital is going to be so much 

1041
00:56:00,200 --> 00:56:04,200
higher for them. 
I mean, plus capital is not the 

1042
00:56:04,200 --> 00:56:07,200
right word, right? 
Just because they, you know, 

1043
00:56:07,200 --> 00:56:09,880
their, their break even is going
to be so much lower for some of 

1044
00:56:09,880 --> 00:56:12,920
these newer companies that they 
can afford to price at like 

1045
00:56:12,920 --> 00:56:15,720
3040% that. 
So it's going to be like, well, 

1046
00:56:15,720 --> 00:56:22,400
look, if I can get 80% of the 
functionality for 50% less, 

1047
00:56:23,280 --> 00:56:24,680
that's a pretty good deal, 
right? 

1048
00:56:24,680 --> 00:56:27,480
And I, and I think that I think 
that's like the big part with 

1049
00:56:28,960 --> 00:56:31,600
software companies and like, I 
think about like a sales force 

1050
00:56:31,600 --> 00:56:34,480
and, and specifically and been 
thinking a lot about that with 

1051
00:56:34,480 --> 00:56:37,520
like ServiceNow, especially 
because AI is going to 

1052
00:56:37,680 --> 00:56:40,600
disintermediate a lot of 
software companies as we know 

1053
00:56:40,600 --> 00:56:43,800
it. 
And this whole concept of system

1054
00:56:43,800 --> 00:56:46,840
of record where you want to be 
the system of record, because 

1055
00:56:46,840 --> 00:56:51,400
that gave you pricing power, it 
would give you stickiness that 

1056
00:56:51,400 --> 00:56:55,920
is going to erode massively in, 
in AI and you know, the future 

1057
00:56:55,920 --> 00:56:59,040
because you're going to have 
interoperability. 

1058
00:56:59,040 --> 00:57:02,200
And the premise of being a 
system of record is that you 

1059
00:57:02,200 --> 00:57:04,320
kind of aren't interoperable. 
You are you have to be the 

1060
00:57:04,320 --> 00:57:07,480
platform and you don't talk well
with a lot of other software 

1061
00:57:07,480 --> 00:57:09,000
companies. 
And so you need to buy the 

1062
00:57:09,000 --> 00:57:11,680
module from the current platform
that you're on. 

1063
00:57:12,120 --> 00:57:15,800
And with AI agents, 
interoperability, MCP servers, 

1064
00:57:15,800 --> 00:57:18,440
all these things, the entire 
thing is breaking open where 

1065
00:57:18,880 --> 00:57:21,960
really they just kind of become 
glorified back end databases, 

1066
00:57:21,960 --> 00:57:24,560
right? 
And whoever can, you know, 

1067
00:57:24,560 --> 00:57:27,280
handle the front end AI 
orchestration is going to really

1068
00:57:27,280 --> 00:57:28,920
win today. 
And I think, you know, that's 

1069
00:57:28,920 --> 00:57:31,080
the service now thesis. 
And I think that's actually the 

1070
00:57:31,080 --> 00:57:33,280
right thesis. 
So it's going to be really, 

1071
00:57:33,280 --> 00:57:36,160
really fascinating. 
And, and my, and I will make one

1072
00:57:36,160 --> 00:57:39,800
last point on Google is, is 
that, you know, Google with all 

1073
00:57:39,800 --> 00:57:45,960
their assets in GCP, we're going
to get to a place where what I 

1074
00:57:45,960 --> 00:57:48,280
just described that I built, 
like I built that through 

1075
00:57:48,280 --> 00:57:53,600
individual prompts using Claude.
And it was very, you know, I 

1076
00:57:53,600 --> 00:57:58,120
want to say inefficient, but it 
was just told me what to do. 

1077
00:57:58,280 --> 00:58:01,120
I downloaded all the modules. 
It was all command line 

1078
00:58:01,120 --> 00:58:04,320
interface. 
But eventually, and we're seeing

1079
00:58:04,320 --> 00:58:07,840
this now with kind of like the 
rise of like lovable replit all 

1080
00:58:07,840 --> 00:58:11,240
these smaller platforms, like 
with Google, you're going to be 

1081
00:58:11,240 --> 00:58:13,320
able to, I'll be able to at one 
point in the future to say, 

1082
00:58:13,320 --> 00:58:15,720
like, look, here's my small 
business, right? 

1083
00:58:15,720 --> 00:58:17,960
We have no technology. 
This is what I want to do, 

1084
00:58:18,680 --> 00:58:20,560
right? 
It will prototype it out and 

1085
00:58:20,560 --> 00:58:23,560
then it will build that 
application for me and it will 

1086
00:58:23,840 --> 00:58:26,920
build the front end and it will 
build the back end like a full 

1087
00:58:26,920 --> 00:58:28,280
stack engineer. 
And it's all going to be 

1088
00:58:28,280 --> 00:58:30,440
integrated and built on Google's
cloud platform. 

1089
00:58:30,440 --> 00:58:33,560
And if you really think about 
that, like the implications of 

1090
00:58:33,560 --> 00:58:38,560
that, you know, that's huge. 
And because they have Gemini and

1091
00:58:38,560 --> 00:58:44,080
it's fully integrated, you know,
across the entire staff, they 

1092
00:58:44,080 --> 00:58:48,400
have owners economics, they have
TP us, then you're really 

1093
00:58:48,400 --> 00:58:51,080
looking at a situation where 
Google is going to be in a great

1094
00:58:51,080 --> 00:58:55,040
spot where the 1st place where 
any entrepreneur or any company 

1095
00:58:55,040 --> 00:58:56,840
that's going to want to build a 
new application is actually 

1096
00:58:56,840 --> 00:59:00,600
going to be on Google, right? 
Because Microsoft with Azure 

1097
00:59:01,160 --> 00:59:05,920
don't actually own their, you 
know, open AI while they have a 

1098
00:59:05,920 --> 00:59:07,960
stake in it, it's also a risk, 
right? 

1099
00:59:08,000 --> 00:59:11,360
And then Amazon, they have a 
stake in Anthropic, but they 

1100
00:59:11,360 --> 00:59:14,120
also don't really have strong 
models as well. 

1101
00:59:14,120 --> 00:59:16,880
So as far as all the hyper 
scalers go, Google's in like a 

1102
00:59:16,880 --> 00:59:20,880
really unique space. 
And I would say that that value 

1103
00:59:20,880 --> 00:59:24,800
long term is enormous, right? 
Like just to basically be like 

1104
00:59:24,800 --> 00:59:28,320
all the, you know, just to be 
able to go to a company and just

1105
00:59:28,480 --> 00:59:30,400
build a application. 
It's not. 

1106
00:59:30,440 --> 00:59:34,480
And I know it sounds like it 
sounds simple, but it's, it's 

1107
00:59:34,640 --> 00:59:37,240
the vision longer term, like 
that's what this is going to 

1108
00:59:37,240 --> 00:59:39,640
turn into. 
And that's why I think a huge 

1109
00:59:39,640 --> 00:59:43,680
part of this too is like people 
are over focusing on search as 

1110
00:59:44,200 --> 00:59:45,600
AI. 
You're not looking enough on, 

1111
00:59:45,800 --> 00:59:48,680
you know, the GCP component and 
what this can mean long term. 

1112
00:59:48,680 --> 00:59:51,440
So. 
Reminds me of when people were 

1113
00:59:51,440 --> 00:59:53,840
focused on video and not 
broadband margins. 

1114
00:59:54,160 --> 00:59:56,800
A little. 
A little, yeah. 

1115
00:59:56,800 --> 00:59:59,280
A little not. 
Not totally, but a little, yeah.

1116
00:59:59,880 --> 01:00:02,920
That's wild, man. 
I, you know, the we, you would 

1117
01:00:02,920 --> 01:00:06,080
have mentioned something about 
your small business. 

1118
01:00:07,000 --> 01:00:11,200
You want to create an agent. 
The other slide that like pops 

1119
01:00:11,200 --> 01:00:14,320
out in my head from that Mary 
Meeker deck is I think they said

1120
01:00:14,320 --> 01:00:18,240
73% of people can't tell the 
difference when they're talking 

1121
01:00:18,240 --> 01:00:21,720
to an AI agent versus a human, 
like in customer service. 

1122
01:00:22,640 --> 01:00:25,000
That's crazy. 
Yeah, it's only going to get 

1123
01:00:25,000 --> 01:00:27,840
even. 
That numbers going to go even 

1124
01:00:27,840 --> 01:00:30,560
higher. 
Yeah, Akram, he said this a long

1125
01:00:30,560 --> 01:00:33,440
time ago, but I think he, I 
think it was, it was prescient 

1126
01:00:33,440 --> 01:00:35,240
and correct. 
But like he was like, man, I 

1127
01:00:35,240 --> 01:00:39,120
could see a firm like American 
Express really benefiting from 

1128
01:00:39,160 --> 01:00:41,440
AI. 
That was like one of his early 

1129
01:00:41,440 --> 01:00:43,160
things that he said. 
And I was like, yeah, that does 

1130
01:00:43,160 --> 01:00:44,040
kind of make a. 
Lot of sense. 

1131
01:00:44,040 --> 01:00:48,720
What was the rationale on that? 
I, I mean, I don't want to speak

1132
01:00:48,720 --> 01:00:52,000
for them, but when I think about
it, the, the things that make 

1133
01:00:52,000 --> 01:00:55,880
sense to me is to the extent 
that you're analyzing something 

1134
01:00:55,880 --> 01:01:00,680
like credit risk and different 
data signals, I could see AI 

1135
01:01:00,920 --> 01:01:05,000
possibly helping early warning 
systems on when to maybe reach 

1136
01:01:05,000 --> 01:01:07,040
out to people. 
And to the extent that you're 

1137
01:01:07,040 --> 01:01:09,640
dealing with customer service, I
could see, you know, I mean, 

1138
01:01:09,640 --> 01:01:12,640
look, if 73% of people can't 
tell the difference right now, 

1139
01:01:12,840 --> 01:01:15,360
you probably need a lot fewer 
customer service people. 

1140
01:01:15,760 --> 01:01:16,560
Yeah. 
But. 

1141
01:01:16,600 --> 01:01:19,880
No, it's he would need to. 
Yeah, well, I mean, let's let's 

1142
01:01:20,000 --> 01:01:21,640
think about it in terms of our 
jobs, right? 

1143
01:01:21,640 --> 01:01:29,600
Like, so I, so I kind of, I say 
this jokingly, but I'm kind of 

1144
01:01:29,600 --> 01:01:34,480
serious about it or become more 
serious about it is like, you 

1145
01:01:34,480 --> 01:01:39,080
know, like I I'm actively trying
to replace myself with AI. 

1146
01:01:39,840 --> 01:01:44,840
Like it's a it's a challenge. 
And a lot of people say to kind 

1147
01:01:44,840 --> 01:01:48,440
of like they don't want it, no 
resistance to that, right? 

1148
01:01:48,440 --> 01:01:50,240
Because it's like, how could you
possibly do that? 

