1
00:00:00,240 --> 00:00:02,800
Secure your Elixir apps with 
paraxial IO. 

2
00:00:03,000 --> 00:00:06,040
From code analysis and bot 
defense to dependency checks, 

3
00:00:06,040 --> 00:00:09,280
they've got you covered. 
Elixir first design with GitHub 

4
00:00:09,280 --> 00:00:11,680
integration. 
Elevate your security at 

5
00:00:11,680 --> 00:00:17,000
paraxial dot IO. 
Hey, I'm Kimberly with Elixir 

6
00:00:17,000 --> 00:00:20,520
mentor Jacob recently added 
recruitment services to help 

7
00:00:20,520 --> 00:00:22,720
companies hire pre vetted Elixir
devs. 

8
00:00:23,280 --> 00:00:26,480
I help with company outreach, so
if you're hiring, I may be 

9
00:00:26,480 --> 00:00:29,480
reaching out, or you can connect
with us anytime at 

10
00:00:29,480 --> 00:00:38,560
elixirmentor.com. 
Testing what's up? 

11
00:00:38,560 --> 00:00:41,480
I'm Jacob Litzow, your Elixir 
mentor and welcome to another 

12
00:00:41,480 --> 00:00:43,840
exciting episode of the Elixir 
Mentor Podcast. 

13
00:00:44,080 --> 00:00:46,840
This is where we discuss 
everything related to Elixir, 

14
00:00:46,840 --> 00:00:48,960
from interviews with 
enthusiasts, pioneers in the 

15
00:00:48,960 --> 00:00:52,680
community to innovative projects
and libraries shaping Elixir's 

16
00:00:52,680 --> 00:00:55,000
future. 
Today I'm joined by George, 

17
00:00:55,000 --> 00:00:56,960
creator of Learn Phoenix live 
view. 

18
00:00:57,320 --> 00:01:00,960
He also went through the 
Gauntlet AI Fellowship and came 

19
00:01:00,960 --> 00:01:03,400
out rethinking how he builds 
software. 

20
00:01:04,239 --> 00:01:06,440
George, it's awesome having you 
today. 

21
00:01:06,440 --> 00:01:07,720
Thanks. 
Thanks for joining me. 

22
00:01:09,240 --> 00:01:11,480
Hi, yeah, thanks for having me. 
Great to be here. 

23
00:01:13,520 --> 00:01:18,520
I hear like a weird like rubbing
noise every once in a while. 

24
00:01:18,520 --> 00:01:26,960
I don't know if your mic is on a
desk or I mean sorry, sorry 

25
00:01:26,960 --> 00:01:28,600
actually. 
Then I had to respond to that 

26
00:01:28,600 --> 00:01:29,640
before. 
We went live. 

27
00:01:33,080 --> 00:01:34,640
Yeah, I mean. 
It should be fine. 

28
00:01:34,640 --> 00:01:36,040
Is the problem or are we good to
go? 

29
00:01:36,760 --> 00:01:38,080
We're good. 
I don't hear it right now. 

30
00:01:38,320 --> 00:01:39,600
Yeah. 
All right, cool. 

31
00:01:41,000 --> 00:01:47,600
So I, I had the the awesome. 
I don't know what I what's the, 

32
00:01:47,600 --> 00:01:50,720
what's the word I'm looking for?
I got to meet George down in 

33
00:01:50,720 --> 00:01:53,640
Austin at XMEX and that was a 
lot of fun. 

34
00:01:53,640 --> 00:01:56,200
That was a good conference. 
Hopefully you weren't one of the

35
00:01:56,200 --> 00:01:58,880
people that got sick because I 
feel like a handful of people 

36
00:01:58,960 --> 00:02:02,560
all got sick after Friday. 
No, not me. 

37
00:02:02,840 --> 00:02:04,840
Yeah, Neither did I. 
Yeah. 

38
00:02:05,720 --> 00:02:07,720
Yeah. 
So you made it through. 

39
00:02:07,720 --> 00:02:11,800
That's good. 
You want to just give people a 

40
00:02:11,800 --> 00:02:15,480
brief background of like who you
are, where you came from, and 

41
00:02:15,600 --> 00:02:17,040
kind of like where you are 
today. 

42
00:02:18,560 --> 00:02:21,200
Yeah, sure. 
So my name is George Millo. 

43
00:02:21,200 --> 00:02:24,400
I am best known in the Elixir 
community for, like you 

44
00:02:24,400 --> 00:02:27,680
mentioned, some courses that 
I've created teaching Elixir in 

45
00:02:27,680 --> 00:02:30,040
Phoenix. 
I used to do all of that under a

46
00:02:30,040 --> 00:02:33,400
pseudonym, a pen name that that 
name was George Arrowsmith. 

47
00:02:33,640 --> 00:02:37,800
I I decided about six months ago
I was going to drop the 

48
00:02:37,800 --> 00:02:41,280
pretense, just start putting my 
real name and face out there. 

49
00:02:41,600 --> 00:02:44,640
But, you know, some of your 
audience may still know me 

50
00:02:44,640 --> 00:02:48,400
better by that that fake name. 
Yeah. 

51
00:02:48,400 --> 00:02:52,240
So I've been a professional 
software engineer since 2013, so

52
00:02:52,480 --> 00:02:57,520
13 years now. 
Unlucky for some, my background,

53
00:02:57,520 --> 00:03:00,960
I got into Alexa through what I 
think is pretty common in the 

54
00:03:00,960 --> 00:03:02,120
Elixir world. 
I am. 

55
00:03:02,520 --> 00:03:05,160
I was a Ruby on Rails developer 
for quite some time. 

56
00:03:05,800 --> 00:03:08,720
Back at the very beginning I was
using PHPI got out of that as 

57
00:03:08,720 --> 00:03:11,400
soon as I could. 
Yeah, I worked with Rails for 

58
00:03:11,400 --> 00:03:15,360
about four or five years, got 
sick of it, not to knock on 

59
00:03:15,360 --> 00:03:18,640
Rails, but yeah, you know, after
just working with it for a long 

60
00:03:18,640 --> 00:03:20,960
time, I it's time for something 
new. 

61
00:03:20,960 --> 00:03:24,560
And there's a lot of cross 
pollination between the Ruby and

62
00:03:24,560 --> 00:03:28,280
Elixir communities, as I'm sure 
everyone watching this knows. 

63
00:03:29,720 --> 00:03:31,840
Played around with Elixir in 
Phoenix, Loved it. 

64
00:03:32,480 --> 00:03:34,360
Phoenix is a fantastic piece of 
technology. 

65
00:03:34,360 --> 00:03:36,880
I feel like it just took 
everything that was great about 

66
00:03:36,880 --> 00:03:40,680
Rails, everything that Rails did
well, and it shedded all the 

67
00:03:40,680 --> 00:03:42,560
things that maybe Rails didn't 
do so well. 

68
00:03:42,560 --> 00:03:47,360
Innovative, great. 
I got into the course business 

69
00:03:48,240 --> 00:03:51,480
partly just out of frustration 
from, you know, when I was 

70
00:03:51,480 --> 00:03:53,960
starting and trying to learn, 
there was a steep learning 

71
00:03:53,960 --> 00:03:56,000
curve. 
People say that Elixir is like 

72
00:03:56,000 --> 00:03:57,360
Ruby. 
It's really not that. 

73
00:03:57,360 --> 00:03:59,280
Similarities are very 
superficial. 

74
00:03:59,520 --> 00:04:02,240
If you've never done functional 
programming before, which I 

75
00:04:02,240 --> 00:04:04,720
hadn't. 
Yeah, There's a lot of things 

76
00:04:04,720 --> 00:04:08,920
you've got to unlearn a lot. 
And the materials that I, I, I 

77
00:04:08,920 --> 00:04:12,800
could find teach myself Elixir 
in Phoenix, I don't know. 

78
00:04:12,840 --> 00:04:14,960
They just weren't very good in 
my opinion, at least not the 

79
00:04:14,960 --> 00:04:18,240
ones that I could find. 
So really I just ended up, I 

80
00:04:18,240 --> 00:04:20,399
mean, it was back in the COVID 
lockdowns, had a lot of free 

81
00:04:20,399 --> 00:04:25,440
time as we all did. 
Yeah, I just started creating 

82
00:04:25,440 --> 00:04:28,360
the really the courses that I 
wish that I had had when I was 

83
00:04:28,480 --> 00:04:31,120
getting started. 
And yeah, they seem to have been

84
00:04:31,280 --> 00:04:33,520
pretty well received. 
It's never been my main gig. 

85
00:04:33,520 --> 00:04:35,520
It's just been an evenings and 
weekends kind of thing. 

86
00:04:36,640 --> 00:04:40,720
But yeah, I've enjoyed doing it 
and good things have come to me 

87
00:04:40,720 --> 00:04:43,320
from it. 
Yeah, that's awesome. 

88
00:04:43,320 --> 00:04:45,960
I feel like you and I had the 
same frustrations like because 

89
00:04:45,960 --> 00:04:47,920
that's how I felt when I was 
learning Elixir. 

90
00:04:47,920 --> 00:04:50,240
I was like, there's this like 
not good learning material. 

91
00:04:50,240 --> 00:04:53,760
And that's what sprung the my 
YouTube channel and just making 

92
00:04:53,760 --> 00:04:58,160
tutorials at 1st and then 
evolved and now as a podcast as 

93
00:04:58,160 --> 00:05:01,480
well. 
Yeah, you know, like when I got 

94
00:05:01,480 --> 00:05:06,360
into Elixir, I, I, I've never 
even heard of OTP or the beam 

95
00:05:06,360 --> 00:05:10,560
or, or, you know, all these cool
things that are somewhat unique 

96
00:05:10,560 --> 00:05:13,000
to Elixir, or at least, you 
know, if you're coming from 

97
00:05:13,000 --> 00:05:15,240
something like Ruby or 
JavaScript, it's miles beyond 

98
00:05:15,280 --> 00:05:17,200
like, like what you may have 
encountered in those. 

99
00:05:17,320 --> 00:05:19,800
I mean, it, it took me even, it 
took me a while to even like 

100
00:05:19,800 --> 00:05:21,800
realize those things were there.
You don't need to learn it 

101
00:05:21,800 --> 00:05:27,880
upfront. 
But yeah, like, like I remember 

102
00:05:27,880 --> 00:05:30,360
the, the first thing I was ever 
doing with Elixir, just to play 

103
00:05:30,360 --> 00:05:33,440
around with it, I was doing some
sort of like, like hacker rank 

104
00:05:33,440 --> 00:05:35,640
problems, that kind of thing. 
Exorcism, you know. 

105
00:05:36,480 --> 00:05:41,360
Just like writing just writing 
code, not actually using all the

106
00:05:41,440 --> 00:05:45,360
the magic of the beam. 
Yeah, I remember it was very, 

107
00:05:46,320 --> 00:05:49,800
very confusing at first because,
you know, it looks as immutable 

108
00:05:49,800 --> 00:05:52,600
and you, you know, you don't 
have wire loops there. 

109
00:05:52,600 --> 00:05:55,320
There's a lot of, like, things 
which are really quite simple 

110
00:05:55,320 --> 00:05:58,000
algorithmic puzzles where like 
just the way that I would have 

111
00:05:58,000 --> 00:06:01,760
done that coming from Ruby, the 
obvious way was not available to

112
00:06:01,760 --> 00:06:02,880
me. 
You know, it might be like, oh, 

113
00:06:02,880 --> 00:06:04,680
OK, it's really obvious how to 
solve this. 

114
00:06:04,680 --> 00:06:07,000
I would just like, initialize A 
variable and then I would 

115
00:06:07,120 --> 00:06:09,600
iterate in a while loop and, 
like, change that variable. 

116
00:06:09,720 --> 00:06:10,960
Oh, wait, what? 
I can't do that. 

117
00:06:11,640 --> 00:06:12,400
Yeah. 
But you know what? 

118
00:06:12,400 --> 00:06:15,360
Once you get used to, once you 
get used to the way Elixir does 

119
00:06:15,360 --> 00:06:16,880
things, things like the pipe 
operator. 

120
00:06:16,880 --> 00:06:18,760
Oh, I miss that so much in other
languages. 

121
00:06:18,760 --> 00:06:20,600
It's just such a beautiful way 
to structure code. 

122
00:06:21,280 --> 00:06:23,560
Patent matching. 
Other languages have got patent 

123
00:06:23,560 --> 00:06:26,120
matching, but I've I've yet to 
use a language that does it as 

124
00:06:26,120 --> 00:06:28,200
well as Elixer does. 
There's just so many little 

125
00:06:28,200 --> 00:06:30,800
niceties in the syntax. 
Yeah, it's a it's a beautiful 

126
00:06:30,800 --> 00:06:32,000
language. 
I love working with it. 

127
00:06:33,040 --> 00:06:35,120
Yeah, yeah, I was the. 
I was the same way. 

128
00:06:35,120 --> 00:06:38,840
I started just writing a simple 
like REST API project when I 

129
00:06:38,840 --> 00:06:44,120
learned Elixir and I like I knew
on paper Elixir could do 

130
00:06:44,120 --> 00:06:47,720
everything that I wanted for 
like, you know, distribution and

131
00:06:47,720 --> 00:06:50,560
fault tolerance, but I didn't 
actually know what it meant yet.

132
00:06:50,560 --> 00:06:55,720
And I was just writing code and 
I struggled with immutability, 

133
00:06:55,720 --> 00:06:59,280
the idea of just transforming 
things 'cause I came from object

134
00:06:59,280 --> 00:07:02,080
oriented. 
It was, it was, it was hard on 

135
00:07:02,080 --> 00:07:04,640
my brain. 
And then when I finally, when it

136
00:07:04,640 --> 00:07:09,520
finally clicked, I was like, I 
wish I learned an immutable 

137
00:07:09,520 --> 00:07:12,880
functional language first before
object oriented programming 

138
00:07:13,360 --> 00:07:15,920
because I think that would have 
made me like grasp programming 

139
00:07:15,920 --> 00:07:19,280
faster. 
Yeah, Yeah, that could be the 

140
00:07:19,280 --> 00:07:20,200
case. 
Yeah. 

141
00:07:21,560 --> 00:07:25,200
I mean, not now that I've seen 
things on the other side. 

142
00:07:25,200 --> 00:07:28,000
It's made me think, I think 
maybe OOP is a bit overrated. 

143
00:07:28,000 --> 00:07:29,960
You know, when you first learn 
it, it kind of makes sense. 

144
00:07:29,960 --> 00:07:32,600
You know, they teach you these 
sort of contrived examples to 

145
00:07:32,600 --> 00:07:36,000
teach you inheritance. 
Oh, yeah, You know, a car 

146
00:07:36,000 --> 00:07:38,920
inherits from a vehicle which 
inherits from this and that. 

147
00:07:38,920 --> 00:07:43,080
But then, you know, when you 
actually start, you know, when 

148
00:07:43,080 --> 00:07:46,160
you actually start building real
things like, you know, so many 

149
00:07:46,160 --> 00:07:48,720
of the classes that you end up 
making, it's not like real world

150
00:07:48,720 --> 00:07:51,800
objects that correspond to 
something concrete where 

151
00:07:51,800 --> 00:07:56,560
there's, it's more obvious 
hierarchy like, you know it. 

152
00:07:58,880 --> 00:08:01,240
Yeah. 
It just really works out as 

153
00:08:01,240 --> 00:08:03,520
smoothly as it does in these 
sort of tutorial examples. 

154
00:08:04,000 --> 00:08:07,400
I also think it's really easy to
like abstract away too much and 

155
00:08:07,400 --> 00:08:09,800
then you just complicate your 
code base for anyone that's not 

156
00:08:09,800 --> 00:08:11,880
familiar with it. 
I don't know. 

157
00:08:11,880 --> 00:08:14,480
Oh yeah, sure. 
And then just the magic 

158
00:08:14,480 --> 00:08:17,200
statefulness of things. 
You can just pass objects around

159
00:08:17,200 --> 00:08:19,080
and change them in multiple 
places. 

160
00:08:19,320 --> 00:08:23,680
Like, yeah. 
Yeah, I worked with Ruby for a 

161
00:08:23,680 --> 00:08:25,520
long time. 
I, I there's a lot of things I 

162
00:08:25,520 --> 00:08:27,520
like about Ruby. 
I think it's a beautiful 

163
00:08:27,520 --> 00:08:29,120
language. 
It just reads very well. 

164
00:08:29,120 --> 00:08:31,880
But like, there's a lot of 
things when you Start learning 

165
00:08:31,880 --> 00:08:33,679
Ruby, you're like, oh, this is 
really cool. 

166
00:08:33,679 --> 00:08:35,919
Like I could do all these things
super succinctly. 

167
00:08:36,880 --> 00:08:38,880
But then and then like once 
you've actually worked with him 

168
00:08:38,880 --> 00:08:42,080
for a while on bigger projects, 
you find that that there this 

169
00:08:42,080 --> 00:08:44,560
flexibility, the fact that like 
everything is an object and 

170
00:08:44,560 --> 00:08:47,120
everything can mutate and, and 
everything's got a zillion 

171
00:08:47,120 --> 00:08:48,760
methods defined on it and 
whatnot. 

172
00:08:49,840 --> 00:08:51,760
That flexibility just comes back
to bite you. 

173
00:08:51,760 --> 00:08:55,080
You know, like something will 
not be the type that you think 

174
00:08:55,080 --> 00:08:56,680
it is. 
You'll get the like no method 

175
00:08:56,680 --> 00:09:00,520
error on nil class. 
And you just end up having to do

176
00:09:00,520 --> 00:09:04,440
this deep dive through your 
stack, trying to trace a 

177
00:09:04,440 --> 00:09:07,160
variable all the way through, 
figure out what, figuring out 

178
00:09:07,160 --> 00:09:09,920
why it's changed. 
You can get these very obscure 

179
00:09:09,920 --> 00:09:12,720
errors that are hard, hard to 
debug because anything can 

180
00:09:12,720 --> 00:09:14,520
change at any point. 
So yeah, that's that's the 

181
00:09:14,520 --> 00:09:18,960
beauty of immutability, just 
everything there is right in 

182
00:09:18,960 --> 00:09:21,680
front of you, which I think 
helps with the AI as well. 

183
00:09:21,720 --> 00:09:27,000
You know the meaning. 
Just like separated processes, 

184
00:09:27,000 --> 00:09:29,040
right? 
Everything is kind of its own 

185
00:09:29,160 --> 00:09:31,800
its own state. 
You don't have to worry about 

186
00:09:31,960 --> 00:09:35,400
keeping things synced. 
Yeah, Yeah, exactly. 

187
00:09:35,400 --> 00:09:42,800
Yeah. 
Yeah, I'm curious to hear about 

188
00:09:42,800 --> 00:09:47,080
what this gauntlet AI thing is. 
What can you tell us about it 

189
00:09:47,080 --> 00:09:49,760
and like what put it on your 
radar and what made you want to 

190
00:09:49,760 --> 00:09:51,840
do it? 
Yeah, sure. 

191
00:09:51,920 --> 00:09:55,120
So I mean, when did ChatGPT come
out? 

192
00:09:55,120 --> 00:09:58,920
Was that like late 2022 I think?
Yeah, that that was when AI. 

193
00:09:59,240 --> 00:10:04,120
Yeah, yeah, that was when AI re.
I mean, obviously everyone, the 

194
00:10:04,120 --> 00:10:07,080
concept of AI has been around 
for as long as I can remember. 

195
00:10:07,080 --> 00:10:10,280
But yeah, it was really ChatGPT 
that, you know, sort of the 

196
00:10:10,280 --> 00:10:12,520
iPhone of AI. 
It just brought us into this new

197
00:10:12,520 --> 00:10:15,560
paradigm. 
And pretty much immediately when

198
00:10:15,560 --> 00:10:19,520
that came out, you know, there 
was a lot of talk about using AI

199
00:10:19,520 --> 00:10:22,440
to write code, people saying, Oh
yeah, this is going to write all

200
00:10:22,440 --> 00:10:24,000
the code for us, it's going to 
replace us. 

201
00:10:24,000 --> 00:10:27,600
But you know, I, I, I think like
a lot, a lot of people, I 

202
00:10:27,960 --> 00:10:30,720
especially in those early days, 
we're all trying to figure it 

203
00:10:30,720 --> 00:10:32,880
out, right? 
Like, like by now, if you're a 

204
00:10:32,880 --> 00:10:35,560
software engineer and you're not
at least trying to use AI in 

205
00:10:35,560 --> 00:10:38,200
your code, like, like you're 
very, very behind. 

206
00:10:39,280 --> 00:10:43,120
But you know, two years ago I 
heard if you're. 

207
00:10:43,400 --> 00:10:49,000
Going to get a hate comment for 
saying that but I had a Dave 

208
00:10:49,000 --> 00:10:51,760
from TV Labs on and he was 
basically telling people if 

209
00:10:51,760 --> 00:10:54,400
they're not using it in their 
daily workflows they're falling 

210
00:10:54,400 --> 00:10:58,360
behind and oh man we got some 
angry comments on YouTube. 

211
00:11:00,000 --> 00:11:02,720
Yeah, I mean, we can definitely 
get into that. 

212
00:11:03,080 --> 00:11:06,480
There's a, there's a ton of 
disagreement about exactly where

213
00:11:06,480 --> 00:11:09,640
this is going and how much you 
should be using it, which is 

214
00:11:11,120 --> 00:11:16,040
it's it's still very early, you 
know, but I, I think like I 

215
00:11:16,040 --> 00:11:18,280
would probably have said that 
two years ago, because I mean, 

216
00:11:18,280 --> 00:11:20,520
the technologies just wasn't 
great for a long time. 

217
00:11:20,520 --> 00:11:22,440
I tried to get to have copilot. 
I think that was a lot of 

218
00:11:22,440 --> 00:11:24,960
people's first. 
Experience I was against it so 

219
00:11:25,040 --> 00:11:29,480
like it was so crappy back back 
when it was first like even like

220
00:11:29,480 --> 00:11:32,000
I would just say within the last
year it's gotten cool, it's 

221
00:11:32,000 --> 00:11:36,880
gotten good right and before 
then it just kind of slowed 

222
00:11:36,960 --> 00:11:39,320
slowed me down. 
I just thought it gave me more 

223
00:11:39,320 --> 00:11:43,400
slop than it was worth and I was
better off just Robert ducking 

224
00:11:43,400 --> 00:11:46,240
with it but not actually letting
it write code. 

225
00:11:47,720 --> 00:11:50,960
Yeah, I think an experience, the
experience that like a lot of us

226
00:11:50,960 --> 00:11:54,400
more experienced guys have is 
that, you know, we've been doing

227
00:11:54,400 --> 00:11:57,720
this for so long, we just got 
our ways of working our habits. 

228
00:11:57,720 --> 00:12:00,040
We're very comfortable, we're 
good at it, right, Just because 

229
00:12:01,040 --> 00:12:04,640
we've had so much practice and 
like getting into that AI driven

230
00:12:04,640 --> 00:12:06,720
paradigm. 
It's just so different. 

231
00:12:07,280 --> 00:12:10,680
And yeah, like, no matter how 
smart you are, it's going to 

232
00:12:10,680 --> 00:12:13,240
slow you down at first because 
you just have to shed a lot of 

233
00:12:13,240 --> 00:12:15,240
old habits and learn new ways of
working. 

234
00:12:15,240 --> 00:12:21,840
And yeah, like, like that. 
I mean, especially, well, you 

235
00:12:21,840 --> 00:12:23,880
know, especially back when the 
models weren't nearly as good as

236
00:12:23,880 --> 00:12:26,920
they are now. 
Like, like, like it's easy to 

237
00:12:26,920 --> 00:12:28,160
see. 
I I certainly had that 

238
00:12:28,160 --> 00:12:30,000
resistance. 
But at the same time, you know, 

239
00:12:30,000 --> 00:12:32,600
there was just so much hype and 
it was clear what it still is 

240
00:12:32,600 --> 00:12:34,280
clear what direction everything 
is going in. 