1149
01:01:51,160 --> 01:01:54,640
I think it's like an 
inevitability, right? 

1150
01:01:54,640 --> 01:01:57,240
Yeah, in many ways. 
And I think that it's not 

1151
01:01:57,640 --> 01:01:59,920
ultimately like, again, like I 
don't think we should think 

1152
01:01:59,920 --> 01:02:01,960
about this or people should 
necessarily think about this as 

1153
01:02:01,960 --> 01:02:03,840
being replaced. 
I think it's like augmented, 

1154
01:02:03,840 --> 01:02:07,720
right? 
Like you're using AI to make 

1155
01:02:07,720 --> 01:02:12,720
yourself better, faster, 
cheaper, right? 

1156
01:02:13,840 --> 01:02:17,720
And when I say cheaper, I mean 
like to be able to do like lower

1157
01:02:17,720 --> 01:02:20,040
your own, like to be able to do 
things at a lower cost, to do 

1158
01:02:20,040 --> 01:02:22,720
more of it, right? 
And therefore that's a huge 

1159
01:02:22,720 --> 01:02:25,480
competitive advantage if done 
properly. 

1160
01:02:25,480 --> 01:02:27,840
And so you're enhancing your own
productivity. 

1161
01:02:28,240 --> 01:02:33,080
You're building things that you 
couldn't do previously or that 

1162
01:02:33,080 --> 01:02:34,840
you were never able to do 
before. 

1163
01:02:35,240 --> 01:02:39,400
You know, I'm doing, you know, 
my own workflow, you know, 

1164
01:02:39,400 --> 01:02:44,600
analysis, like for, for example,
the FCC has a bunch of great 

1165
01:02:44,600 --> 01:02:49,440
data on the cable space, right? 
So those are very, very large 

1166
01:02:49,440 --> 01:02:52,440
files. 
And you know, in a world of 

1167
01:02:52,440 --> 01:02:54,280
Excel, like you can't analyze 
them. 

1168
01:02:54,880 --> 01:03:00,680
But in a world of AI where you 
understand how databases work 

1169
01:03:00,680 --> 01:03:02,920
and you understand how to ask 
questions, which I think is 

1170
01:03:02,920 --> 01:03:04,600
actually probably the most 
important thing out of all of 

1171
01:03:04,600 --> 01:03:07,440
this is just, you know, if 
you're if you're not good at 

1172
01:03:07,440 --> 01:03:09,320
asking questions like you're not
going to be good at. 

1173
01:03:09,440 --> 01:03:11,240
Yeah, right. 
Comp engineering is real. 

1174
01:03:11,240 --> 01:03:13,800
Just like you could actually be 
a really good Googler and people

1175
01:03:13,800 --> 01:03:16,280
would laugh at you when you said
that, but it's actually true. 

1176
01:03:17,160 --> 01:03:20,560
When you ask it, will you ask 
the the model that you're 

1177
01:03:21,280 --> 01:03:24,040
working with if there's a better
way to ask the question? 

1178
01:03:24,040 --> 01:03:25,760
Like will you? 
Will you use it to help you 

1179
01:03:25,760 --> 01:03:28,800
reason through things or do you 
have to learn how to prompt it? 

1180
01:03:28,880 --> 01:03:32,360
Yeah. 
So I So what I'll do is I will. 

1181
01:03:32,360 --> 01:03:34,840
So usually what will happen, or 
This is why I started doing more

1182
01:03:34,840 --> 01:03:40,480
recently, is you can create a 
project in any of these apps, 

1183
01:03:40,480 --> 01:03:43,960
but really like I use thought as
my go to and you create a 

1184
01:03:43,960 --> 01:03:47,080
project and you can unsert 
project knowledge and think 

1185
01:03:47,080 --> 01:03:48,520
about this is like a knowledge 
graph. 

1186
01:03:48,520 --> 01:03:52,120
Because what ends up happening 
is that a lot of these AIS, they

1187
01:03:52,120 --> 01:03:54,400
have context limit window 
limits. 

1188
01:03:54,600 --> 01:03:56,600
You hit your limitations for the
chats. 

1189
01:03:57,680 --> 01:04:00,360
And so the project knowledge 
base is actually really helpful 

1190
01:04:00,360 --> 01:04:03,200
way to kind of get around some 
of that. 

1191
01:04:03,640 --> 01:04:06,000
And you give it like core 
information data. 

1192
01:04:06,000 --> 01:04:09,520
So maybe you would, in this 
case, I would upload, you know, 

1193
01:04:09,560 --> 01:04:13,680
a sample file of like 1000 
different rows of data, 15 

1194
01:04:13,680 --> 01:04:17,000
columns, right? 
And and then at the same time, I

1195
01:04:17,000 --> 01:04:20,240
would upload any type of 
developer docs or like the 

1196
01:04:20,240 --> 01:04:23,360
schema definitions that come 
from the data provider. 

1197
01:04:23,920 --> 01:04:28,360
And I put that in the project 
knowledge space and then I would

1198
01:04:28,440 --> 01:04:31,640
create a chat and I would ask 
it, I would give it my use case,

1199
01:04:31,800 --> 01:04:34,920
be specific, what I'm trying to 
achieve, what my outcome is. 

1200
01:04:34,920 --> 01:04:39,280
And I'll prompt it and saying 
like you are an expert, you 

1201
01:04:39,280 --> 01:04:41,240
know, you are a data engineer, 
right? 

1202
01:04:42,720 --> 01:04:45,840
You are a data engineer, but 
you've also, you know, helped 

1203
01:04:45,840 --> 01:04:50,680
build the FCC's, you know, 
national Broadband System and 

1204
01:04:50,680 --> 01:04:54,240
you have first hand experience 
working at charter comp, you 

1205
01:04:54,240 --> 01:04:56,640
know, like something like that. 
And, and all those prompts 

1206
01:04:56,640 --> 01:04:58,920
actually matter. 
And I'll ask it for a prompt 

1207
01:04:58,920 --> 01:05:01,120
than what I'm trying to achieve.
It will give me a prompt. 

1208
01:05:01,120 --> 01:05:05,000
I'll review the prompt, I'll 
other the prompt, and then I 

1209
01:05:05,000 --> 01:05:07,720
will take that prompt and I'll 
give it to probably two other a 

1210
01:05:07,720 --> 01:05:11,000
is and ask it to tell me what it
thinks about the prompt. 

1211
01:05:11,600 --> 01:05:14,800
Interesting it will. 
Give me other edits and insights

1212
01:05:14,880 --> 01:05:16,040
that I wouldn't have thought 
previously. 

1213
01:05:16,040 --> 01:05:18,960
Almost like a peer review is 
kind of the way to think about 

1214
01:05:18,960 --> 01:05:20,920
it. 
And then once I have that prompt

1215
01:05:20,920 --> 01:05:24,200
and I feel like I haven't missed
anything, I will take it, put it

1216
01:05:24,200 --> 01:05:27,840
into the instructions section of
the of the project chat. 

1217
01:05:28,400 --> 01:05:32,840
And then I'll just start asking 
for step by step instructions on

1218
01:05:32,840 --> 01:05:35,600
how to do things. 
And I think the way, the best 

1219
01:05:35,720 --> 01:05:39,240
way to do it is like, you can't,
like AI doesn't do one shots 

1220
01:05:39,240 --> 01:05:42,040
very well. 
AI is not just going to be give 

1221
01:05:42,040 --> 01:05:44,200
you the answer right away, but 
you have to chunk it. 

1222
01:05:44,200 --> 01:05:47,360
You have to think in steps and 
sometimes very small steps. 

1223
01:05:47,360 --> 01:05:50,760
And so if you retrain your 
thinking and expectations, 

1224
01:05:50,760 --> 01:05:54,120
almost like you're putting 
together like, you know, like I,

1225
01:05:54,120 --> 01:05:58,080
I've put together tons of like 
research process notes because 

1226
01:05:58,080 --> 01:06:00,080
we had analysts that come in and
I'm like, Hey, I need you to 

1227
01:06:00,080 --> 01:06:02,840
update this. 
And I've done a million times, 

1228
01:06:02,840 --> 01:06:05,080
but it's nuanced, right? 
You got to go to the government 

1229
01:06:05,080 --> 01:06:07,400
website, put this in Excel, 
update this chart to all these 

1230
01:06:07,400 --> 01:06:08,440
things. 
So I've written like very 

1231
01:06:08,440 --> 01:06:13,800
detailed step by And you know, 
there's some things there's 

1232
01:06:13,800 --> 01:06:17,360
always some first hand knowledge
of like little quirks that you 

1233
01:06:17,360 --> 01:06:20,200
have to deal with, but it's very
similar in the sense of AI. 

1234
01:06:20,240 --> 01:06:23,160
And then it gives you steps and 
then you just follow those 

1235
01:06:23,160 --> 01:06:24,560
steps. 
And then if you have a problem 

1236
01:06:24,560 --> 01:06:27,960
with something or you get stuck,
you know, you could actually 

1237
01:06:27,960 --> 01:06:31,160
take screenshots of your, of 
your page, right? 

1238
01:06:31,200 --> 01:06:34,960
Or in the case of like AI Dev 
studio from Google, they have 

1239
01:06:34,960 --> 01:06:37,320
this really cool functionality 
where you can do screen share. 

1240
01:06:37,720 --> 01:06:41,840
So literally it's you're sitting
there with a Gemini open screen 

1241
01:06:41,840 --> 01:06:44,200
sharing and it can see 
everything you're doing. 

1242
01:06:44,200 --> 01:06:47,320
Now again, compliance like this 
is not stuff I'm doing. 

1243
01:06:48,000 --> 01:06:50,560
You know, this is all this is 
stuff I'm doing as like 

1244
01:06:50,560 --> 01:06:54,320
projects, just my own. 
But you can use these tool. 

1245
01:06:55,560 --> 01:06:58,280
Look at, you know, it's 
basically like, see what you're 

1246
01:06:58,280 --> 01:07:00,400
doing. 
And then that case they're like,

1247
01:07:00,400 --> 01:07:02,320
OK, I see on your screen you're 
trying to do this. 

1248
01:07:02,320 --> 01:07:05,080
I see this code and it can give 
you insights as if you're 

1249
01:07:05,080 --> 01:07:08,280
sitting there with a senior 
software developer, an engineer,

1250
01:07:08,520 --> 01:07:12,200
like over your shoulder. 
And that's something that I 

1251
01:07:12,200 --> 01:07:15,520
would have to do. 
Previously we've had the higher 

1252
01:07:15,520 --> 01:07:18,080
talent to do that. 
I would have to get on their 

1253
01:07:18,080 --> 01:07:20,160
schedule. 
We'd only be able to do like one

1254
01:07:20,160 --> 01:07:23,120
hour sprints. 
It would not be very efficient 

1255
01:07:23,120 --> 01:07:24,520
probably. 
Then the meeting ends. 

1256
01:07:24,520 --> 01:07:26,320
You hope you both remember what 
you talked about. 

1257
01:07:26,400 --> 01:07:28,080
You. 
Get back together, you redo it. 

1258
01:07:28,440 --> 01:07:30,440
It's, yeah. 
And it's just like, you know, 

1259
01:07:30,440 --> 01:07:34,080
and it's like, and then and it's
also like a lot of this stuff is

1260
01:07:34,120 --> 01:07:36,400
just like, I need to sleep more.
So it's like I'm up. 

1261
01:07:36,400 --> 01:07:38,440
I have an idea. 
I'm doing it like 1:00 AM. 

1262
01:07:38,720 --> 01:07:40,880
It's like not calling a software
engineer do that. 

1263
01:07:40,880 --> 01:07:43,320
Like, so it's always on that 
kind of a thing. 

1264
01:07:43,600 --> 01:07:45,800
I mean, the biggest issue is 
just burning through credits and

1265
01:07:45,800 --> 01:07:48,080
all these things, right? 
That's why I have like, that's 

1266
01:07:48,080 --> 01:07:50,480
why I'm like spending $1000 a 
month on like all these 

1267
01:07:50,480 --> 01:07:54,280
different subscriptions because 
I'm literally just like burning 

1268
01:07:54,280 --> 01:07:57,760
through tokens even after 
becoming more efficient. 

1269
01:07:57,760 --> 01:07:59,720
Like I'm hitting these 
limitations like very, very 

1270
01:07:59,720 --> 01:08:02,000
quickly. 
But but yeah, that's, that's 

1271
01:08:02,000 --> 01:08:05,320
kind of like, I mean, and so, 
OK, so sorry, my point before, I

1272
01:08:05,320 --> 01:08:07,560
just keep going off. 
But point before is like with 

1273
01:08:07,560 --> 01:08:09,800
the some of this data set now 
it's like, OK, I couldn't do 

1274
01:08:09,800 --> 01:08:13,000
this in Excel. 
So, you know, let me have the 

1275
01:08:13,000 --> 01:08:16,279
data in there, put it into a let
me get into a database. 

1276
01:08:16,279 --> 01:08:18,439
Let's write a Python script on 
it, right. 

1277
01:08:18,479 --> 01:08:23,479
And it's like, wow, this is like
took took me, you know, 4 hours 

1278
01:08:23,479 --> 01:08:27,680
where before it would have taken
me like, you know, 20 to 30 

1279
01:08:27,680 --> 01:08:29,399
hours working with the software 
engineer. 

1280
01:08:30,040 --> 01:08:32,920
Yeah, it's amazing. 
It's really just amazing what it

1281
01:08:32,920 --> 01:08:36,760
can do as long as you're kind 
of, you know, willing to put in 

1282
01:08:36,760 --> 01:08:40,520
the effort and, again, 
understand what it is and what 

1283
01:08:40,520 --> 01:08:42,960
it isn't. 
Like it's not omnipotent. 

1284
01:08:43,200 --> 01:08:45,439
It's not going to give you the 
right answer right away. 

1285
01:08:45,439 --> 01:08:48,840
It's going to make mistakes. 
Just like to your point earlier 

1286
01:08:48,840 --> 01:08:50,960
in the conversation, Bill, it's 
like an intern, an analyst. 

1287
01:08:51,279 --> 01:08:52,560
People are going to make 
mistakes. 

1288
01:08:52,560 --> 01:08:55,920
The question is, is the margin 
of error better or worse? 

1289
01:08:56,359 --> 01:08:59,520
Yeah, Yeah, that's right. 
Do you think so? 

1290
01:08:59,520 --> 01:09:01,520
I had a guy that was was at my 
house. 

1291
01:09:01,520 --> 01:09:04,560
It was one of the first times 
I've actually gotten to just sit

1292
01:09:04,560 --> 01:09:06,240
through a software 
implementation. 

1293
01:09:06,240 --> 01:09:11,720
And he was working with a large 
firm and data quality really 

1294
01:09:11,720 --> 01:09:14,920
mattered, right. 
And like they were going through

1295
01:09:14,960 --> 01:09:17,120
all of the checks. 
It was all weekend long. 

1296
01:09:17,120 --> 01:09:19,680
They were just finding errors 
and then fixing errors. 

1297
01:09:19,680 --> 01:09:21,279
And it was three of them on the 
phone. 

1298
01:09:21,760 --> 01:09:25,319
And I thought to myself, I was 
like, why in the future, why 

1299
01:09:25,319 --> 01:09:28,640
can't a computer just like, do 
this and run through all the 

1300
01:09:28,640 --> 01:09:30,960
iterations and find the errors 
and fix itself? 

1301
01:09:30,960 --> 01:09:34,000
Now I understand you've got it, 
you've got to train it and 

1302
01:09:34,000 --> 01:09:36,240
you've got to know the answers 
or you've got to know the 

1303
01:09:36,240 --> 01:09:40,279
questions to ask and whatnot. 
But man, like, to the extent 

1304
01:09:40,279 --> 01:09:43,760
that you start reducing that 
kind of friction of a, my first 

1305
01:09:43,760 --> 01:09:45,720
thought was, boy, that's a 
sticky relationship. 

1306
01:09:45,720 --> 01:09:47,920
And my second thought was, I 
wonder if it'll ever get less 

1307
01:09:47,920 --> 01:09:50,160
sticky over time. 
Yeah, right. 

1308
01:09:50,160 --> 01:09:51,520
If computers can kind of. 
Like. 

1309
01:09:52,000 --> 01:09:54,920
Troubleshoot all that. 
So I mean, coding has been like 

1310
01:09:54,920 --> 01:10:00,200
the killer app so far with AI. 
But if you look at what Google 

1311
01:10:00,200 --> 01:10:04,160
just announced with their jewels
coding agent open AI release 

1312
01:10:04,160 --> 01:10:07,320
something to forget the what 
their what their forget the name

1313
01:10:07,320 --> 01:10:12,400
of what they called it. 
But basically you can take your 

1314
01:10:12,400 --> 01:10:16,760
entire code base and you can 
have an agent, a coding agent 

1315
01:10:16,760 --> 01:10:20,640
that's does just that and it can
scan through up and down back, 

1316
01:10:20,680 --> 01:10:23,360
you know, up, down, sideways, 
your entire code base and look 

1317
01:10:23,360 --> 01:10:25,760
for errors and they can change 
things in real time. 

1318
01:10:25,760 --> 01:10:29,480
And that's so that that part of 
it is going to get changed, is 

1319
01:10:29,480 --> 01:10:31,840
going to change too. 
So you're going to need as far 

1320
01:10:31,840 --> 01:10:35,280
as like engineers go, you're 
going to need a lot fewer to do 

1321
01:10:35,280 --> 01:10:39,440
things, but you're still going 
to need the top tier, right, 

1322
01:10:39,440 --> 01:10:42,040
Because you're still going to 
need some type of human 

1323
01:10:42,040 --> 01:10:45,600
validation, human verification. 
But that's where the that's 

1324
01:10:45,600 --> 01:10:48,040
actually probably going to be 
like an entirely like, you know 

1325
01:10:48,040 --> 01:10:51,160
how there's like ad verification
that will verify like all these 

1326
01:10:51,160 --> 01:10:53,560
things to see if an ad is 
fraudulent, if there's a 

1327
01:10:53,560 --> 01:10:55,720
watermark, all these things, 
it's going to be some right. 

1328
01:10:55,920 --> 01:11:00,640
You're going to have entire 
companies whose, you know, focus

1329
01:11:00,640 --> 01:11:03,640
is on creating like a 
verification system for AI to 

1330
01:11:03,640 --> 01:11:06,280
make sure that it's doing what 
it's supposed to be doing and 

1331
01:11:06,280 --> 01:11:10,680
that it's it's accurate. 
So and then that kind of goes 

1332
01:11:10,680 --> 01:11:12,440
into the whole quality component
of software. 

1333
01:11:12,480 --> 01:11:15,760
So that's yeah, I mean, it's, 
it's, it's pretty remarkable. 

1334
01:11:15,760 --> 01:11:19,200
And the speed at which you can 
develop software and fix code 

1335
01:11:19,200 --> 01:11:23,960
bases is, is really, it's really
something else Like it's, you 

1336
01:11:23,960 --> 01:11:27,280
know, who's going to who's going
to be writing lines of code. 

1337
01:11:28,120 --> 01:11:29,520
You know, it's just not going to
happen. 

1338
01:11:29,520 --> 01:11:30,400
You're going to be tweaking 
code. 

1339
01:11:31,640 --> 01:11:35,200
And and then also for our job, 
man, like have you like the way 

1340
01:11:35,200 --> 01:11:39,000
I the way I think it's kind of 
going to play out right is I 

1341
01:11:39,000 --> 01:11:44,160
think that you're basically 
going to have like model 

1342
01:11:44,160 --> 01:11:48,440
curators or model builders. 
So every single hedge fund by 

1343
01:11:48,440 --> 01:11:51,800
side. 
So like eventually you're going 

1344
01:11:51,800 --> 01:11:55,000
to need somebody on the team 
whose sole job is to basically 

1345
01:11:55,000 --> 01:12:01,000
fine tune models and build 
agents that specifically for, 

1346
01:12:03,080 --> 01:12:07,080
you know, individual teams or 
asset managers, processes and 

1347
01:12:07,080 --> 01:12:09,040
work flows. 
You're going to basically try 

1348
01:12:09,040 --> 01:12:11,360
to, and then the agents are 
going to be competing with each 

1349
01:12:11,360 --> 01:12:13,880
other just like you would hire 
someone for a specific role for 

1350
01:12:13,880 --> 01:12:15,120
an analyst role. 
But you're going to be doing 

1351
01:12:15,120 --> 01:12:18,760
that using AI and you're going 
to be training it on your own 

1352
01:12:18,760 --> 01:12:22,760
data, but you're also going to 
be probably have, you know, 

1353
01:12:22,960 --> 01:12:27,800
security specific agents or 
security specific models, right?

1354
01:12:27,800 --> 01:12:30,000
So that's the other thing when I
said before, like I'm trying to 

1355
01:12:30,000 --> 01:12:33,120
replace myself with AI, it's 
because I realized that I think,

1356
01:12:33,720 --> 01:12:37,640
you know, like on the South side
today, you're like, you have 

1357
01:12:37,640 --> 01:12:39,320
people that are like the axe 
right? 

1358
01:12:39,400 --> 01:12:41,280
In certain, in certain, certain 
sectors. 

1359
01:12:41,280 --> 01:12:43,600
And it's like, I, OK, I go to 
this person for this task. 

1360
01:12:43,600 --> 01:12:45,000
I go to that person for that 
task. 

1361
01:12:45,600 --> 01:12:50,000
And, you know, and I, I think 
that for better or worse for me,

1362
01:12:50,000 --> 01:12:52,840
right, like I'm get like Roku's 
already on my gravestone, like 

1363
01:12:52,840 --> 01:12:54,840
better or worse, like that's, 
that's what's happening. 