241
00:12:34,280 --> 00:12:36,600
I was just thinking, man, I've 
got to, I've got to figure this 

242
00:12:36,600 --> 00:12:39,800
out. 
And yeah, so about a year ago, 

243
00:12:42,200 --> 00:12:45,120
everyone on Twitter who I 
followed was just talking about 

244
00:12:45,120 --> 00:12:47,960
this new program that was going 
on boot camp type thing, 

245
00:12:47,960 --> 00:12:53,320
Gauntlet AI Gauntlet is a like, 
So just to be clear, I'm not 

246
00:12:53,320 --> 00:12:55,520
like, I don't run Gauntlet. 
I'm not affiliated with it. 

247
00:12:55,520 --> 00:12:57,360
Like like like it's not my 
business. 

248
00:12:57,360 --> 00:13:01,360
I was just a participant. 
But Gauntlet is a intensive 10 

249
00:13:01,360 --> 00:13:04,400
week training program in 
building software with AI. 

250
00:13:05,520 --> 00:13:08,600
Yeah, very intensive. 
Yeah, really, almost every 

251
00:13:08,600 --> 00:13:10,360
waking minute is not an 
exaggeration. 

252
00:13:10,360 --> 00:13:14,360
But yeah, just a series of 
projects, challenges, deadlines.

253
00:13:14,560 --> 00:13:17,760
Everybody gets together and the 
goal is just, you know, let's go

254
00:13:17,760 --> 00:13:20,560
absolutely all in on LLM 
maximalism. 

255
00:13:21,360 --> 00:13:24,160
We're going to be building as 
much as we can as fast as we 

256
00:13:24,160 --> 00:13:26,080
can. 
Just throwing caution to the 

257
00:13:26,080 --> 00:13:28,520
wind with AIAI has got to build 
everything. 

258
00:13:28,520 --> 00:13:30,720
Let's just see what's possible. 
Let's see what we learn. 

259
00:13:31,600 --> 00:13:33,880
Yeah, totally free program, all 
expenses paid. 

260
00:13:33,920 --> 00:13:37,720
I mean, fundamentally it's a, 
it's a recruitment platform, you

261
00:13:37,720 --> 00:13:40,080
know, you make it to the end and
they'll they'll get, they've got

262
00:13:40,080 --> 00:13:44,080
a bunch of hiring partners lined
up who they'll, yeah, they'll 

263
00:13:44,560 --> 00:13:47,200
get you some interviews with. 
So that's how it's funded. 

264
00:13:47,240 --> 00:13:48,920
And yeah, it was one of those 
things. 

265
00:13:48,920 --> 00:13:50,400
It just came along at the right 
time for me. 

266
00:13:50,400 --> 00:13:54,960
You know, I been doing a bunch 
of freelance work. 

267
00:13:55,000 --> 00:13:57,600
I just finished a freelance 
contract actually. 

268
00:13:57,600 --> 00:14:01,480
So yeah, just trying to figure 
out what to do next. 

269
00:14:02,320 --> 00:14:04,320
And yeah, I definitely had my 
doubts about this court. 

270
00:14:04,320 --> 00:14:07,000
And I think it sounded too good 
to be true, to be honest. 

271
00:14:07,000 --> 00:14:10,280
But I was part of the second 
cohort. 

272
00:14:11,600 --> 00:14:13,280
Yeah, everyone was talking about
it. 

273
00:14:13,280 --> 00:14:17,480
I just DM Ed some people who 
were doing the first cohort on 

274
00:14:17,480 --> 00:14:19,360
Twitter. 
Just ask them about it, you 

275
00:14:19,360 --> 00:14:22,760
know, got on a call with a few 
of them and yeah, everyone was 

276
00:14:22,760 --> 00:14:23,840
like, hey, you've got to do it. 
It's great. 

277
00:14:23,840 --> 00:14:26,800
So I thought, why not, You know,
I'll take, I'll take a chance on

278
00:14:26,800 --> 00:14:27,680
this thing. 
And. 

279
00:14:27,680 --> 00:14:31,000
Yeah, TLDR it was great, learned
a lot. 

280
00:14:31,000 --> 00:14:33,280
It was fantastic. 
I would highly recommend it. 

281
00:14:34,560 --> 00:14:40,640
That's awesome. 
So essentially it you learned 

282
00:14:40,640 --> 00:14:45,880
how to vibe code, is that like a
fair, a fair thing to say? 

283
00:14:45,880 --> 00:14:48,520
Like you weren't writing code, 
you were leveraging it to 

284
00:14:48,520 --> 00:14:54,640
actually vibe products that were
written well or designed well at

285
00:14:54,640 --> 00:14:57,560
least. 
Well, as well as we could make 

286
00:14:57,560 --> 00:15:01,160
them given the time available. 
I mean, it was almost just like 

287
00:15:01,160 --> 00:15:06,200
a series of hackathons, you 
know, like the hackathon 

288
00:15:06,200 --> 00:15:09,320
mentality, you know, here's 
here's the task, here's a week. 

289
00:15:09,320 --> 00:15:11,720
I mean, most hackathons are not 
a week long. 

290
00:15:11,720 --> 00:15:15,920
But you know, it's just a series
of projects, a series, a series 

291
00:15:15,920 --> 00:15:18,280
of assignments. 
Hey, build this, see what you 

292
00:15:18,280 --> 00:15:21,800
can produce. 
And yeah, like, like, give me a 

293
00:15:21,800 --> 00:15:24,240
cool demo at the end. 
So, yeah, you know, we're not 

294
00:15:24,240 --> 00:15:27,280
we're not necessarily going for 
like the the most kind of 

295
00:15:27,280 --> 00:15:30,640
robust, perfect production ready
products that that are going to 

296
00:15:30,640 --> 00:15:32,440
like get out there and have huge
amounts of users. 

297
00:15:32,440 --> 00:15:35,400
It's more just about yeah, yeah,
you know what, what what cool 

298
00:15:35,400 --> 00:15:37,360
thing can we build in a week 
that would make a make a good 

299
00:15:37,360 --> 00:15:42,440
demo vibe coding. 
I hate that term, but we don't 

300
00:15:44,120 --> 00:15:48,720
we we don't seem to have settled
on on like what the what, what, 

301
00:15:49,240 --> 00:15:52,200
what else to call it, you know, 
agentic engineering perhaps, 

302
00:15:52,200 --> 00:15:54,120
but. 
Yeah, I feel like yeah, it's. 

303
00:15:54,320 --> 00:15:57,440
Widely recognized. 
We definitely need a better name

304
00:15:57,440 --> 00:16:00,680
for it because I feel like I 
just saw a post the other day 

305
00:16:00,680 --> 00:16:04,080
and I don't remember who did it 
now and maybe it was the 

306
00:16:04,080 --> 00:16:08,200
primogen but he was saying like 
the worst part about like the AI

307
00:16:08,200 --> 00:16:13,560
hype is everyone from the NFT 
and crypto world came over and 

308
00:16:13,560 --> 00:16:17,320
are now vibe coders and they're 
just annoying and being noisy in

309
00:16:17,320 --> 00:16:20,080
the space. 
Oh, yeah, for sure. 

310
00:16:22,160 --> 00:16:25,880
I mean, my, my take is that 
whatever you want to call it, 

311
00:16:25,880 --> 00:16:28,920
like, yeah, the technology is 
not perfect yet and we're all 

312
00:16:28,920 --> 00:16:31,600
still figuring this out. 
But there is no doubt in my mind

313
00:16:31,600 --> 00:16:34,800
that like, eventually we'll, 
everything is going to be 

314
00:16:34,800 --> 00:16:38,520
written by AI in some capacity 
and there'll be very little code

315
00:16:38,520 --> 00:16:40,480
written by hand. 
You've just got to compare the 

316
00:16:40,480 --> 00:16:43,560
state of the technology now to 
even a year ago, six months ago 

317
00:16:43,600 --> 00:16:45,240
for that matter. 
You know, when I started calling

318
00:16:45,280 --> 00:16:52,400
it, I think Claude 3.7 had just 
come out, which was, it's pretty

319
00:16:52,400 --> 00:16:54,840
good, definitely a lot better 
than the things before it, but 

320
00:16:54,840 --> 00:16:57,160
it still required a lot of hand 
holding. 

321
00:17:00,440 --> 00:17:01,960
Yeah, yeah. 
But it was very far from 

322
00:17:01,960 --> 00:17:04,400
perfect. 
I think the real big shift, and 

323
00:17:04,400 --> 00:17:07,400
you see this in the discourse as
well, just, you know, when a lot

324
00:17:07,400 --> 00:17:09,760
of a lot of other people started
to have this same realization. 

325
00:17:09,760 --> 00:17:16,839
It was kind of Q3 to Q4 last 
year when Opus Opus 4 came out. 

326
00:17:16,839 --> 00:17:20,800
Later Opus 4.5 and, you know, 
codecs started to get really 

327
00:17:20,800 --> 00:17:22,280
good as well. 
That was when you really started

328
00:17:22,280 --> 00:17:25,839
to see serious people. 
And yeah, not, not just not so 

329
00:17:25,839 --> 00:17:29,120
many slop slingers, you know, 
people, even people like Andre 

330
00:17:29,120 --> 00:17:32,680
Carpathy and, yeah, people 
who've got very highly regarded 

331
00:17:32,680 --> 00:17:34,600
programmers. 
They're, you know, they're, 

332
00:17:35,000 --> 00:17:37,560
that's even people like them 
starting to say, yeah, wow, this

333
00:17:37,560 --> 00:17:39,080
technology is good. 
It's getting very good at 

334
00:17:39,080 --> 00:17:40,320
writing code. 
Yeah. 

335
00:17:41,160 --> 00:17:43,320
Yeah. 
Curious how you feel. 

336
00:17:43,320 --> 00:17:48,720
How do you feel about like, 
writing less code and reviewing 

337
00:17:48,720 --> 00:17:54,000
more and being more like the 
overseer than actually like a 

338
00:17:54,000 --> 00:17:57,640
software developer from, you 
know, what we're used to, you 

339
00:17:57,640 --> 00:18:00,960
know, writing code, hacking 
things out, building. 

340
00:18:00,960 --> 00:18:04,920
Do you feel like they're like, I
like have my ups and downs with 

341
00:18:04,920 --> 00:18:08,240
it sometimes I'm like stoked 
about it because I can like do 

342
00:18:08,240 --> 00:18:11,320
so much more as a single 
developer. 

343
00:18:11,920 --> 00:18:17,680
But I'd also like part of me 
like misses the old old way of 

344
00:18:17,680 --> 00:18:21,520
doing things too. 
You know what, I I thought I 

345
00:18:21,520 --> 00:18:25,560
would miss writing code and I 
did it for so long, but I 

346
00:18:25,560 --> 00:18:27,000
honestly don't miss the old way 
at all. 

347
00:18:27,000 --> 00:18:32,160
I can't imagine going back. 
I wonder, It makes me think 

348
00:18:32,160 --> 00:18:34,760
about like, what was it that I 
liked about writing code? 

349
00:18:34,760 --> 00:18:37,680
I mean, it wasn't, it wasn't 
just the typing and the cranking

350
00:18:37,680 --> 00:18:39,720
out syntax. 
It was the thinking and the 

351
00:18:39,720 --> 00:18:42,160
solving problems and the 
learning. 

352
00:18:42,160 --> 00:18:46,760
And I've never learnt so much so
fast as now when I mean AI. 

353
00:18:46,760 --> 00:18:49,120
First of all, it's just a 
fantastic research tool. 

354
00:18:49,160 --> 00:18:53,040
You know, when there's something
I'm stuck on, you know, in the 

355
00:18:53,040 --> 00:18:55,400
olden days, I might spend a long
time reading through a lot of 

356
00:18:55,400 --> 00:18:59,200
very tedious and confusing 
documentation, you know, 

357
00:19:00,400 --> 00:19:04,880
Googling my error messages and, 
you know, getting stuck so, so 

358
00:19:04,880 --> 00:19:07,320
much time. 
It's funny looking back, like, 

359
00:19:07,320 --> 00:19:11,440
like how much time we used to 
spend, just just how much time 

360
00:19:11,440 --> 00:19:13,320
was waste. 
You know, you'd have a bug, 

361
00:19:13,320 --> 00:19:15,720
you'd get stuck on a problem. 
And like, like, you know, you 

362
00:19:15,880 --> 00:19:20,800
you could spend days just 
tediously trawling through Stack

363
00:19:20,800 --> 00:19:22,400
Overflow and not do you look at 
it. 