1364
01:12:55,000 --> 01:12:57,240
But I, I do think I know that 
company better than probably 

1365
01:12:57,240 --> 01:12:59,480
anyone else does. 
Again, for better for worse. 

1366
01:12:59,480 --> 01:13:01,680
And good models spend a ton of 
time on it. 

1367
01:13:01,760 --> 01:13:05,280
So people come to me, right, and
they pay me because they want to

1368
01:13:05,280 --> 01:13:08,200
know more about the company and 
they want to know what we think 

1369
01:13:08,200 --> 01:13:12,000
about certain things. 
Well, I can create a model and 

1370
01:13:12,000 --> 01:13:18,680
fine tune an LLM that's 
basically a reflection of all my

1371
01:13:18,680 --> 01:13:23,520
knowledge and data and insights 
on a specific company, not just 

1372
01:13:23,520 --> 01:13:26,800
a rag, which is like databasing,
querying for real time 

1373
01:13:26,800 --> 01:13:30,040
information and, and, you know, 
retrieving and providing like a 

1374
01:13:30,040 --> 01:13:34,880
text, a text response with 
sources, like actually train a 

1375
01:13:34,880 --> 01:13:38,800
model to think like me and with 
its knowledge. 

1376
01:13:38,800 --> 01:13:42,920
And because if you think of like
the way and, and there's more 

1377
01:13:42,920 --> 01:13:45,600
that goes into it, but like the 
way that you fine tune a model 

1378
01:13:45,720 --> 01:13:50,520
is just the Q&A. 
So you, you give it a bunch of 

1379
01:13:50,520 --> 01:13:55,360
Q&A, like thousands and 
thousands of examples and it 

1380
01:13:55,360 --> 01:13:59,720
just, well, and in the model, 
you know, depending on how good 

1381
01:13:59,720 --> 01:14:02,040
the model is and the weights you
give, it will train. 

1382
01:14:02,600 --> 01:14:06,880
And so then in theory, you could
have like Andrew Friedman's 

1383
01:14:06,880 --> 01:14:11,440
Roku, lol, or Andrew Friedman's 
Roku agent, right? 

1384
01:14:11,560 --> 01:14:15,160
And that's infinitely scalable 
to the extent that if you want 

1385
01:14:15,160 --> 01:14:18,880
information from it then and you
know, you can just chat with it 

1386
01:14:18,880 --> 01:14:20,600
whenever you want. 
You pay a certain amount for 

1387
01:14:20,600 --> 01:14:23,280
that, right? 
But I trained it, right. 

1388
01:14:23,280 --> 01:14:26,880
It's so therefore it's unique. 
And to my point before about 

1389
01:14:26,880 --> 01:14:29,640
like the efficiencies of certain
models, like compounding 

1390
01:14:29,640 --> 01:14:32,960
inefficiencies, like you're 
probably going to want to use 

1391
01:14:32,960 --> 01:14:35,760
that model instead of somebody 
else's model on the street, 

1392
01:14:36,160 --> 01:14:37,680
right? 
Because it was trained by me. 

1393
01:14:37,840 --> 01:14:40,640
That's how you can extend your 
own knowledge and your own 

1394
01:14:40,640 --> 01:14:46,320
expertise right in, in, in an AI
gentic type of world and, or, 

1395
01:14:46,360 --> 01:14:48,360
and that's also how you can 
easily be replaced, right? 

1396
01:14:48,360 --> 01:14:51,320
Because then all of a sudden, 
you know, you basically are 

1397
01:14:51,320 --> 01:14:55,720
breaking down like time, like 
like their time doesn't become a

1398
01:14:55,720 --> 01:14:58,600
friction. 
It's like when, you know, it's 

1399
01:14:58,600 --> 01:15:01,600
like when content, when Netflix 
came in, like video went just 

1400
01:15:01,600 --> 01:15:04,400
streaming, right? 
There's no more prime time five 

1401
01:15:04,400 --> 01:15:08,000
to seven, like you're no longer 
bounded by my schedule, right? 

1402
01:15:08,000 --> 01:15:10,760
You want to access information. 
And so you are reducing that 

1403
01:15:11,080 --> 01:15:12,880
friction. 
You're increasing information 

1404
01:15:12,880 --> 01:15:14,880
liquidity, you're getting to 
scale faster. 

1405
01:15:15,120 --> 01:15:18,280
And then I can monetize that and
that's why. 

1406
01:15:18,320 --> 01:15:19,840
And that's kind of a version of 
the future. 

1407
01:15:19,840 --> 01:15:22,920
So I think you're going to have 
like, you know, someone who 

1408
01:15:22,920 --> 01:15:25,560
specialized in AI that 
understands prompt engineering, 

1409
01:15:25,560 --> 01:15:29,240
that can actually train models, 
but we're not there today yet 

1410
01:15:29,240 --> 01:15:32,560
because training is expensive 
and it's hard to do or sorry, 

1411
01:15:32,560 --> 01:15:34,880
fine tuning is expensive and 
it's hard to do. 

1412
01:15:36,280 --> 01:15:40,160
Eventually, I think like the 
next, it's either going to be a 

1413
01:15:40,160 --> 01:15:43,640
hyper scale or somebody's going 
to come in and just make it so 

1414
01:15:43,640 --> 01:15:46,760
much easier for somebody to fine
tune a model based on their own 

1415
01:15:46,760 --> 01:15:50,240
expertise and their own 
knowledge base and scale that 

1416
01:15:50,240 --> 01:15:53,400
out that you're going to have 
like, you know, and set like 

1417
01:15:53,600 --> 01:15:56,320
there's marketplaces that emerge
for agents. 

1418
01:15:56,320 --> 01:15:58,920
There's already that case, but 
it's going to be more like open,

1419
01:15:59,480 --> 01:16:03,840
right, where you could have the 
Bill Brewster, you know, 

1420
01:16:03,840 --> 01:16:08,520
business group podcast, LLF, 
right? 

1421
01:16:08,520 --> 01:16:13,520
And it's basically proprietary. 
It's fine-tuned on every single 

1422
01:16:13,520 --> 01:16:16,600
transcript, every single 
question that you've ever asked 

1423
01:16:16,600 --> 01:16:19,800
anybody, right? 
And you can use that two fold. 

1424
01:16:19,800 --> 01:16:23,400
You could use that one either 
for internal purposes. 

1425
01:16:23,400 --> 01:16:26,320
So when you're preparing for the
next podcast, right? 

1426
01:16:26,400 --> 01:16:29,720
You can ask it for a series of 
questions and you can give it 

1427
01:16:29,800 --> 01:16:32,080
information on your guest, 
right? 

1428
01:16:32,200 --> 01:16:35,000
And you can do deep research on 
a topic, but it's also going to 

1429
01:16:35,000 --> 01:16:38,080
know your style, how you think, 
types of questions that you ask,

1430
01:16:38,360 --> 01:16:41,680
because all even in the world of
AI, authenticity is going to be 

1431
01:16:41,680 --> 01:16:44,200
even more important, right? 
And so. 

1432
01:16:44,400 --> 01:16:46,520
Yeah, although people won't know
if it's actually me. 

1433
01:16:46,520 --> 01:16:49,360
If it gets too good, they'll 
just they'll just have videos. 

1434
01:16:49,640 --> 01:16:52,000
Yeah, that's totally true. 
But I mean, it could help you 

1435
01:16:52,000 --> 01:16:54,240
like, you know, in that sense 
become more efficient, ask 

1436
01:16:54,240 --> 01:16:57,000
better questions faster. 
And then, you know, on the other

1437
01:16:57,000 --> 01:16:59,920
hand, too, it could be a form of
a product, right, where it's 

1438
01:16:59,920 --> 01:17:04,200
like this knowledge base, you 
can ask, ask questions to it and

1439
01:17:04,600 --> 01:17:06,600
you know, it will give you 
answers. 

1440
01:17:07,560 --> 01:17:09,800
And then, and actually you could
use a rag for this today if you 

1441
01:17:09,800 --> 01:17:12,600
really wanted to. 
But but I think like the future 

1442
01:17:12,600 --> 01:17:15,200
is going to be like fine tune 
models because that's just 

1443
01:17:15,560 --> 01:17:18,000
otherwise everything just 
becomes a wrapper, right? 

1444
01:17:18,000 --> 01:17:22,920
And everyone just has a 
different UXUI on the same 

1445
01:17:22,920 --> 01:17:24,760
model. 
And I think that, you know, 

1446
01:17:24,760 --> 01:17:27,640
while I think everyone's going 
to want to use the same model to

1447
01:17:27,640 --> 01:17:31,160
fine tune, but everyone's going 
to use the best model to fine 

1448
01:17:31,160 --> 01:17:32,920
tune, right? 
Because there's a cost 

1449
01:17:32,920 --> 01:17:34,920
implications and quality 
implications. 

1450
01:17:35,320 --> 01:17:37,760
But I think the fine tune 
element is something that a lot 

1451
01:17:37,760 --> 01:17:41,520
of people aren't talking about 
today, which I think to me at 

1452
01:17:41,520 --> 01:17:42,920
least seems like an 
inevitability. 

1453
01:17:43,880 --> 01:17:48,360
Because you're also, what that 
also does is it breaks down the 

1454
01:17:48,360 --> 01:17:53,080
problem of context. 
Because right now, you know, 

1455
01:17:53,400 --> 01:17:56,640
with like Claude, like you can 
only have like 100,200 thousand 

1456
01:17:56,640 --> 01:18:00,920
tokens, which means that just 
like a conversation over time, 

1457
01:18:00,920 --> 01:18:04,480
the longest conversation goes 
kind of either maybe you get 

1458
01:18:04,480 --> 01:18:07,040
tired, right? 
So your energy levels drop over 

1459
01:18:07,040 --> 01:18:11,400
the course of the podcast or 
conversation, or you've kind of 

1460
01:18:11,440 --> 01:18:13,000
forget what you're initially 
talking about. 

1461
01:18:13,000 --> 01:18:16,840
You go off topic, right? 
And or you can't recall what 

1462
01:18:16,840 --> 01:18:19,680
you're initially talking about. 
The models work the same way in 

1463
01:18:19,680 --> 01:18:24,120
the form of context windows. 
And, and that's why there's 

1464
01:18:24,120 --> 01:18:28,840
limits on the chats, right? 
And that's why, you know, yeah, 

1465
01:18:28,840 --> 01:18:31,120
that's why there's limits on the
length of the chats that you hit

1466
01:18:31,280 --> 01:18:33,680
a lot of time. 
That's also why, like, I like to

1467
01:18:33,880 --> 01:18:37,400
use sometimes, like multiple 
models because if I get stuck on

1468
01:18:37,400 --> 01:18:39,480
one, then I get the Gemini, 
right? 

1469
01:18:39,480 --> 01:18:42,000
And then Gemini's like, looks at
it differently with the 

1470
01:18:42,000 --> 01:18:43,680
different model, with different 
context. 