364
00:19:22,920 --> 00:19:24,600
Do you look at it as waste 
though? 

365
00:19:24,600 --> 00:19:27,920
Or do you look at it as like 
those hard moments that really 

366
00:19:27,920 --> 00:19:32,080
made you learn something new? 
Some of it was definitely waste,

367
00:19:32,360 --> 00:19:33,080
you know. 
Yeah, yeah. 

368
00:19:33,240 --> 00:19:35,960
I mean, you, you do learn stuff.
Yeah. 

369
00:19:35,960 --> 00:19:38,480
That, that is how you learn, You
know, you learn by encountering 

370
00:19:38,480 --> 00:19:40,560
it, encountering problems and 
things you don't know what to, 

371
00:19:40,720 --> 00:19:42,680
don't know how to solve and 
then, you know, learning how to 

372
00:19:42,680 --> 00:19:47,600
solve them. 
But, you know, nowadays, yeah, 

373
00:19:47,600 --> 00:19:49,680
it happens all the time. 
You know, I'll get stuck on a 

374
00:19:49,680 --> 00:19:51,960
bug or whatever, and I'll ask 
the AI what's going on here, 

375
00:19:51,960 --> 00:19:54,240
explain it to me, and it'll just
give me the answer like that. 

376
00:19:54,240 --> 00:19:56,960
Whereas, you know, it'll take 
me, it'll take me 5 minutes and 

377
00:19:56,960 --> 00:20:00,000
I've learned it, you know, in 
the past, you know, I might have

378
00:20:00,000 --> 00:20:02,760
spent the whole day just just 
trying to solve it. 

379
00:20:03,160 --> 00:20:05,120
And, you know, reading a lot of 
stuff that didn't help before 

380
00:20:06,120 --> 00:20:10,240
the rate of learning it's. 
How do you, how do you handle, 

381
00:20:10,240 --> 00:20:13,120
how do you force yourself to pay
attention and keep learning 

382
00:20:13,120 --> 00:20:17,560
versus going brain dead and just
trusting an LLM more than you 

383
00:20:17,560 --> 00:20:19,920
should? 
Like what is your, what's your 

384
00:20:19,920 --> 00:20:21,840
trick there? 
Because I feel like I've talked 

385
00:20:21,840 --> 00:20:25,080
to a lot of people that just 
they, they feel like they're 

386
00:20:25,080 --> 00:20:30,320
slipping with, you know, this 
they're, they're solving part of

387
00:20:30,320 --> 00:20:35,400
their brain. 
And I still feel like I am 

388
00:20:35,400 --> 00:20:39,120
pretty good at like, 'cause you 
see code written, you, you know,

389
00:20:39,120 --> 00:20:42,280
whether it's well written or 
not, you can make adjustments, 

390
00:20:42,280 --> 00:20:44,440
blah, blah, blah. 
But like the overall 

391
00:20:44,440 --> 00:20:47,960
architecture, how do you make 
sure you're like, involved in 

392
00:20:47,960 --> 00:20:49,560
actually thinking through 
problems? 

393
00:20:51,320 --> 00:20:53,400
Yeah, I mean, it can be a 
challenge. 

394
00:20:53,400 --> 00:20:56,880
I mean, my, my biggest challenge
is just like when you're working

395
00:20:56,880 --> 00:20:59,600
with AI, you spend a lot of time
waiting for the AI to complete 

396
00:20:59,600 --> 00:21:02,560
and there's a ton of context 
switching. 

397
00:21:02,760 --> 00:21:05,400
You know, I'll, I'll usually 
have multiple tabs open. 

398
00:21:05,400 --> 00:21:08,320
I'll be working on more than one
thing at once just because 

399
00:21:08,320 --> 00:21:09,600
otherwise, like, what else am I 
going to do? 

400
00:21:09,600 --> 00:21:11,360
I might say something running 
and it's going to take 15 

401
00:21:11,360 --> 00:21:16,480
minutes to get back to me, but I
don't know. 

402
00:21:16,520 --> 00:21:21,160
I, I, I, I feel like the way 
that I work with AI is I, I 

403
00:21:21,160 --> 00:21:24,040
don't necessarily read 
everything that it's producing, 

404
00:21:24,040 --> 00:21:25,920
or at least not like super 
attentively. 

405
00:21:25,920 --> 00:21:28,520
I might give it a skim, but you 
know, that doesn't mean that I 

406
00:21:28,520 --> 00:21:30,680
don't understand what it's 
producing or that I'm not paying

407
00:21:30,680 --> 00:21:32,480
attention. 
You know, if I've got a task 

408
00:21:32,480 --> 00:21:35,800
that I want to solve, you know, 
Peter Steinberger, the creator 

409
00:21:35,800 --> 00:21:39,200
of Open Floor, he's written some
blog posts about like how he 

410
00:21:39,200 --> 00:21:42,840
works with AI. 
And I like what he has to say 

411
00:21:42,840 --> 00:21:44,640
because I think it's pretty 
similar to the way I work. 

412
00:21:44,640 --> 00:21:47,960
You know, if there's a task I 
want to work on, I will, I'll 

413
00:21:47,960 --> 00:21:51,040
spend a long time planning and 
like just trying to get it into 

414
00:21:51,040 --> 00:21:53,440
my head exactly what it is that 
we're going to do. 

415
00:21:54,680 --> 00:21:58,360
I'll start by codecs and Florida
code. 

416
00:21:58,360 --> 00:22:01,080
Those are my main drivers now. 
But you know, I'll ask the LLM. 

417
00:22:03,520 --> 00:22:06,600
I'll start by just trying to 
understand the code, right? 

418
00:22:06,600 --> 00:22:09,320
Like I have some understanding 
already because I've already 

419
00:22:09,320 --> 00:22:11,040
been working on it, right? 
But it might be a big code base,

420
00:22:11,040 --> 00:22:12,400
don't necessarily understand 
everything. 

421
00:22:12,400 --> 00:22:15,280
So I'll spend a long time saying
here's what I want to build, you

422
00:22:15,280 --> 00:22:17,120
know, help me, help me flesh out
this idea. 

423
00:22:17,120 --> 00:22:19,400
You know, what am I not seeing? 
What are the edge cases? 

424
00:22:20,000 --> 00:22:22,520
And then, you know, tell it to 
remind me exactly what does the 

425
00:22:22,520 --> 00:22:24,560
existing code do? 
Where is everything, where 

426
00:22:24,560 --> 00:22:26,600
everything's located? 
What's the architecture? 

427
00:22:27,800 --> 00:22:29,600
Not just that having that kind 
of conversation with it. 

428
00:22:29,600 --> 00:22:34,960
I kind of the, the most 
important thing that, that I, 

429
00:22:35,880 --> 00:22:38,520
the most important thing to me 
when working with AI, with code,

430
00:22:38,520 --> 00:22:41,800
it's just understanding like the
better your understanding of, of

431
00:22:41,800 --> 00:22:44,640
like exactly what it is that you
want to build and exactly like 

432
00:22:44,640 --> 00:22:47,160
how the thing, how it all works,
how it fits together, what 

433
00:22:47,160 --> 00:22:52,120
you're going to change and where
like, like my job as I see it is

434
00:22:52,120 --> 00:22:56,520
to just like work with the AI 
until we both have a shared 

435
00:22:56,520 --> 00:22:59,240
mutual understanding, a high 
level understanding of what it 

436
00:22:59,240 --> 00:23:02,520
is that we're going to build. 
And, and then once I'm confident

437
00:23:02,520 --> 00:23:04,600
that like I understand what 
needs to be built and that the 

438
00:23:04,600 --> 00:23:07,440
LLM unders, yeah, the LLM shares
my understanding, I let it RIP 

439
00:23:07,440 --> 00:23:11,440
and I get great results. 
So yeah, I, I, I feel like I am,

440
00:23:12,480 --> 00:23:14,480
I feel like I'm still staying on
top of what the code does and 

441
00:23:14,480 --> 00:23:19,680
where everything is, you know, 
bad things happen and like, you 

442
00:23:19,680 --> 00:23:22,160
know, the, the, the NFT Bros 
revive coding. 

443
00:23:23,960 --> 00:23:26,320
Yeah, But I had a great time 
recently. 

444
00:23:26,320 --> 00:23:28,280
It was verification debt. 
You know, we used to talk about 

445
00:23:28,280 --> 00:23:30,560
technical debt, which is 
certainly still a thing. 

446
00:23:30,560 --> 00:23:34,280
But the danger now with AI is 
cognitive debt and verification 

447
00:23:34,280 --> 00:23:36,960
debt is that because you're not,
you're not typing everything 

448
00:23:36,960 --> 00:23:38,280
out. 
And so not every line of code 

449
00:23:38,280 --> 00:23:39,480
necessarily goes through your 
brain. 

450
00:23:39,480 --> 00:23:43,240
A gap can, a gap can grow 
between what the code does and 

451
00:23:43,240 --> 00:23:45,200
your understanding of what it 
does and your understanding of 

452
00:23:45,200 --> 00:23:48,240
how well it works. 
And like, like the sloppy 

453
00:23:48,240 --> 00:23:51,280
approach where you're just kind 
of firing things out without 

454
00:23:51,280 --> 00:23:53,520
taking a lot of time to plan and
really understand what it's 

455
00:23:53,520 --> 00:23:56,120
building that like, that's just 
a recipe for disaster. 

456
00:23:58,200 --> 00:23:58,640
Yeah. 
Yeah. 

457
00:23:58,640 --> 00:24:03,840
Like, but it it. 
That's what happens when you go 

458
00:24:03,840 --> 00:24:07,200
too fast. 
Yeah, I get that. 

459
00:24:07,440 --> 00:24:10,080
I definitely think the planning 
is important and I've started to

460
00:24:10,080 --> 00:24:15,400
see a lot in the aspect of like 
spec driven development and I 

461
00:24:15,400 --> 00:24:19,480
don't know, like are you 
familiar with that term at all? 

462
00:24:20,680 --> 00:24:21,880
Yeah, I'm familiar with the 
idea. 

463
00:24:22,040 --> 00:24:27,480
This is this is when you, I, I, 
I mean, it's kind of TDD kind of

464
00:24:27,480 --> 00:24:28,920
thing, you know, you'd start 
with the. 

465
00:24:29,920 --> 00:24:32,640
Yeah, yeah, it makes sense, I 
think. 

466
00:24:32,640 --> 00:24:34,680
I mean, that that's one way that
you, you, you solve that 

467
00:24:34,680 --> 00:24:37,280
verification problem. 
Yeah. 

468
00:24:37,280 --> 00:24:40,760
I mean, vibe coding tests, I 
think is is a particularly bad 

469
00:24:40,760 --> 00:24:44,000
idea. 
Well, I'm not not well if we're 

470
00:24:44,000 --> 00:24:46,560
talking about vibe coding in the
like very vibey sense where 

471
00:24:46,560 --> 00:24:49,200
where you're not actually paying
close attention to what the 

472
00:24:49,200 --> 00:24:50,720
tests do. 
Yeah, yeah. 

473
00:24:52,160 --> 00:24:53,600
Well, AI loves to do things 
like. 

474
00:24:54,080 --> 00:24:57,800
Just make them pass right? 
Yeah, well, if you tell it to 

475
00:24:57,800 --> 00:25:01,320
write the test first, it can do.
It'll write tests that do at a 

476
00:25:01,320 --> 00:25:03,440
glance they look like they work,
but then you actually read it 

477
00:25:03,440 --> 00:25:06,800
and it says, you know, people 
set it all up, but then we just 

478
00:25:06,800 --> 00:25:08,440
do assert true or something. 
It. 

479
00:25:08,920 --> 00:25:11,640
Doesn't actually test anything 
so that's when you've really got

480
00:25:11,640 --> 00:25:15,920
to make sure that that like it's
doing what you think it's doing 

481
00:25:15,920 --> 00:25:19,320
but yeah, that's another big 
aspect of working with AI. 

482
00:25:19,320 --> 00:25:22,080
You know it's constraints. 
It's because it can be, because 

483
00:25:22,200 --> 00:25:24,880
AI moves so fast and it can be 
very hard to keep up mentally 

484
00:25:24,880 --> 00:25:27,200
with what it's doing. 
Part of the skill of working 

485
00:25:27,200 --> 00:25:30,960
with it is just like setting up 
your workflow and your 

486
00:25:30,960 --> 00:25:33,800
harnesses, your methodology in 
such a way that like, like you 

487
00:25:33,800 --> 00:25:37,880
can be reasonably confident that
even if the AI is getting a bit 

488
00:25:37,880 --> 00:25:40,040
of head ahead of where you are 
mentally, like they're, they're 

489
00:25:40,040 --> 00:25:42,000
just things keeping it on the 
rails, things keeping it in 

490
00:25:42,000 --> 00:25:45,560
track. 
And yeah, like thing thing is 

491
00:25:45,560 --> 00:25:47,680
like just having a good test 
suite and making sure that the 

492
00:25:47,680 --> 00:25:50,360
tests are good and avoiding 
those situations where the AI is

493
00:25:50,360 --> 00:25:53,160
kind of taking shortcuts and 
hacking its way around the test.

494
00:25:53,160 --> 00:25:54,560
That's one way to stay on the 
rails. 

495
00:25:54,840 --> 00:25:58,040
Can you walk us through like a 
typical like workflow for you? 

496
00:25:58,040 --> 00:26:04,160
Like if you're adding a feature 
to existing code base, how do 

497
00:26:04,160 --> 00:26:07,680
you use AI to even begin the 
feature, plan the feature, and 

498
00:26:07,680 --> 00:26:09,560
then eventually implement the 
feature? 

499
00:26:09,560 --> 00:26:12,800
What do you? 
Yeah, I mean, I, I think I just 

500
00:26:12,800 --> 00:26:15,560
described my workflow really. 
I don't use a huge amount of 

501
00:26:15,560 --> 00:26:19,560
like, you know, a lot of people 
talk about these super 

502
00:26:19,560 --> 00:26:21,920
complicated setups they've got 
with like, you know, a product 

503
00:26:21,920 --> 00:26:25,880
manager agent and a code 
reviewer agent and a senior 

504
00:26:25,880 --> 00:26:28,760
engineer agent. 
All these different skills and 

505
00:26:28,880 --> 00:26:30,960
yeah. 
Yeah, yeah. 

506
00:26:30,960 --> 00:26:34,400
I, I, I don't know, I see a lot 
of people who've got very, very 

507
00:26:34,400 --> 00:26:37,760
impressive markdown collections,
but, but I don't know a lot, a 

508
00:26:37,760 --> 00:26:40,480
lot of the, a lot of those 
people can't really point. 

509
00:26:41,160 --> 00:26:42,680
I'm not really sure what they've
actually built. 

510
00:26:43,480 --> 00:26:48,800
I, I, yeah, like I said it, it's
about understanding. 

511
00:26:48,800 --> 00:26:50,640
You know, you don't need a 
product manager agent. 

512
00:26:50,640 --> 00:26:54,400
You, you, you need to, you need 
to have a, a sense of what you 

513
00:26:54,400 --> 00:26:55,760
need to have product sense, 
right? 

514
00:26:56,000 --> 00:26:58,800
I feel like you need to be the 
product manager like it's your 

515
00:26:58,800 --> 00:26:59,800
product. 
Right. 

516
00:27:00,880 --> 00:27:03,040
Yeah, exactly. 
Yeah. 

517
00:27:03,040 --> 00:27:06,720
So, so you know, I do everything
in the terminal. 

518
00:27:06,720 --> 00:27:10,520
I'll fire up codecs. 
I mean, it varies so much 

519
00:27:10,520 --> 00:27:12,120
depending on what it is that I 
want to build. 

520
00:27:12,120 --> 00:27:14,720
But you know, I'll just start 
by. 

521
00:27:16,760 --> 00:27:20,640
I'll spend a long time like 
really kind of going over 

522
00:27:20,640 --> 00:27:22,440
everything and fleshing 
everything out and getting the 

523
00:27:22,440 --> 00:27:24,880
details into my head before, 
before I start to try and write,

524
00:27:25,160 --> 00:27:28,280
before I actually tell it to, 
you know, actually make any 

525
00:27:28,280 --> 00:27:32,200
changes, you know? 
I'm curious like do you, do you 

526
00:27:32,440 --> 00:27:36,440
like, will you plan like a 
really large feature and let AI 

527
00:27:36,440 --> 00:27:40,080
kind of take it or do you break 
it down into the small steps and

528
00:27:40,560 --> 00:27:45,640
approach a huge feature like 
kind of implementation function 

529
00:27:45,640 --> 00:27:49,280
at a time kind of thing? 
And I think the, the, the 

530
00:27:49,280 --> 00:27:51,920
limiting factor for me that the,
the, the unit of work that I, I,

531
00:27:51,920 --> 00:27:56,160
I, I like, try to not exceed. 
It's, it's just like how, how 

532
00:27:56,160 --> 00:27:58,520
much can I juggle in my head at 
once, right? 

533
00:27:59,360 --> 00:28:03,440
And that, that, that's not just 
in terms of like, like, what are

534
00:28:03,440 --> 00:28:05,880
we working on with the AI? 
You know, it's like if I'm going

535
00:28:05,880 --> 00:28:08,760
to open APR or, you know, if I 
want to get this merged into 

536
00:28:08,760 --> 00:28:12,040
main, you know, what's like, 
well, it's like a self-contained

537
00:28:12,040 --> 00:28:15,360
unit of work that that could 
theoretically be deployed even 

538
00:28:15,360 --> 00:28:17,400
if it doesn't have like every 
last little thing that we want 

539
00:28:17,400 --> 00:28:19,880
it to, to, to, to be fully 
production ready. 

540
00:28:19,880 --> 00:28:24,480
But can we at least get it 
merged, You know, because the 

541
00:28:24,560 --> 00:28:28,120
the more we just let that grow, 
the, the more that is like still

542
00:28:28,120 --> 00:28:31,000
needing review and final review 
and approval, the more that, you

543
00:28:31,000 --> 00:28:32,400
know, I just can't keep that in 
my head. 

544
00:28:32,400 --> 00:28:35,360
I forget what, what, what I 
can't verify at all. 

545
00:28:35,440 --> 00:28:39,840
But yeah, I'll take a similar 
approach with with AI, you know,

546
00:28:39,840 --> 00:28:44,720
I'll, I'll, I'll talk to codecs.
We'll try and figure out, I'll 

547
00:28:44,720 --> 00:28:47,640
try and flesh out the plan. 
But you know, if I feel like 

548
00:28:47,640 --> 00:28:50,440
it's getting a bit too big, even
if, you know, I might say, I 

549
00:28:50,440 --> 00:28:53,040
might even ask codecs, you know,
what's, what's the first thing 

550
00:28:53,040 --> 00:28:54,560
we can build? 
What is something that's nice 

551
00:28:54,560 --> 00:29:00,360
and self-contained here? 
If yeah, if it's something super

552
00:29:00,360 --> 00:29:03,800
big and you know, it's just, I 
think I'm going to have to, you 

553
00:29:03,880 --> 00:29:05,080
know, it's just more than I can 
juggle. 

554
00:29:05,080 --> 00:29:07,520
That's when I'll start to do 
things like, you know, let's 

555
00:29:07,560 --> 00:29:10,280
write a plan, let's write a 
file, the full plan, write a 

556
00:29:10,280 --> 00:29:14,640
checklist. 
I don't, some people, some 

557
00:29:14,640 --> 00:29:17,000
people I work with like, like to
do that for, for much smaller 

558
00:29:17,000 --> 00:29:19,440
tasks and just like, like keep 
everything much more documented 

559
00:29:19,440 --> 00:29:22,240
from the beginning. 
I, I, I, I find it's not really 

560
00:29:22,240 --> 00:29:24,080
worth it. 
I, I don't like to do that 

561
00:29:24,080 --> 00:29:26,360
unless I feel like we're 
actually just past a certain 

562
00:29:26,640 --> 00:29:29,120
level of complexity and size, 
you know, if it's something that

563
00:29:29,120 --> 00:29:31,400
I can kind of do in one chat and
self contain. 

564
00:29:31,520 --> 00:29:34,320
And yeah, I, I feel like the, 
the code itself is the 

565
00:29:34,320 --> 00:29:37,720
checklist. 
It's just creating that markdown

566
00:29:37,720 --> 00:29:39,880
file. 
It's just another file to juggle

567
00:29:39,880 --> 00:29:41,600
and yeah, it just gets messy 
but. 

568
00:29:41,680 --> 00:29:44,760
Yeah, that makes sense. 
Yeah, I do think a lot of people

569
00:29:44,760 --> 00:29:48,840
will over complicate their whole
like AI workflow and I'm not 

570
00:29:48,840 --> 00:29:51,000
sure why. 
And maybe it's like they want to

571
00:29:51,000 --> 00:29:55,320
just have something so 
complicated that other people 

572
00:29:55,320 --> 00:29:58,280
might not understand it and then
they just think it works better.

573
00:29:58,280 --> 00:30:00,960
But I'm kind of like, I'm kind 
of in your boat. 

574
00:30:00,960 --> 00:30:04,000
I'll, I'll plan things, but I 
also don't ever want to push 

575
00:30:04,000 --> 00:30:07,760
code that's more than, you know,
a couple modules changed because

576
00:30:07,760 --> 00:30:11,760
I don't want to review APR that 
it's like 50 modules that I have

577
00:30:11,760 --> 00:30:15,400
to review. 
I want a handful of changes that

578
00:30:15,400 --> 00:30:19,760
are easy to read and look at. 
Yeah, yeah, exactly. 

579
00:30:19,760 --> 00:30:25,800
I think something I'm thinking 
more and more like just having, 

580
00:30:26,640 --> 00:30:28,560
you know, I mean, I've been on 
this journey for a year now. 

581
00:30:28,560 --> 00:30:31,880
I've spent a huge amount of time
working with AI and like really 

582
00:30:31,880 --> 00:30:33,960
learning how, you know, really 
trying to get the most out of 

583
00:30:33,960 --> 00:30:35,600
it. 
But the further I get into it, 

584
00:30:35,920 --> 00:30:37,760
it certainly is a skill in 
itself. 

585
00:30:37,760 --> 00:30:40,040
Like, you know, you've got to 
learn how to use AI, you've got 

586
00:30:40,040 --> 00:30:42,040
to learn how the tools work. 
You've got to try different 

587
00:30:42,040 --> 00:30:44,160
things, understand what models 
are good at, what they're bad 

588
00:30:44,160 --> 00:30:48,800
at, but nothing can replace 
like, like all the skills that 

589
00:30:48,800 --> 00:30:51,960
we used to need before AI, just 
understanding architecture, 

590
00:30:51,960 --> 00:30:54,920
understanding clean code, 
understanding, you know, 

591
00:30:55,040 --> 00:30:58,120
security and just just every 
little thing that goes into 

592
00:30:58,120 --> 00:31:02,280
making a, a good, an app that 
not only works and is secure and

593
00:31:02,280 --> 00:31:04,960
like, like, you know, meets the 
business requirements, but just 

594
00:31:04,960 --> 00:31:08,400
like, like it is, is 
maintainable and all the things 

595
00:31:08,400 --> 00:31:12,000
we care about as, as developers,
something that, you know, we can

596
00:31:12,000 --> 00:31:13,360
just keep on working with 
smoothly. 

597
00:31:13,360 --> 00:31:18,120
Like like those skills have not 
changed, which is why I think 

598
00:31:18,120 --> 00:31:21,200
that there's a lot of worry at 
the moment. 

599
00:31:21,560 --> 00:31:24,360
I was certainly worried for a 
while that AI was going to take 

600
00:31:24,360 --> 00:31:27,760
all of our jobs. 
I it might take some jobs. 

601
00:31:27,800 --> 00:31:30,920
I I mean, no one really knows 
what the future holds, but like 

602
00:31:31,040 --> 00:31:35,240
for now, at least as I see it, 
like I think I'm senior 

603
00:31:35,240 --> 00:31:37,880
engineers and people who've got 
that, you know, people who just 

604
00:31:37,880 --> 00:31:40,800
have a very deep technical 
understanding and just good 

605
00:31:40,800 --> 00:31:42,360
coders. 
Those are the people who make 

606
00:31:42,360 --> 00:31:47,760
the best AIAI driven developers.
Yeah. 