1471
01:18:44,080 --> 01:18:47,320
And it can sometimes solve 
problems that Claude got stuck 

1472
01:18:47,320 --> 01:18:50,280
on vice versa. 
And I think that's, you know, 

1473
01:18:50,360 --> 01:18:54,800
going to be a really big deal. 
We're not there today, but 

1474
01:18:54,800 --> 01:18:57,720
that's like, I think it would be
a pretty cool product, right, 

1475
01:18:57,720 --> 01:19:02,160
for like media for, you know, 
information in general and 

1476
01:19:02,160 --> 01:19:04,680
accessing it. 
It's going to be really 

1477
01:19:04,680 --> 01:19:06,480
transformative to a lot of 
people's experiences. 

1478
01:19:07,760 --> 01:19:11,920
So going back to a previous 
question, what, what do we think

1479
01:19:11,920 --> 01:19:14,920
this does to employment? 
Are we are we going to 20% 

1480
01:19:14,920 --> 01:19:19,560
unemployment or do we create 
enough new work and efficiency? 

1481
01:19:21,080 --> 01:19:25,360
I think that my perspective is 
really hard because I everyone 

1482
01:19:25,360 --> 01:19:27,320
internalizes it through their 
own lens. 

1483
01:19:27,320 --> 01:19:31,560
And I've come to realize, and I 
know I'm not trying to sound 

1484
01:19:31,560 --> 01:19:34,400
arrogant at this and it's 
actually probably A and, and 

1485
01:19:34,400 --> 01:19:38,200
this, but I've just come to 
realize that like not everyone 

1486
01:19:38,320 --> 01:19:40,960
kind of thinks like me in terms 
of like the ability to self 

1487
01:19:40,960 --> 01:19:45,360
disrupt and like find new 
technologies and you know, 

1488
01:19:46,640 --> 01:19:50,240
approach things from like this 
like total abundance mindset. 

1489
01:19:50,240 --> 01:19:52,560
So I try to. 
And the reason why I'm saying 

1490
01:19:52,560 --> 01:19:55,400
that is because like if, if, if 
I, if my base case was that 

1491
01:19:55,400 --> 01:19:58,360
everyone was like that right, 
then probably, yeah, like 

1492
01:19:58,360 --> 01:20:00,120
they're probably be like 30% 
unemployment. 

1493
01:20:00,480 --> 01:20:02,600
Yeah, Yeah. 
Well, everybody would be fine 

1494
01:20:02,600 --> 01:20:04,760
disintermediating their own job,
right? 

1495
01:20:05,000 --> 01:20:07,120
Yeah, exactly. 
Just, I mean, productivity would

1496
01:20:07,120 --> 01:20:09,520
go up massively, right? 
So there's that offset from an 

1497
01:20:09,520 --> 01:20:11,280
economic growth perspective. 
Yeah. 

1498
01:20:11,280 --> 01:20:13,640
Like, you know, everyone would 
be figuring out how to do like 

1499
01:20:13,640 --> 01:20:17,440
3-4 jobs and using AI agents to 
replace themselves. 

1500
01:20:17,960 --> 01:20:20,720
So for that reason though, I 
think like it's going to be a 

1501
01:20:20,720 --> 01:20:23,440
slower rate of adoption. 
But I do think that the core of 

1502
01:20:23,440 --> 01:20:31,240
it's still the same in the sense
that a lot of jobs today will go

1503
01:20:31,240 --> 01:20:35,800
wet just like they've gone away 
in past technology cycles. 

1504
01:20:35,880 --> 01:20:37,680
Now. 
I think what makes this a little

1505
01:20:37,680 --> 01:20:40,960
bit different is that, and you, 
and you called this out too, 

1506
01:20:40,960 --> 01:20:44,720
it's like the speed at by which 
everything's happening is a lot 

1507
01:20:44,720 --> 01:20:47,880
different. 
Oh, as technology moves faster 

1508
01:20:47,880 --> 01:20:53,680
and, and it disrupts faster, you
still have the same impact on 

1509
01:20:53,680 --> 01:20:57,120
unemployment, right? 
In the sense of companies 

1510
01:20:57,120 --> 01:20:59,480
deciding that they can automate 
things and they're trying to 

1511
01:20:59,480 --> 01:21:02,400
drive margins in the slower 
growth environment and they're 

1512
01:21:02,400 --> 01:21:05,080
getting efficiency gains. 
And therefore, instead of 10 

1513
01:21:05,080 --> 01:21:07,480
engineers, they only need 2 
engineers and they're going to 

1514
01:21:07,480 --> 01:21:09,840
pay, you know, those two 
engineers may be a little bit 

1515
01:21:09,840 --> 01:21:13,040
more, but maybe they keep 5, 
right, and then pay all 5 less. 

1516
01:21:13,040 --> 01:21:16,320
Like there's going to be kind of
a deflationary shock that comes 

1517
01:21:16,320 --> 01:21:18,360
from all this. 
In the long run, it's good for 

1518
01:21:18,360 --> 01:21:22,240
corporate margins. 
But what I think the problem is 

1519
01:21:22,240 --> 01:21:25,840
that since the technology is 
moving so quickly, the ability 

1520
01:21:25,840 --> 01:21:29,440
to retrain people is going. 
I don't think that is 

1521
01:21:29,440 --> 01:21:35,520
necessarily going to speed up as
quickly, right as the rate of AI

1522
01:21:35,520 --> 01:21:38,000
adoption and the impact on 
business models and layoffs 

1523
01:21:38,000 --> 01:21:40,120
happened. 
So for that sense, you could 

1524
01:21:40,200 --> 01:21:42,560
have kind of have like an asset 
liability mismatch. 

1525
01:21:43,080 --> 01:21:44,600
Yeah, maybe one way to think 
about it. 

1526
01:21:44,920 --> 01:21:47,000
Yeah. 
Whereas if someone was like. 

1527
01:21:47,000 --> 01:21:49,840
Call it a societal asset 
liability mismatch. 

1528
01:21:49,840 --> 01:21:52,520
Exactly, yeah. 
And you know, the risk is that 

1529
01:21:52,720 --> 01:21:55,200
if you do too much of that, do 
you just risk? 

1530
01:21:55,200 --> 01:21:57,200
Man, the politics could get 
crazy. 

1531
01:21:57,200 --> 01:22:00,400
Yeah, crazier or unions like, I 
mean, that's why it's like, you 

1532
01:22:00,400 --> 01:22:02,160
know, it's like the concept of 
like AI, right? 

1533
01:22:02,160 --> 01:22:03,880
It's isn't like it's 
interesting, but it's also not 

1534
01:22:03,880 --> 01:22:05,240
crazy, right? 
Yeah. 

1535
01:22:05,600 --> 01:22:09,000
You know, this idea that you're 
going to have, you know, way 

1536
01:22:09,000 --> 01:22:11,880
more regulation around where AI 
can be embedded and what it 

1537
01:22:11,880 --> 01:22:13,720
can't be and who can actually 
make decisions. 

1538
01:22:13,720 --> 01:22:16,560
It's kind of not, It's, it's 
like, I mean the unions with 

1539
01:22:16,560 --> 01:22:18,680
this, one of the things I've 
learned researching like the 

1540
01:22:18,680 --> 01:22:22,640
port strike that happened more 
recently is that a lot of the a 

1541
01:22:22,800 --> 01:22:25,680
lot of those crane operators 
that are taking the container 

1542
01:22:25,680 --> 01:22:30,720
ships from the boats as part of 
their contracts, They actually 

1543
01:22:30,720 --> 01:22:36,080
have to like a robot can't do 
basically everything, like take 

1544
01:22:36,080 --> 01:22:38,480
it off whole thing can be 
automated, but someone has to be

1545
01:22:38,480 --> 01:22:40,720
in the seat. 
And then the person operating 

1546
01:22:40,720 --> 01:22:43,920
the crane has to actually, per 
terms of the union contract, 

1547
01:22:45,320 --> 01:22:50,000
annually lower the lower the 
container for the last 10 feet. 

1548
01:22:51,240 --> 01:22:54,760
So maybe that's like something 
similar. 

1549
01:22:54,800 --> 01:22:58,600
I mean, you can also like, 
really go dystopian and think 

1550
01:22:58,600 --> 01:23:01,560
about it in terms of like, does 
this open up the potential for 

1551
01:23:01,560 --> 01:23:04,400
like, universal basic income? 
To the. 

1552
01:23:04,520 --> 01:23:07,640
Extent that, you know, the 
economy and all these functions 

1553
01:23:07,640 --> 01:23:11,600
are going to be able to be 
operated using AI, but you know,

1554
01:23:11,760 --> 01:23:16,160
and, and it's going to be so 
accretive to corporate profit 

1555
01:23:16,160 --> 01:23:21,920
margins, but at some point lose 
the demand function, right? 

1556
01:23:21,920 --> 01:23:25,160
The demand works against you 
because AI doesn't consume in a 

1557
01:23:25,160 --> 01:23:27,720
similar way. 
And so in order to sustain 

1558
01:23:27,720 --> 01:23:30,960
levels of demand, you need to 
have some type of form of like 

1559
01:23:31,200 --> 01:23:34,320
universal basic income, which 
probably comes in the form of 

1560
01:23:34,320 --> 01:23:37,400
like a higher tax on corporate 
profits because the share of 

1561
01:23:37,640 --> 01:23:41,120
corporate profits as percentage 
GDP goes so high because 

1562
01:23:41,360 --> 01:23:43,560
productivity is through the 
roof, right. 

1563
01:23:43,680 --> 01:23:47,840
But even though labor demands 
down and so eventually it kind 

1564
01:23:47,840 --> 01:23:51,520
of comes full circle and you 
have to close the loop and 

1565
01:23:51,520 --> 01:23:54,720
basically give people money so 
they can go out and maintain 

1566
01:23:54,720 --> 01:23:58,320
their lifestyle and also buy 
things to keep everything going.

1567
01:23:58,400 --> 01:24:01,520
Again, kind of dystopian, but 
also not crazy. 

1568
01:24:01,600 --> 01:24:04,440
But there will be like a Charlie
and the Chocolate Factory, the 

1569
01:24:04,440 --> 01:24:07,560
Johnny Depp version part of this
too, where Charlie's dad's 

1570
01:24:07,560 --> 01:24:09,080
working at the Chocolate 
Factory. 

1571
01:24:09,560 --> 01:24:12,520
I'm sorry, working at a factory,
putting squirt on the tops of 

1572
01:24:12,520 --> 01:24:14,880
toothpaste, right? 
And then all of a sudden machine

1573
01:24:14,880 --> 01:24:19,000
comes in to replace him, he gets
laid off and then at the end of 

1574
01:24:19,000 --> 01:24:22,000
the movie, he comes back, gets 
retrained as like a robot 

1575
01:24:22,000 --> 01:24:24,520
operator, right? 
Fix the robots. 

1576
01:24:24,520 --> 01:24:26,720
Like there's going to be, you 
know, an element of that for 

1577
01:24:26,720 --> 01:24:29,080
sure. 
But I think this concept of, you

1578
01:24:29,080 --> 01:24:32,960
know, labor to capital shifting 
and is, is very, very real. 

1579
01:24:33,000 --> 01:24:35,880
And it's it's going to, you 
know, impact. 

1580
01:24:36,000 --> 01:24:39,240
Absolutely. 
Every part of the economy, yeah.

1581
01:24:39,240 --> 01:24:41,800
I don't think many jobs won't be
some more. 

1582
01:24:42,160 --> 01:24:45,000
And I do think it will probably 
be resulting structurally higher

1583
01:24:45,000 --> 01:24:48,440
unemployment because of that. 
And as soon as we have like the 

1584
01:24:48,440 --> 01:24:52,480
worst thing that could happen is
let's say like, let's say in 

1585
01:24:52,480 --> 01:24:54,840
like a couple years when this 
technology is like really 

1586
01:24:54,840 --> 01:24:58,240
pervasive is if you have like a 
recession or some type of event 

1587
01:24:58,240 --> 01:25:02,520
that causes corporations to 
cyclically lay people off 

1588
01:25:03,240 --> 01:25:05,320
because the. 
Then they maybe lean into the 

1589
01:25:05,320 --> 01:25:07,480
efficiency and then the jobs 
don't come back. 

1590
01:25:07,560 --> 01:25:08,880
Exactly. 
That's exactly right. 

1591
01:25:09,960 --> 01:25:14,920
So yeah, man, it's, I think the,
the core of it is like, and my 

1592
01:25:14,920 --> 01:25:18,920
approach is like, you just have 
to like, you can't dismiss it. 

1593
01:25:19,600 --> 01:25:24,320
It's real, it's coming, it's 
scary, it's threatening, it's 

1594
01:25:24,320 --> 01:25:27,200
exciting. 
It's all these things you 

1595
01:25:27,280 --> 01:25:31,400
depending on where you are on 
your career and what you're, you

1596
01:25:31,400 --> 01:25:34,760
know, like what your net worth 
is, right, and like what your 

1597
01:25:34,760 --> 01:25:38,120
incentives are, you're going to 
be more your your priorities are

1598
01:25:38,120 --> 01:25:40,760
probably going to be different, 
right, in many ways. 

1599
01:25:41,040 --> 01:25:44,040
And but if I was like, you know,
I think the biggest thing is 

1600
01:25:44,040 --> 01:25:47,280
like, and I see this happening 
with like a lot of analysts that

1601
01:25:47,280 --> 01:25:49,280
are younger coming out of 
college, like an existential 

1602
01:25:49,280 --> 01:25:53,440
crisis around knowledge workers.
And what's the value of 

1603
01:25:53,440 --> 01:25:55,840
somebody, a four year education 
if somebody is coming out of 

1604
01:25:55,840 --> 01:25:57,840
school, coming an analyst, 
right? 

1605
01:25:57,840 --> 01:26:03,520
If I 20 dollar $200 a month 
subscription to a chat AI 

1606
01:26:03,520 --> 01:26:06,640
service is so much more 
efficient, so much better. 

1607
01:26:06,640 --> 01:26:09,800
And it is like, I mean, I 
honestly, it is like, I can't, 

1608
01:26:10,280 --> 01:26:13,240
it's, it's just the reality. 
You can't dismiss it that 

1609
01:26:13,280 --> 01:26:15,000
there's margin for error like 
everything else. 

1610
01:26:15,000 --> 01:26:20,280
But I've been able to substitute
like, you know, 5060% of a lot 

1611
01:26:20,280 --> 01:26:23,040
of the work, you know, to AI 
directly. 

1612
01:26:23,120 --> 01:26:26,760
Even I created a, you know, an 
application model that can 

1613
01:26:26,760 --> 01:26:28,680
create slide decks for us, 
right? 

1614
01:26:28,680 --> 01:26:31,720
Something that on on our 
knowledge base of analysis that 

1615
01:26:31,720 --> 01:26:35,680
would something that would take.
God slide decks take forever. 

1616
01:26:35,680 --> 01:26:39,640
I probably built 10,000 slides 
like forever and now I can 

1617
01:26:39,640 --> 01:26:43,080
literally give it point to my 
knowledge base, which is 

1618
01:26:43,080 --> 01:26:45,480
basically a synthesis of all of 
our research that we've done. 

1619
01:26:46,000 --> 01:26:49,400
And I can't say, and it can 
create an outline and I say 

1620
01:26:49,400 --> 01:26:52,720
create this slide and it's not 
perfect, but I make edits to it,

1621
01:26:53,240 --> 01:26:55,680
Yeah. 
And it saves me a ton of time. 

1622
01:26:55,680 --> 01:26:58,000
And so I think. 
And if you don't have to worry 

1623
01:26:58,000 --> 01:27:00,960
about like do the numbers tie 
and you just know they do and 

1624
01:27:00,960 --> 01:27:02,280
you can trust it, that's. 
Nice. 

1625
01:27:02,280 --> 01:27:05,360
I mean, like, you have to still 
give it like conclusions, but 

1626
01:27:05,400 --> 01:27:07,840
you know, we'll get there 
eventually where it can do a lot

1627
01:27:07,840 --> 01:27:11,080
of this, like some of these 
other things as well, like more.

1628
01:27:11,080 --> 01:27:14,640
Well, I just mean as as easy as 
like, you know, you get typos 

1629
01:27:14,640 --> 01:27:17,040
and stuff when you do things 
manually to the extent that 

1630
01:27:17,040 --> 01:27:20,520
something and, and, and those 
kind of those things, when the 

1631
01:27:20,520 --> 01:27:23,040
numbers don't tie, that's the 
stuff that undermines your 

1632
01:27:23,040 --> 01:27:25,040
fundamental like everything 
else, right? 

1633
01:27:25,040 --> 01:27:27,880
That's the stuff people, at 
least in my career, it is. 

1634
01:27:27,880 --> 01:27:30,520
That's what people notice and 
then you start getting. 

1635
01:27:30,520 --> 01:27:34,040
So if you don't have to worry 
about that, it frees your mind 

1636
01:27:34,040 --> 01:27:37,360
up to be creative in other ways.
Yeah, I think the the biggest 

1637
01:27:37,360 --> 01:27:41,320
part, the hardest part of this 
all is that you and I got to 

1638
01:27:41,320 --> 01:27:43,960
where we are. 
Like we're sitting here and I 

1639
01:27:43,960 --> 01:27:47,560
spent 10 years hunting and 
pecking, reading every single 

1640
01:27:47,560 --> 01:27:52,280
filing, building models from 
scratch, like all the, all the 

1641
01:27:52,280 --> 01:27:56,520
learnings that come that 
associated with that, that are 

1642
01:27:56,640 --> 01:28:00,720
very, very like beneficial. 
Because for me. 

1643
01:28:00,720 --> 01:28:03,680
And then like with AI, it's like
you already know this stuff. 

1644
01:28:04,240 --> 01:28:06,400
You know what kind of questions 
to ask, you know what to look 

1645
01:28:06,400 --> 01:28:08,120
for. 
You can ask better questions 

1646
01:28:08,120 --> 01:28:10,960
because you've gone through that
whole kind of experience. 

1647
01:28:11,120 --> 01:28:13,560
You have experience, right? 
So you can just ask better 

1648
01:28:13,560 --> 01:28:16,080
questions, tighter questions, 
you can learn how to think 

1649
01:28:16,080 --> 01:28:17,400
critically. 
You've learned how to think 

1650
01:28:17,400 --> 01:28:19,840
critically more because you've 
made mistakes. 

1651
01:28:19,840 --> 01:28:24,200
That's the biggest issue that I 
see with AI, especially if 

1652
01:28:24,200 --> 01:28:27,520
you're like, you know, younger 
analysts, is that on the one 

1653
01:28:27,520 --> 01:28:30,000
hand, I want to say I've been 
telling people and I agree. 

1654
01:28:30,000 --> 01:28:32,320
And I, it's like you have to 
figure out how to use this to 

1655
01:28:32,320 --> 01:28:34,280
make yourself more efficient, 
right? 

1656
01:28:34,280 --> 01:28:36,960
Because this is the future. 
But on the other hand, there's 

1657
01:28:36,960 --> 01:28:41,760
like this paradox of you have 
to, you know, learn how to 

1658
01:28:41,760 --> 01:28:44,400
become more efficient. 
Like you have to learn how to 

1659
01:28:44,400 --> 01:28:45,840
learn. 
You have to learn how to ask the

1660
01:28:45,840 --> 01:28:47,520
right questions. 
You have to learn how to think 

1661
01:28:47,520 --> 01:28:49,480
critically. 
It's not something that's net 

1662
01:28:49,480 --> 01:28:52,400
doesn't necessarily come natural
to some people that you can't, 

1663
01:28:52,440 --> 01:28:55,520
you know, because otherwise, you
know, just like, just like if a 

1664
01:28:55,520 --> 01:28:57,760
junior analyst goes on, does 
research and gives you back 

1665
01:28:57,760 --> 01:29:00,480
work, you look at it, you're 
like, you know, that's wrong, 

1666
01:29:00,480 --> 01:29:01,800
that's wrong, that's wrong, 
that's wrong. 

1667
01:29:02,120 --> 01:29:04,120
Or you look at a slide deck, you
look at it for 2 seconds, right,

1668
01:29:04,120 --> 01:29:07,600
That's wrong. 
And the analyst is like, how did

1669
01:29:07,600 --> 01:29:09,960
you figure that out? 
How did you see that so fast? 

1670
01:29:10,680 --> 01:29:12,280
It's like, well, it's not 
because AI, it's just because 

1671
01:29:12,280 --> 01:29:14,240
it's this experience. 
I've done this a million times, 

1672
01:29:14,680 --> 01:29:16,400
right? 
And that's, you know, I think 

1673
01:29:16,400 --> 01:29:20,400
that's that element is still 
really important and AI can't 

1674
01:29:20,400 --> 01:29:22,400
solve for that yet. 
But but The thing is like, if 

1675
01:29:22,400 --> 01:29:24,560
you the whole fine tuning 
element that I mentioned before,

1676
01:29:24,560 --> 01:29:28,720
because we're not at that stage,
like in theory, you can actually

1677
01:29:28,720 --> 01:29:31,960
fine tune a model for just that,
right? 

1678
01:29:31,960 --> 01:29:33,640
So like pick up on those 
nuances. 

1679
01:29:34,080 --> 01:29:37,200
But for younger people, it's a 
it is a it is stuff, right? 

1680
01:29:37,200 --> 01:29:40,840
Because you're competing against
AI in many ways, there's going 

1681
01:29:40,840 --> 01:29:44,080
to be just less demand for labor
because of it, like fewer new 

1682
01:29:44,080 --> 01:29:47,240
job openings, because so much of
the basic grunt work is, can 

1683
01:29:47,240 --> 01:29:52,000
just be, you know, replaced. 
So it's kind of, they're going 

1684
01:29:52,000 --> 01:29:56,080
to have to use AI to accelerate 
their learning curve, right? 