607
00:31:47,760 --> 00:31:51,080
Like, like if you're, if you're 
not so experienced, if you're a 

608
00:31:51,080 --> 00:31:54,520
junior engineer, like like 
there's a lot of yeah, yeah. 

609
00:31:54,520 --> 00:31:56,600
Like, like you can add negative 
value. 

610
00:31:56,600 --> 00:31:58,920
Definitely because AI moves. 
So when you're when you're 

611
00:31:58,920 --> 00:32:00,600
working with AI, you move so 
fast, right? 

612
00:32:00,600 --> 00:32:02,920
So like bad decisions compound 
very quickly. 

613
00:32:02,920 --> 00:32:05,640
You know, if you choose the 
wrong, if you choose, if you're 

614
00:32:05,640 --> 00:32:07,880
building with the wrong 
technology, if if you if you 

615
00:32:07,880 --> 00:32:10,200
make a poor architectural 
choice, then you know, it might 

616
00:32:10,320 --> 00:32:12,320
in the past, it might be months 
until that really becomes a 

617
00:32:12,320 --> 00:32:14,360
problem because it just takes so
long to build anything. 

618
00:32:14,360 --> 00:32:16,320
But you know, now with AI, 
you're moving so much faster. 

619
00:32:16,320 --> 00:32:18,840
There's there's bad decisions 
that you make early can come 

620
00:32:18,840 --> 00:32:23,240
back to bite you much sooner. 
So, yeah, that's when that that,

621
00:32:23,240 --> 00:32:25,160
that that understanding that 
really only comes with 

622
00:32:25,160 --> 00:32:28,360
experience and, and, and just, 
you know, having worked on big 

623
00:32:28,360 --> 00:32:31,160
problems and seen the same 
things over and over again. 

624
00:32:31,160 --> 00:32:33,760
And, and, and just like 
understanding how things go 

625
00:32:33,760 --> 00:32:35,400
wrong, that that's where it's 
really valuable. 

626
00:32:37,640 --> 00:32:42,680
I'm not sure. 
I, I, I do wonder, like, I don't

627
00:32:42,680 --> 00:32:45,560
know what it's like to be a 
junior developer working with AI

628
00:32:45,560 --> 00:32:49,880
coming into all of this now. 
I, I, yeah, I I have a concern. 

629
00:32:49,960 --> 00:32:54,320
Like like not I kind. 
Of feel bad for a. 

630
00:32:54,320 --> 00:32:59,520
Junior 'cause I don't know how 
they're going to get their foot 

631
00:32:59,520 --> 00:33:02,800
in the door and, like, get 
trained and become senior 

632
00:33:02,800 --> 00:33:07,120
because there's so much, like, 
I'm overwhelmed with AI, right, 

633
00:33:07,120 --> 00:33:10,320
'cause it helps me so much, but 
it's also made me have to 

634
00:33:10,320 --> 00:33:14,000
understand so many things I 
don't think I would have had to 

635
00:33:14,000 --> 00:33:18,000
learn without AI. 
Like, now I have to be very 

636
00:33:18,000 --> 00:33:21,520
capable of understanding my 
DevOps deployment. 

637
00:33:21,520 --> 00:33:25,320
Like all this stuff, like 
security, like you were saying, 

638
00:33:25,560 --> 00:33:28,800
it's kind of just opened the 
door and made our responsibility

639
00:33:28,800 --> 00:33:32,400
larger. 
And we're just like, now an 

640
00:33:32,400 --> 00:33:36,440
orchestrator of all these 
subtasks underneath us, but 

641
00:33:36,760 --> 00:33:40,320
we're still having to understand
how everything works together. 

642
00:33:41,920 --> 00:33:44,640
Yeah. 
You know, I, I have heard people

643
00:33:44,640 --> 00:33:47,560
saying I, I think former guest 
on your podcast, I've heard talk

644
00:33:47,560 --> 00:33:50,320
about this. 
Yeah, it's a brutal market at 

645
00:33:50,320 --> 00:33:53,040
the moment for junior engineers 
because AI has made senior 

646
00:33:53,040 --> 00:33:56,080
engineers so much more powerful 
that there's just, there's just 

647
00:33:56,080 --> 00:33:57,640
less need to hire juniors 
anymore. 

648
00:33:57,640 --> 00:34:00,520
But that's not a sustainable 
strategy, right? 

649
00:34:00,720 --> 00:34:04,480
And because seniors are going to
get old and retire and someone's

650
00:34:04,480 --> 00:34:06,440
got to, someone's got to come up
from beneath them. 

651
00:34:07,280 --> 00:34:13,239
I, I, I mean, I don't know, 
like, I mean, I guess the 

652
00:34:13,239 --> 00:34:15,920
advantage if you are a junior 
now is like, it has never been 

653
00:34:15,920 --> 00:34:18,000
easier to learn really, really 
fast. 

654
00:34:18,000 --> 00:34:21,199
Like I said earlier, AI is just 
the most powerful research and 

655
00:34:21,199 --> 00:34:23,080
learning tool in the history of 
man. 

656
00:34:24,000 --> 00:34:26,320
Yeah, I know it doesn't always 
get everything right and it 

657
00:34:26,320 --> 00:34:28,120
gives you wrong answers 
sometimes, blah, blah, blah. 

658
00:34:28,120 --> 00:34:31,719
But like the speed up that you 
get, the ease of research, the 

659
00:34:31,719 --> 00:34:34,960
way that you can just ask such 
specific questions and, you 

660
00:34:34,960 --> 00:34:37,400
know, get clarification on every
little thing you're unsure 

661
00:34:37,400 --> 00:34:40,560
about. 
Do just deep research, search 

662
00:34:40,560 --> 00:34:43,280
the whole Internet and get back 
teeth filled reports on obscure 

663
00:34:43,280 --> 00:34:46,120
things like I'm jealous of 
juniors. 

664
00:34:46,320 --> 00:34:50,560
Yeah, I would say like. 
As long as you're not being lazy

665
00:34:50,600 --> 00:34:54,480
as a junior and you're willing 
to put in the time and learn and

666
00:34:54,480 --> 00:34:57,080
like actually ask AI. 
Like, hey, if there's a piece of

667
00:34:57,080 --> 00:34:59,840
code you don't understand, have 
it explain it to you. 

668
00:34:59,840 --> 00:35:01,560
Like, that's the coolest thing 
ever, right. 

669
00:35:01,560 --> 00:35:03,840
If you're ever in, like, a 
complicated module and you're 

670
00:35:03,840 --> 00:35:08,400
like, what is going on here? 
You now have a tool that can, 

671
00:35:08,400 --> 00:35:12,240
like, track everything a 
function is doing and explain it

672
00:35:12,240 --> 00:35:14,440
to you. 
Oh, absolutely. 

673
00:35:14,440 --> 00:35:15,040
Yeah. 
I I, I I. 

674
00:35:15,080 --> 00:35:17,880
I think the the one 
characteristic that everybody 

675
00:35:17,880 --> 00:35:20,720
needs, but juniors especially, 
it's just curiosity. 

676
00:35:20,960 --> 00:35:24,120
You, you, you need to be 
inquisitive. 

677
00:35:24,120 --> 00:35:27,120
You need to always be, be asking
like, like, why does it work 

678
00:35:27,120 --> 00:35:29,040
that way? 
You know, what's what's best 

679
00:35:29,040 --> 00:35:30,560
practice? 
Is this a good idea? 

680
00:35:30,560 --> 00:35:35,440
What are my options? 
You were talking about just then

681
00:35:35,440 --> 00:35:38,760
about, you know, how like we're 
taking on so many more 

682
00:35:38,760 --> 00:35:41,640
responsibilities and we're doing
broader things as seniors. 

683
00:35:42,080 --> 00:35:45,360
I saw a really it was, it was, 
it was either it was one of the 

684
00:35:45,360 --> 00:35:47,800
big AI labs. 
They were tweeting, I can't 

685
00:35:47,800 --> 00:35:51,040
remember who it might have been 
open AI, something about their 

686
00:35:51,040 --> 00:35:52,880
hiring. 
And they made a really 

687
00:35:52,880 --> 00:35:55,840
interesting point. 
They said that, yeah, they're, 

688
00:35:55,880 --> 00:35:58,600
they're finding that their 
developers are bifurcating, like

689
00:35:58,680 --> 00:36:00,960
the people who are crushing it, 
They're they're going one of two

690
00:36:00,960 --> 00:36:03,120
ways, right? 
Like people are either becoming 

691
00:36:03,120 --> 00:36:06,760
like hyper generalists, you 
know, people who who are just 

692
00:36:06,760 --> 00:36:10,160
like, you know, great coders 
with great product sense, great 

693
00:36:10,160 --> 00:36:12,400
design sense, they're doing dev 
OPS, They're they're just 

694
00:36:12,720 --> 00:36:14,040
handling every, everything they 
can. 

695
00:36:14,400 --> 00:36:16,840
Or you've got people who are, 
are like hyper, hyper 

696
00:36:16,840 --> 00:36:18,680
specialists. 
So, you know, you know, but 

697
00:36:18,760 --> 00:36:21,560
maybe the special special isn't,
aren't necessarily about 

698
00:36:21,560 --> 00:36:23,160
engineering. 
You know, it's, it's about like,

699
00:36:23,160 --> 00:36:26,400
you know, that that interfit, 
that overlap between that 

700
00:36:26,400 --> 00:36:28,400
overlap between like technical 
skills and then the things that 

701
00:36:28,400 --> 00:36:30,480
are like more specific to your 
business and the industry that 

702
00:36:30,480 --> 00:36:32,360
you're in. 
Just having that really deep 

703
00:36:32,360 --> 00:36:35,400
knowledge. 
Yeah, that matches my 

704
00:36:35,400 --> 00:36:38,760
experience. 
I think AI makes it really easy.

705
00:36:38,840 --> 00:36:41,520
Well, I don't know if I would 
say easy, but it makes it easier

706
00:36:41,520 --> 00:36:43,440
to be a generalist. 
I agree. 

707
00:36:43,440 --> 00:36:46,760
And I feel like I was gonna say 
I think. 

708
00:36:46,840 --> 00:36:50,200
A is made me more of a 
generalist, but it's kind of 

709
00:36:50,200 --> 00:36:52,560
fits my personality. 
I like to build things that 

710
00:36:52,560 --> 00:36:56,520
solve problems that I have and 
it allowed me I so I built 

711
00:36:56,520 --> 00:37:02,400
Killswitch, which is like an 
encrypted document store but has

712
00:37:02,400 --> 00:37:05,600
dead man switch capabilities. 
So like if I stop checking in, I

713
00:37:05,600 --> 00:37:08,400
can have my encrypted documents 
get sent to, you know, my loved 

714
00:37:08,400 --> 00:37:10,560
ones or people that I want to 
have access to them. 

715
00:37:10,960 --> 00:37:15,360
And like, as far as like 
product, like UI design and 

716
00:37:15,360 --> 00:37:19,360
everything like I would not have
got a clean product without LLMS

717
00:37:19,360 --> 00:37:21,800
helping me. 
Oh, absolutely. 

718
00:37:21,800 --> 00:37:25,720
Yeah, yeah, yeah, I. 
Mean that's when yeah, yeah, 

719
00:37:25,800 --> 00:37:29,280
that's that that's one of the 
most amazing things with working

720
00:37:29,280 --> 00:37:32,040
with LLMS is just like sometimes
I want to step into something 

721
00:37:32,040 --> 00:37:36,920
totally new like that. 
Like I recently made a desktop 

722
00:37:36,920 --> 00:37:41,920
app like for I mean, there was 
just some something we had on my

723
00:37:41,920 --> 00:37:45,160
team where where we had a web 
version and we were to cut a 

724
00:37:45,160 --> 00:37:47,400
Long story short, we were like, 
maybe we could bundle this up 

725
00:37:47,400 --> 00:37:50,080
into a desktop thing Electron, 
which I've never worked with 

726
00:37:50,080 --> 00:37:52,440
before and I'm sure I could have
figured it out anyway, but like 

727
00:37:53,200 --> 00:37:55,680
just stepping into that or I've 
done this with mobile apps as 

728
00:37:55,680 --> 00:37:57,200
well. 
I have very little experience 

729
00:37:57,200 --> 00:37:58,680
with mobile apps. 
I've always been a full stack 

730
00:37:58,680 --> 00:38:01,720
web developer. 
But yeah, just just like diving 

731
00:38:01,720 --> 00:38:04,400
into that totally new domain and
getting something out super 

732
00:38:04,400 --> 00:38:05,360
fast. 
Yeah. 