1685
01:29:56,240 --> 01:29:58,480
Like I said, everything's moving
so much faster. 

1686
01:29:58,480 --> 01:30:02,240
So you're going to have to have 
to ask questions, you're going 

1687
01:30:02,240 --> 01:30:04,760
to be have to be intellectually 
curious and you're going to have

1688
01:30:04,760 --> 01:30:08,720
to use AI to make you learn 
skills that you previously 

1689
01:30:08,720 --> 01:30:10,720
couldn't learn. 
You got there's you're going to 

1690
01:30:10,720 --> 01:30:13,480
have to create roles for 
yourself that didn't previously 

1691
01:30:13,480 --> 01:30:17,720
exist. 
And that is something that I'm 

1692
01:30:17,720 --> 01:30:21,880
not even sure like our 
educational system is readily 

1693
01:30:21,880 --> 01:30:25,120
equipped to adapt because while 
the Internet was disruptive in 

1694
01:30:25,120 --> 01:30:30,200
the sense of it was an 
information, you know, of that 

1695
01:30:30,200 --> 01:30:32,840
like information, just the rate 
of information flow just broke 

1696
01:30:32,840 --> 01:30:35,560
open right now. 
But you can still anchor on like

1697
01:30:35,680 --> 01:30:38,480
analytical rigor and critical 
like learning skills. 

1698
01:30:38,520 --> 01:30:41,720
You know, now it's AI because 
it's going to be prevalent 

1699
01:30:41,720 --> 01:30:43,840
everywhere. 
I think a lot of the biases that

1700
01:30:43,840 --> 01:30:46,680
we talked about in terms of this
preservation mindset and come 

1701
01:30:46,680 --> 01:30:50,280
and see mindset versus abundance
mindset is going to be endemic 

1702
01:30:50,280 --> 01:30:54,200
to a lot of institutions that 
they're going to view it equally

1703
01:30:54,200 --> 01:30:56,160
as a threat. 
Professors are going to view it 

1704
01:30:56,240 --> 01:30:59,000
as a threat, right? 
And therefore they're not going 

1705
01:30:59,000 --> 01:31:03,800
to teach people how to use the 
tools, right to actually, 

1706
01:31:03,800 --> 01:31:06,960
because you can use AI to help 
you with critical thinking to 

1707
01:31:06,960 --> 01:31:08,200
solve problems. 
You just. 

1708
01:31:08,200 --> 01:31:10,920
Yeah, I was, I was thinking 
since you've been saying this 

1709
01:31:10,920 --> 01:31:13,520
and my wife said this, so I 
should give her credit for it. 

1710
01:31:13,920 --> 01:31:19,680
But the the way I think that 
effective teaching may occur is,

1711
01:31:19,680 --> 01:31:22,840
is more is closer to the law 
school experience where you're 

1712
01:31:22,840 --> 01:31:26,000
teaching people how to think and
asking like, like more Socratic 

1713
01:31:26,000 --> 01:31:30,280
method type stuff, I think could
prove a good way to educate 

1714
01:31:30,280 --> 01:31:32,520
people. 
Because working on people's 

1715
01:31:32,520 --> 01:31:35,160
logic and working on people's 
ability, I mean, you said it 

1716
01:31:35,160 --> 01:31:37,800
earlier, your ability to ask 
good questions is going to be a 

1717
01:31:37,800 --> 01:31:40,200
good or a key differentiator, 
right. 

1718
01:31:40,200 --> 01:31:44,240
So I think maybe that's, that's 
how the professors that that 

1719
01:31:44,240 --> 01:31:47,360
want to do a service, that's 
that's how maybe they can lean 

1720
01:31:47,360 --> 01:31:48,960
into their role. 
Yeah. 

1721
01:31:48,960 --> 01:31:51,360
Because, yeah, I mean, like, I 
don't know, obviously there's 

1722
01:31:51,360 --> 01:31:53,240
still going to be a need for 
research professors. 

1723
01:31:53,240 --> 01:31:56,360
I don't want you know, but a lot
of the research might be able to

1724
01:31:56,360 --> 01:32:00,520
be outsourced. 
Yeah, and, and, and then, you 

1725
01:32:00,520 --> 01:32:04,640
know, it's like your ability to 
learn, think it's infinite. 

1726
01:32:04,640 --> 01:32:07,400
It's everything that you could 
possibly really want to know at 

1727
01:32:07,400 --> 01:32:09,800
your fingertips, right. 
I mean, the danger of like we 

1728
01:32:09,880 --> 01:32:11,560
talked about before is just 
verification. 

1729
01:32:11,560 --> 01:32:16,160
Like, are you, you know, 
equipped and able to call BS if 

1730
01:32:16,160 --> 01:32:19,040
it's wrong, right. 
Like that's a thing. 

1731
01:32:19,040 --> 01:32:22,360
But in theory, you know, if you 
anchor in Google search, if you 

1732
01:32:22,360 --> 01:32:25,880
anchor in other things or if you
create agents that are basically

1733
01:32:25,880 --> 01:32:29,360
like designed like I once had a,
one of my best meetings that 

1734
01:32:29,360 --> 01:32:33,360
I've ever had was it was a 
little awkward at first, but it 

1735
01:32:33,360 --> 01:32:34,400
ended up being a really good 
meetings. 

1736
01:32:34,400 --> 01:32:37,480
It's really seasoned PM like 
former like one of the biggest 

1737
01:32:37,480 --> 01:32:41,400
hedge funds out there and follow
our work. 

1738
01:32:41,440 --> 01:32:44,920
They were we, we came 
independently to a different 

1739
01:32:44,920 --> 01:32:50,280
research conclusion on a stock 
and the PM like set up a meeting

1740
01:32:50,720 --> 01:32:55,000
and it was like his analyst who 
I've talked to a lot and me and 

1741
01:32:55,000 --> 01:32:59,160
he's like you're saying this, 
you're saying that, but you're 

1742
01:32:59,160 --> 01:33:01,080
both are looking at the same 
data. 

1743
01:33:02,520 --> 01:33:05,480
Yeah. 
Figure it out, right? 

1744
01:33:05,920 --> 01:33:08,720
And it was like a very interest.
It was a great it was. 

1745
01:33:08,720 --> 01:33:10,840
I mean, it was definitely a 
little terrifying and 

1746
01:33:10,840 --> 01:33:13,520
intimidating, right? 
Yeah. 

1747
01:33:14,400 --> 01:33:19,040
But I kind of do the same thing 
with like a lot of AI is that I 

1748
01:33:19,040 --> 01:33:23,400
will train agents or I'll put 
the different models against 

1749
01:33:23,400 --> 01:33:25,320
each other and say like, look, 
you're saying this, you're 

1750
01:33:25,320 --> 01:33:28,120
saying that or you know, what do
you think? 

1751
01:33:28,520 --> 01:33:31,280
Like what are the pros and cons?
You basically do the same type 

1752
01:33:31,280 --> 01:33:34,480
of optimization, just an 
optimization and you have them 

1753
01:33:34,480 --> 01:33:37,760
break it down in steps and you 
can get to, you know, at a 

1754
01:33:37,760 --> 01:33:40,640
minimum what it does is it for 
me at least, it helps me. 

1755
01:33:40,640 --> 01:33:44,080
It expands like my capacity for 
analysis because I have all 

1756
01:33:44,080 --> 01:33:46,360
these questions in my head all 
the time, right? 

1757
01:33:46,440 --> 01:33:50,240
But I, I can't, my ability to 
execute all of it is not there. 

1758
01:33:50,480 --> 01:33:52,640
I do have a learning disability 
that I've struggled with my 

1759
01:33:52,640 --> 01:33:55,240
entire life to help. 
It makes writing harder, makes 

1760
01:33:55,400 --> 01:33:58,680
kind of managing things harder. 
And AI has been incredibly 

1761
01:33:58,680 --> 01:34:01,560
helpful in that because it's 
just like an ability to it. 

1762
01:34:01,600 --> 01:34:05,720
It actually, it can solve for a 
lot of that in terms of writing,

1763
01:34:05,720 --> 01:34:10,560
synthesizing information. 
And so that's I think that's 

1764
01:34:10,600 --> 01:34:14,800
probably and, and for me, that's
been a, a, a huge value add and 

1765
01:34:14,800 --> 01:34:17,440
that optimization piece, I think
is something that maybe people 

1766
01:34:17,440 --> 01:34:20,080
aren't leaning to as much. 
Again, this isn't just like 

1767
01:34:20,080 --> 01:34:22,560
summary, right? 
This isn't just like, hey, 

1768
01:34:22,560 --> 01:34:24,880
summarize this, help me write an
e-mail faster. 

1769
01:34:25,840 --> 01:34:28,200
A lot of the applications now 
and a lot of people are using 

1770
01:34:28,200 --> 01:34:30,280
the free models, like the free 
versions. 

1771
01:34:30,680 --> 01:34:33,080
It's like you can't, like you 
just got to go AI first. 

1772
01:34:33,120 --> 01:34:36,760
All in experiment with the 
different models. 

1773
01:34:36,760 --> 01:34:39,600
Ask the same question to 
different models, even within 

1774
01:34:39,600 --> 01:34:40,120
the. 
Same. 

1775
01:34:40,120 --> 01:34:43,400
I find that very interesting and
then comparing the answers and 

1776
01:34:43,400 --> 01:34:45,960
trying to figure out. 
I like to read the reasoning 

1777
01:34:45,960 --> 01:34:48,880
steps too and and see you know 
how are they thinking about 

1778
01:34:48,880 --> 01:34:49,960
things. 
Totally. 

1779
01:34:50,080 --> 01:34:53,360
And, and it, it makes me think 
harder, right? 

1780
01:34:53,400 --> 01:34:55,920
It makes me think differently. 
And I'm OK if it frames my 

1781
01:34:55,920 --> 01:34:58,960
thinking differently, if it's 
influencing me because I just at

1782
01:34:58,960 --> 01:35:01,240
the end of the day, I just know 
that this is where the world's 

1783
01:35:01,240 --> 01:35:03,600
going. 
Like it is like it's, this is 

1784
01:35:03,600 --> 01:35:06,040
where it's going. 
So it's either get on board and 

1785
01:35:06,040 --> 01:35:09,760
you're either going to ride the 
waves or you're going to be, you

1786
01:35:09,760 --> 01:35:15,320
know, run over by it. 
And I fully intend to stay on 

1787
01:35:15,320 --> 01:35:17,720
top of the waves, not fall off 
the surfboard. 

1788
01:35:18,960 --> 01:35:21,080
Well, I appreciate you coming on
and talking about it. 

1789
01:35:21,160 --> 01:35:24,680
It's funny, my dad asked me at 
dinner just this past week what 

1790
01:35:24,680 --> 01:35:29,160
I think about AI And I, I gave 
him about a 10 minute discussion

1791
01:35:29,160 --> 01:35:30,960
and I'm just going to send him 
this and I'm going to be like, I

1792
01:35:30,960 --> 01:35:34,000
think you should listen to this.
And I I think he will enjoy it 

1793
01:35:34,000 --> 01:35:35,880
very much. 
I appreciate that and thanks. 