733
00:38:05,400 --> 00:38:07,520
You know, it's not as good as a 
super experienced mobile 

734
00:38:07,520 --> 00:38:10,600
developer would have made. 
But, you know, just how far was 

735
00:38:10,600 --> 00:38:12,880
I able to get in a day where, 
you know, previously I might 

736
00:38:12,880 --> 00:38:15,000
have the whole electron 
electron. 

737
00:38:15,000 --> 00:38:18,480
Is cool like. 
The fact that you can just like.

738
00:38:18,480 --> 00:38:22,560
Bundle a JavaScript app up or a 
web app essentially right? 

739
00:38:22,560 --> 00:38:26,440
And now you have a native app. 
Yeah, yeah, yeah. 

740
00:38:26,440 --> 00:38:29,920
It's very cool. 
There was something I was gonna 

741
00:38:29,920 --> 00:38:36,600
say. 
Yeah, yeah, yeah. 

742
00:38:36,600 --> 00:38:38,760
The generalist thing, again, I 
think it just comes back to 

743
00:38:38,760 --> 00:38:40,680
curiosity. 
I, I just like learning, you 

744
00:38:40,680 --> 00:38:42,600
know, I, I, I like 
understanding. 

745
00:38:42,640 --> 00:38:46,200
I, I, I, I want to understand as
much as I can about, about what 

746
00:38:46,200 --> 00:38:48,680
it just, we're doing, you know, 
that could be the code itself. 

747
00:38:48,680 --> 00:38:51,880
It could be things like, I don't
know, we're hosting on AWSI want

748
00:38:51,880 --> 00:38:54,800
to actually get a really good 
understanding of, of, you know, 

749
00:38:55,080 --> 00:38:58,000
all the, all these different 
things in our infrastructure. 

750
00:38:58,000 --> 00:38:59,520
How do they work? 
How do I work with them? 

751
00:39:00,280 --> 00:39:01,880
Because these are the things 
that make you work better with, 

752
00:39:01,920 --> 00:39:06,480
with AI as well, you know, the, 
the, the secret to writing a 

753
00:39:06,480 --> 00:39:08,920
good prompt and getting, getting
what you want out of AI. 

754
00:39:08,920 --> 00:39:11,280
It's being specific. 
You know, don't, don't say make 

755
00:39:11,280 --> 00:39:13,800
me an app, Say make me a Phoenix
app. 

756
00:39:15,200 --> 00:39:16,560
Yeah. 
Make me a Phoenix app. 

757
00:39:16,560 --> 00:39:18,560
Where, you know, here, here are 
the ecto schemas. 

758
00:39:18,560 --> 00:39:20,040
They're connected with these 
foreign keys. 

759
00:39:20,280 --> 00:39:22,880
You know, watch out for race 
conditions or, I don't know, 

760
00:39:22,880 --> 00:39:25,760
like you got to know the jargon.
You got to know what it is. 

761
00:39:25,760 --> 00:39:27,840
And yeah, yeah. 
Like the deeper you're 

762
00:39:27,840 --> 00:39:30,960
understanding, This is why I'm 
learning so much because you 

763
00:39:30,960 --> 00:39:33,520
know the things that I things 
that I understand. 

764
00:39:33,520 --> 00:39:35,880
When I understand the domain 
well and I know exactly what it 

765
00:39:35,880 --> 00:39:38,040
is that I want to build. 
It's just so easy to work with 

766
00:39:38,240 --> 00:39:41,680
AI because I can type out if you
don't talk to an. 

767
00:39:41,680 --> 00:39:45,240
LLM like a software engineer, 
you're not going to get the same

768
00:39:45,240 --> 00:39:48,280
output, you know what I mean? 
Like if you're not using the 

769
00:39:48,280 --> 00:39:52,080
right terminology, then the 
right things like yeah, just 

770
00:39:52,080 --> 00:39:55,560
like, hey, build me A to do list
app like it's gonna just 

771
00:39:55,720 --> 00:39:59,360
blindly, it'll probably build 
you something, but you're not 

772
00:39:59,360 --> 00:40:02,080
picking the tech stack, you're 
not picking any architectural 

773
00:40:02,080 --> 00:40:04,440
decisions. 
You're not creating your schemas

774
00:40:04,800 --> 00:40:07,000
like. 
You're leaving a lot out there 

775
00:40:07,000 --> 00:40:10,640
to just. 
Drift in the wrong direction, 

776
00:40:10,640 --> 00:40:12,680
right? 
Yeah. 

777
00:40:12,680 --> 00:40:15,800
The reason I'm learning so much 
so fast is that like almost all 

778
00:40:15,800 --> 00:40:18,480
my time now is spent on the 
things that I don't understand. 

779
00:40:18,480 --> 00:40:20,520
Because the things that I 
already understand, well, I just

780
00:40:20,520 --> 00:40:23,120
crank out the great prompt. 
It's very specific and I'm just 

781
00:40:23,120 --> 00:40:24,920
telling the AI exactly what to 
do. 

782
00:40:24,920 --> 00:40:28,120
And yeah, there's no 
ambiguities, which just rolls 

783
00:40:28,120 --> 00:40:31,720
through it and does it. 
And it's it's the things that 

784
00:40:31,720 --> 00:40:34,960
I'm not so sure of, things I 
haven't worked with before, 

785
00:40:34,960 --> 00:40:38,240
maybe aspects of our own code 
base that, you know, I'm just 

786
00:40:38,920 --> 00:40:41,240
not so familiar with or maybe 
I've forgotten the specifics 

787
00:40:41,240 --> 00:40:44,720
that I'm spending most of my 
time just interrogating that 

788
00:40:44,720 --> 00:40:46,520
asking codecs, how it works, 
where do things go? 

789
00:40:47,800 --> 00:40:52,520
I think I think non-technical 
PMSIII think they've got to be a

790
00:40:52,520 --> 00:40:55,440
lot more afraid for their jobs 
right now than than engineers 

791
00:40:55,440 --> 00:40:56,480
have. 
I think great. 

792
00:40:56,920 --> 00:41:00,240
Yeah. 
And a lot of non-technical 

793
00:41:00,240 --> 00:41:05,560
roles, actually, I think, you 
know, even if we're not writing 

794
00:41:05,560 --> 00:41:08,800
code anymore, like just, we 
still have to know how to read. 

795
00:41:08,840 --> 00:41:12,320
It and understand it, you know. 
Yeah, well, even if you're not. 

796
00:41:12,320 --> 00:41:15,520
Reading code, just understanding
tech, understanding computers, 

797
00:41:15,640 --> 00:41:17,560
like, like how things work, 
right? 

798
00:41:17,800 --> 00:41:19,400
Yeah. 
How things work right like. 

799
00:41:19,880 --> 00:41:22,640
The best, the best, the best PMS
to work with, the best 

800
00:41:22,640 --> 00:41:24,960
recruiters to work with the the 
best people in those kind of 

801
00:41:24,960 --> 00:41:26,760
roles which are software 
adjacent. 

802
00:41:26,760 --> 00:41:29,240
They're the ones who actually 
understand what it all does, 

803
00:41:29,240 --> 00:41:31,960
even if it's not their like core
day-to-day responsibility to 

804
00:41:31,960 --> 00:41:34,560
like work with it directly. 
You know, people who you can, 

805
00:41:35,160 --> 00:41:37,080
people who you can communicate 
with effectively because you 

806
00:41:37,080 --> 00:41:39,880
speak the same language. 
So yeah. 

807
00:41:39,880 --> 00:41:43,880
And part of the reason we've had
this split between technical and

808
00:41:43,880 --> 00:41:47,520
non-technical people is just 
like it's just a bandwidth 

809
00:41:47,520 --> 00:41:48,560
thing. 
You know, there's a lot of work 

810
00:41:48,560 --> 00:41:51,840
that needs to be done where like
if you're going to take the time

811
00:41:51,840 --> 00:41:54,600
to like learn to code and like, 
like get deep into the tech, 

812
00:41:54,600 --> 00:41:56,520
then you're not going to have 
the time to do all those other 

813
00:41:56,520 --> 00:41:58,280
non-technical jobs that need 
doing. 

814
00:41:58,280 --> 00:42:02,680
But yeah, like, like now more of
the non-technical stuff is being

815
00:42:02,680 --> 00:42:07,040
done by AI, more of the 
technical stuff is being speed 

816
00:42:07,040 --> 00:42:10,560
up by, by AI or like, you know, 
made a lot easier. 

817
00:42:10,560 --> 00:42:13,520
It's easier to learn. 
So, so those pressures just 

818
00:42:13,520 --> 00:42:15,880
aren't there so much anymore. 
Yeah. 

819
00:42:15,880 --> 00:42:17,720
Yeah. 
I, I, I think that blurring, 

820
00:42:17,720 --> 00:42:19,520
it's already happening. 
It's like I said, people are 

821
00:42:19,520 --> 00:42:21,000
going to become generalists. 
You've got to become a 

822
00:42:21,000 --> 00:42:24,240
generalist. 
Yeah, we got a couple. 

823
00:42:24,240 --> 00:42:28,800
Comments from Bry J Bry, he 
says, hey, George, I don't know.

824
00:42:29,800 --> 00:42:33,480
And he said the ability to learn
is incredible if you have a 

825
00:42:33,480 --> 00:42:37,240
sense of what what the real 
learning is. 

826
00:42:37,440 --> 00:42:40,440
And I think that's just like 
asking AI the right questions 

827
00:42:40,440 --> 00:42:42,040
right as you're building things.
Yeah. 

828
00:42:43,120 --> 00:42:46,160
If we stop Brian Bryce. 
It might be. 

829
00:42:46,960 --> 00:42:49,400
Yeah, yeah, yeah. 
Hi, Brian, if that's you. 

830
00:42:49,440 --> 00:42:50,640
Yeah, sorry, we used to work 
together. 

831
00:42:51,280 --> 00:42:52,440
Oh, awesome. 
Very cool. 

832
00:42:53,200 --> 00:42:56,880
And then he said that if we 
stopped all AI progress right 

833
00:42:56,880 --> 00:43:00,440
now, it would take us years to 
realize the benefits just to 

834
00:43:00,440 --> 00:43:02,400
education alone, he said. 
Yeah. 

835
00:43:02,400 --> 00:43:04,920
Brian Bryce. 
So yeah. 

836
00:43:05,600 --> 00:43:09,960
Yeah, great to hear from you. 
Yeah, I'm actually working in Ed

837
00:43:09,960 --> 00:43:12,560
tech now. 
So that's, yeah, it's not just 

838
00:43:12,560 --> 00:43:15,360
programming education. 
Yeah. 

839
00:43:16,800 --> 00:43:19,600
And yeah, coding is certainly 
not the only thing we're we're 

840
00:43:19,800 --> 00:43:25,680
that I use AI to learn about. 
I mean, it's yeah, it's amazing.

841
00:43:26,440 --> 00:43:27,440
I wish that it is. 
Yeah. 

842
00:43:27,480 --> 00:43:30,120
I just imagine if this had 
existed when I was in school. 

843
00:43:30,120 --> 00:43:33,800
I think I don't know. 
A lot of people are certainly. 

844
00:43:33,800 --> 00:43:37,880
I'm not really sure how the 
education sector is going to 

845
00:43:37,880 --> 00:43:40,640
cope because I'm sure there is 
rampant cheating going on. 

846
00:43:40,640 --> 00:43:42,280
Oh, yeah, it did. 
I would. 

847
00:43:42,280 --> 00:43:46,520
Have been in so much trouble. 
I'm actually glad social media 

848
00:43:46,520 --> 00:43:50,240
and AI did not exist when I was 
in school because I would have 

849
00:43:50,240 --> 00:43:53,280
gotten in so much trouble. 
Oh, social media. 

850
00:43:53,280 --> 00:43:54,920
Certainly existed when I was in 
school. 

851
00:43:54,960 --> 00:43:58,240
That one definitely did get me. 
That one definitely did get me 

852
00:43:58,240 --> 00:44:02,120
in trouble. 
Yeah, yeah. 