1794
01:35:35,880 --> 01:35:38,000
Yeah, thanks for, I mean for 
reaching out and we should 

1795
01:35:38,000 --> 01:35:41,360
definitely do more follow-ups 
like as things progress because 

1796
01:35:41,360 --> 01:35:43,440
it's moving so much fat, it's 
moving so quickly, right. 

1797
01:35:43,440 --> 01:35:45,560
I know you said earlier that 
you're glad you didn't do any 

1798
01:35:45,560 --> 01:35:48,120
recordings in last month because
it would proven a lot of things 

1799
01:35:48,120 --> 01:35:49,960
wrong. 
And but look, I'm totally 

1800
01:35:49,960 --> 01:35:52,880
comfortable and OK with the fact
that maybe we'll look back six 

1801
01:35:52,880 --> 01:35:55,240
months a year from now, two 
months from now, and everything 

1802
01:35:55,240 --> 01:35:57,280
has come out of my mouth and 
everything just talked about 

1803
01:35:57,280 --> 01:35:59,960
just just proven utterly and 
totally wrong. 

1804
01:36:00,040 --> 01:36:02,360
And that's all this. 
This is different. 

1805
01:36:02,720 --> 01:36:07,760
So I, I think one of the issues 
as, as a podcast at this stage 

1806
01:36:07,760 --> 01:36:12,000
is like I've noticed with my own
listening, I don't listen to 

1807
01:36:12,000 --> 01:36:15,440
everybody's episodes anymore. 
There's so many podcasts out 

1808
01:36:15,440 --> 01:36:19,080
there that it's like I almost 
don't even there are some that I

1809
01:36:19,080 --> 01:36:22,800
want a weekly, you know, sort of
like something put in my head. 

1810
01:36:23,520 --> 01:36:27,120
My commitment to my listeners, 
at least going forward is I'm 

1811
01:36:27,120 --> 01:36:29,440
going to try to make sure that 
if something's in their ear, 

1812
01:36:29,720 --> 01:36:32,440
it's because I want it to be and
it's not because I'm trying to 

1813
01:36:32,440 --> 01:36:37,360
Juke some weekly release. 
So Spotify recommends me, right?

1814
01:36:37,360 --> 01:36:43,240
And I just, there was so much 
macro news flow and I found it 

1815
01:36:43,240 --> 01:36:46,520
so interesting and Anna, it was 
moving the market so much. 

1816
01:36:46,520 --> 01:36:50,760
It was what I was focused on. 
So I, I don't know, man, I 

1817
01:36:50,760 --> 01:36:53,880
probably would have ended up 
chasing some sort of, you know, 

1818
01:36:53,880 --> 01:36:57,440
macro guru of the of the day. 
I mean, I had I had Andy 

1819
01:36:57,440 --> 01:37:00,320
constant on that was kind of 
macro E, but I'd followed him 

1820
01:37:00,320 --> 01:37:03,200
for a while and it wasn't it was
inspired by what was going on, 

1821
01:37:03,200 --> 01:37:05,920
but it wasn't chasing it. 
And I just I think it had I been

1822
01:37:05,920 --> 01:37:10,040
booking a lot, I just think I 
would have probably gone after 

1823
01:37:10,040 --> 01:37:12,840
the wrong content. 
So I'm glad I didn't, if that 

1824
01:37:12,840 --> 01:37:15,400
makes sense. 
I say plenty of stuff that ends 

1825
01:37:15,400 --> 01:37:17,560
up being wrong is not 
concerning. 

1826
01:37:17,560 --> 01:37:19,840
Yeah, that's why I like always 
when you reach out, it's always 

1827
01:37:19,840 --> 01:37:22,320
love coming on here because like
I can't, it's like feels 

1828
01:37:22,320 --> 01:37:24,520
authentic and just, you know, 
genuine. 

1829
01:37:24,520 --> 01:37:27,760
And that's that's important. 
And yeah, I think, you know, 

1830
01:37:27,760 --> 01:37:30,760
we're going to, you know, maybe 
maybe the next step at some 

1831
01:37:30,760 --> 01:37:33,520
point is like we talked on like 
kind of high level across the 

1832
01:37:33,520 --> 01:37:35,400
board and a lot of these 
different like AI topics. 

1833
01:37:35,400 --> 01:37:38,680
But yeah, I feel like a lot of 
them each and itself, like even 

1834
01:37:38,680 --> 01:37:41,840
like the Google discussion by 
itself could be like on search 

1835
01:37:41,840 --> 01:37:44,640
and what's happening could be 
like it's own separate things. 

1836
01:37:45,520 --> 01:37:47,880
Yeah, well, let's do a follow up
and like, I don't know, like 2 

1837
01:37:47,880 --> 01:37:49,720
months or so. 
Let's let people listen to this,

1838
01:37:49,720 --> 01:37:52,960
digest it and hopefully get, you
know, start playing with the 

1839
01:37:52,960 --> 01:37:55,560
tools and then come back on. 
Yeah, I I love talking. 

1840
01:37:55,720 --> 01:37:56,760
I. 
Think I think it's great 

1841
01:37:57,000 --> 01:37:59,520
listening to this like it's all 
one big experiment, right. 

1842
01:37:59,520 --> 01:38:02,680
So like if you have any 
questions or thoughts or 

1843
01:38:02,720 --> 01:38:04,920
anything that like as you're 
listening to this, any 

1844
01:38:04,920 --> 01:38:08,960
inspiration, any ideas? 
Any questions like don't keep 

1845
01:38:08,960 --> 01:38:10,640
them. 
I mean, you know, don't keep 

1846
01:38:10,640 --> 01:38:12,760
them to yourself. 
Like if you want like throw them

1847
01:38:12,760 --> 01:38:16,200
out there like on the X or send 
them in to Bill and we can talk 

1848
01:38:16,200 --> 01:38:19,360
about next time. 
Or you know, it's I think part 

1849
01:38:19,360 --> 01:38:23,800
of this is just that sharing 
learnings for all of us is 

1850
01:38:23,800 --> 01:38:27,320
something that in a world where 
informational advantages and 

1851
01:38:27,320 --> 01:38:31,520
edge is so closely held close to
the best, This is this is a 

1852
01:38:31,520 --> 01:38:35,400
technology, this is a phase that
we're all entering together that

1853
01:38:35,400 --> 01:38:39,320
I think we actually all benefit 
tremendously from sharing our 

1854
01:38:39,320 --> 01:38:42,560
experiences and our learnings. 
Because at the end of the day, 

1855
01:38:42,960 --> 01:38:44,560
you know, this does impact all 
of us. 

1856
01:38:44,560 --> 01:38:48,600
And that's foster community of 
intellectual curiosity related 

1857
01:38:48,600 --> 01:38:52,120
to that I think is really cool. 
And I know your audience is like

1858
01:38:52,240 --> 01:38:56,000
the perfect audience for that. 
And hopefully we get some good 

1859
01:38:56,000 --> 01:38:57,560
feedback and thoughts from it as
well. 

1860
01:38:58,120 --> 01:39:00,480
Indeed. 
Well, have a good one. 

1861
01:39:00,480 --> 01:39:02,200
I heard your kids screaming in 
the background. 

1862
01:39:02,200 --> 01:39:04,240
How's that going? 
It's going well, man. 

1863
01:39:04,280 --> 01:39:08,360
Like he's he's just the three 
and a half 18 months the two 

1864
01:39:08,360 --> 01:39:10,800
boys are beating the shit out of
each other like as they. 

1865
01:39:10,800 --> 01:39:12,600
Shocking to stop, man. 
Yeah, they're trying. 

1866
01:39:12,600 --> 01:39:15,600
They're trying to. 
Keeping them alive is proving to

1867
01:39:15,600 --> 01:39:19,000
be an interesting daily task in 
his first BMX biking race on the

1868
01:39:19,000 --> 01:39:19,960
starter bikes. 
The other. 

1869
01:39:20,040 --> 01:39:22,520
Day which? 
Is pretty cool and you know some

1870
01:39:22,520 --> 01:39:24,760
he's just a total. 
He's just a killer. 

1871
01:39:24,760 --> 01:39:27,760
He's just like running around 
getting into everything. 

1872
01:39:28,640 --> 01:39:30,480
But dude, they're happy, man. 
And that's like the only thing 

1873
01:39:30,480 --> 01:39:35,760
that matters, right? 
Like honestly, it's that's my 

1874
01:39:35,760 --> 01:39:37,920
goal, right? 
My goal is to make sure that 

1875
01:39:38,720 --> 01:39:43,560
they're just happy and have a 
fulfilled life and being present

1876
01:39:43,560 --> 01:39:46,000
as a father or something. 
That's really top of mind for me

1877
01:39:46,000 --> 01:39:47,240
too. 
And it's hard to do. 

1878
01:39:47,240 --> 01:39:52,960
But I don't think any, I don't 
think any type of AI is going to

1879
01:39:53,560 --> 01:39:56,920
replace that any type any. 
No, I don't think so either. 

1880
01:39:57,280 --> 01:40:01,640
The mine are, are now, you know,
getting older and starting to 

1881
01:40:01,640 --> 01:40:05,080
think about teenage years. 
And one of the things that I've 

1882
01:40:05,080 --> 01:40:09,000
been pushing myself to do is 
make sure that, you know, I'm 

1883
01:40:09,000 --> 01:40:12,080
leading by example and, and 
making sure that it's not just 

1884
01:40:12,080 --> 01:40:15,240
talk because, you know, they, 
they watch and they learn from 

1885
01:40:15,240 --> 01:40:18,680
from who we are. 
And I don't know, it's been, I 

1886
01:40:18,680 --> 01:40:22,200
want to, to the point of this 
whole conversation, right? 

1887
01:40:22,200 --> 01:40:26,640
I I want to show them. 
A father that's able to adapt 

1888
01:40:26,680 --> 01:40:30,600
and able to embrace new things 
and not be so scared. 

1889
01:40:30,600 --> 01:40:32,680
But it has been a little bit 
difficult, man. 

1890
01:40:32,680 --> 01:40:35,800
It's been, it's been an 
interesting six months or a 

1891
01:40:35,800 --> 01:40:38,400
year, but I'm mostly through it.
We'll see. 

1892
01:40:38,560 --> 01:40:41,120
Yeah, you know, it's life's a 
journey. 

1893
01:40:41,760 --> 01:40:45,640
No doubt, no doubt. 
All right, my man. 

1894
01:40:45,840 --> 01:40:47,760
We'll talk soon. 
I look forward to the follow. 

1895
01:40:47,760 --> 01:40:48,240
Up. 
Yeah. 

1896
01:40:48,320 --> 01:40:49,240
Thanks, Bill. 
Appreciate. 

1897
01:40:49,240 --> 01:41:27,180
It all right, take care. 
None. 

1898
01:41:53,440 --> 01:41:53,520
The.