853
00:44:02,160 --> 00:44:06,160
I didn't have Facebook. 
Until, you know, two years into 

854
00:44:06,160 --> 00:44:09,680
college. 
So I got, I got a little saved 

855
00:44:09,680 --> 00:44:12,160
there. 
Yeah, I I was part of that 

856
00:44:12,160 --> 00:44:13,840
first. 
Generation where where social 

857
00:44:13,840 --> 00:44:16,840
media was like just becoming a 
big thing when I was a teenager 

858
00:44:16,840 --> 00:44:17,680
and in school. 
OK. 

859
00:44:18,480 --> 00:44:19,400
Yeah, that. 
Would be hard. 

860
00:44:19,440 --> 00:44:22,920
I'm like really grateful that 
didn't happen because there 

861
00:44:22,920 --> 00:44:25,840
would have been some drinking 
parties on there and it would 

862
00:44:25,840 --> 00:44:29,480
have been bad. 
It was a strange transitional 

863
00:44:29,480 --> 00:44:31,200
period. 
You know, this stuff existed and

864
00:44:31,200 --> 00:44:33,680
all, all the kids were on it, 
but it was still just a young 

865
00:44:33,680 --> 00:44:38,120
person thing. 
And I mean, mostly it was weird.

866
00:44:38,120 --> 00:44:39,640
If your parents were on 
Facebook, you didn't want them 

867
00:44:39,640 --> 00:44:40,760
on there. 
That was embarrassing, dude. 

868
00:44:40,840 --> 00:44:43,520
That's when Facebook. 
Went downhill is. 

869
00:44:43,520 --> 00:44:46,440
When all. 
The parents joined like it was 

870
00:44:46,440 --> 00:44:49,760
cool, and it was just college 
kids, and then they opened it up

871
00:44:49,760 --> 00:44:52,400
to the world and you're like, 
well, now Facebook's kind of 

872
00:44:52,400 --> 00:44:54,400
lame. 
Yeah. 

873
00:44:55,640 --> 00:45:02,480
Yeah, man, I've got a one year 
old and I I'm excited for him in

874
00:45:02,480 --> 00:45:06,160
many ways growing up in this new
AI world, but at the same time, 

875
00:45:06,160 --> 00:45:07,800
like there's a lot of ways it 
could go wrong. 

876
00:45:08,280 --> 00:45:17,480
Yeah, it's God knows, I think. 
Yeah, in the right hands, it's 

877
00:45:17,480 --> 00:45:20,320
an extremely powerful tool. 
And we can bring all kinds of 

878
00:45:20,320 --> 00:45:25,160
benefits to programmers, to kids
in schools all over the shop. 

879
00:45:25,160 --> 00:45:27,360
But yeah, yeah, it's not hard 
to. 

880
00:45:27,480 --> 00:45:29,440
It's not hard to imagine ways it
can go wrong. 

881
00:45:29,440 --> 00:45:31,400
I mean, we're already seeing 
that in some respects. 

882
00:45:31,400 --> 00:45:35,880
But, yeah, I mean, none of us 
know where this is going, 

883
00:45:35,880 --> 00:45:42,320
really. 
I just feel like, well, we were 

884
00:45:42,320 --> 00:45:44,120
talking about whether or not 
programmers are going to be 

885
00:45:44,120 --> 00:45:46,560
unemployed. 
Like, I don't, I don't know if 

886
00:45:46,640 --> 00:45:50,160
AI is going to replace us, but I
am 100% certain that that like 

887
00:45:51,000 --> 00:45:53,320
if anyone gets, if anyone does 
get replaced, it's going to be 

888
00:45:53,320 --> 00:45:57,080
the people who, who, who are 
late or who are skeptical or 

889
00:45:57,080 --> 00:45:59,760
who, well, even people who are 
technical. 

890
00:46:00,000 --> 00:46:02,280
I I I just feel like but refuse 
to integrate. 

891
00:46:02,400 --> 00:46:06,600
With it a little bit dude, it's 
it's insane whether you like it 

892
00:46:06,600 --> 00:46:10,120
or not. 
Like so I agree with you. 100% 

893
00:46:10,200 --> 00:46:15,680
but it is crazy like the the. 
Holdouts. 

894
00:46:15,680 --> 00:46:18,240
How angry they get when you make
those comments. 

895
00:46:18,760 --> 00:46:22,800
Like dude, I don't know, I just,
I think it's better to enter 

896
00:46:22,800 --> 00:46:25,120
this being as optimistic as 
possible. 

897
00:46:25,440 --> 00:46:28,800
Like none of us know where it's 
going, but I do always think 

898
00:46:28,800 --> 00:46:32,040
there's going to be a technical 
fit for everyone that knows how 

899
00:46:32,040 --> 00:46:33,360
things work. 
You know? 

900
00:46:33,360 --> 00:46:36,720
Like, you might not need a team 
of 20 engineers anymore, but at 

901
00:46:36,720 --> 00:46:41,000
least you'll have A-Team of 
like, really, like maybe five 

902
00:46:41,000 --> 00:46:42,720
really smart guys that all work 
together. 

903
00:46:43,080 --> 00:46:45,400
And then I think we'll just have
more and more companies 

904
00:46:45,400 --> 00:46:50,480
springing up that can be more 
and more useful solutions to 

905
00:46:50,480 --> 00:46:53,680
problems that we have. 
Yeah, you know when. 

906
00:46:53,680 --> 00:46:56,120
You and I met at X Mex. 
What was that December? 

907
00:46:56,680 --> 00:46:59,600
I remember there was a speaker 
who asked at some point, how 

908
00:46:59,600 --> 00:47:02,360
many of you do not use AI at all
to write code? 

909
00:47:02,600 --> 00:47:05,600
And maybe like 10 to 15% of the 
audience put their hands up, 

910
00:47:05,600 --> 00:47:06,960
which, you know, it's a 
minority. 

911
00:47:06,960 --> 00:47:11,200
But I mean, first of all, this 
isn't, this isn't a, this isn't 

912
00:47:11,200 --> 00:47:13,960
a random sample of developers. 
It's, it's developers who have 

913
00:47:14,200 --> 00:47:16,120
the inclination to attend the 
conferences. 

914
00:47:16,120 --> 00:47:19,600
So I, I feel like that, that 
that's a, that's a select crowd 

915
00:47:19,600 --> 00:47:23,680
there. 
But yeah, like, I just thought, 

916
00:47:23,680 --> 00:47:25,000
really, you're not using it at 
all. 

917
00:47:25,000 --> 00:47:28,040
Like, yeah, I can understand if 
you you're not, you don't want 

918
00:47:28,040 --> 00:47:31,480
to be a maximalist, but that's 
certainly not where I am. 

919
00:47:32,720 --> 00:47:34,160
Yeah. 
Yeah, I don't. 

920
00:47:34,160 --> 00:47:36,000
Know I think it, I think it's 
wild. 

921
00:47:36,000 --> 00:47:42,120
I, I mean, I also have buddies 
that are just like insanely into

922
00:47:42,240 --> 00:47:46,760
AI and you know, running agents 
upon agents and like they can 

923
00:47:46,760 --> 00:47:49,680
barely keep track of what work 
is being done. 

924
00:47:50,040 --> 00:47:53,840
And I'm not there because I feel
like I only can fit so much in 

925
00:47:53,840 --> 00:47:58,480
my brain and I don't want to be 
running all these different 

926
00:47:58,480 --> 00:48:02,320
instances of you. 
Know a Claude? 

927
00:48:02,320 --> 00:48:05,000
Agent building me all these 
things that like I can't like 

928
00:48:05,000 --> 00:48:08,120
keep up with I that overwhelms 
me and maybe, I don't know, 

929
00:48:08,120 --> 00:48:12,440
maybe that's a a bad. 
Thing that I have to get over. 

930
00:48:12,640 --> 00:48:15,600
I have to like let go of some 
control I don't know. 

931
00:48:17,720 --> 00:48:20,320
Yeah. 
I mean, I don't write anything 

932
00:48:20,320 --> 00:48:22,960
by hand anymore. 
I haven't. 

933
00:48:22,960 --> 00:48:27,880
I rarely do. 
I do, but not often, yeah. 

934
00:48:27,960 --> 00:48:32,360
I I mean, I don't AI is a great 
code reviewer, not just a great,

935
00:48:32,560 --> 00:48:35,520
a great code writer. 
Really just the, the, the 

936
00:48:35,520 --> 00:48:39,320
limiting factor is the 
bottleneck is my own brain. 

937
00:48:39,520 --> 00:48:43,000
Like I just try really hard to 
just not let it run off ahead of

938
00:48:43,000 --> 00:48:46,880
me and build things that where I
don't have a where, where, you 

939
00:48:46,880 --> 00:48:48,880
know, I don't have to be 
intimately familiar with every 

940
00:48:48,880 --> 00:48:51,720
last little little line. 
But as soon as that gap starts 

941
00:48:51,720 --> 00:48:55,560
to open up between, between just
just my mental map of the code 

942
00:48:55,560 --> 00:49:01,160
and what it does and where 
everything sits and like, yeah, 

943
00:49:02,680 --> 00:49:05,480
so yeah, I am not in that stage 
either like I am. 

944
00:49:05,840 --> 00:49:09,640
I am working all day with AI, 
with cloud code codecs open in 

945
00:49:09,640 --> 00:49:12,760
multiple tabs at once and having
a lot of things run. 

946
00:49:12,760 --> 00:49:16,720
But that like that more 
complicated kind of pipeline 

947
00:49:16,720 --> 00:49:19,880
where you're like, you've got 
some big elaborate set up where 

948
00:49:19,880 --> 00:49:21,880
different agents with different 
specialties. 

949
00:49:23,800 --> 00:49:26,200
Maybe we'll get there 
eventually, but I don't know. 

950
00:49:26,200 --> 00:49:29,360
I feel like that just runs too 
far ahead of me and I don't 

951
00:49:29,360 --> 00:49:31,560
trust it yet. 
Me too. 

952
00:49:31,840 --> 00:49:35,000
I just. 
Imagine like these like vibe 

953
00:49:35,000 --> 00:49:40,040
coder agent orchestrator guys 
just going to be like on a spin 

954
00:49:40,120 --> 00:49:43,960
a spin board like a DJ soon, 
just like, you know, doing the 

955
00:49:43,960 --> 00:49:47,760
thing just blasting through. 
Tasks. 

956
00:49:47,760 --> 00:49:51,080
And using a what? 
I don't even know what you call,

957
00:49:51,080 --> 00:49:54,360
what do you call like the DJ 
where you have the spin board in

958
00:49:54,360 --> 00:49:55,520
front of you? 
Is that what it's called? 

959
00:49:55,520 --> 00:50:00,560
I don't know the spin board. 
What like the like a like a DJ 

960
00:50:00,560 --> 00:50:01,920
like? 
Who's mixing music? 

961
00:50:01,920 --> 00:50:04,520
Like isn't, isn't that just a 
DJ? 

962
00:50:05,120 --> 00:50:08,960
Yeah, but what do you call like?
The actual like the the record 

963
00:50:08,960 --> 00:50:11,320
part that spins is that. 
Does that have a name? 

964
00:50:11,320 --> 00:50:13,520
I don't even know. 
I don't know. 

965
00:50:14,200 --> 00:50:16,720
I just imagine our. 
Keyboards changing to something 

966
00:50:16,720 --> 00:50:19,800
like that right? 
For like swiping between tasks 

967
00:50:19,800 --> 00:50:23,680
and agents and stuff faster and 
approve approve instead of 

968
00:50:23,680 --> 00:50:26,600
having a full keyboard. 
Yeah, this is. 

969
00:50:26,600 --> 00:50:29,040
Something I I, I've been 
thinking about because, you 

970
00:50:29,040 --> 00:50:31,000
know, like a turntable, that's 
what it's called. 

971
00:50:31,600 --> 00:50:32,920
Thank you. 
Thank you, Brian. 

972
00:50:35,560 --> 00:50:43,120
Yeah, yeah, you know, I, I, I 
feel like sometimes I feel, I 

973
00:50:43,400 --> 00:50:46,560
feel like I just just like the 
way that I am interfacing with 

974
00:50:46,560 --> 00:50:51,520
my AI is definitely sub optimal 
and and that like I might have 

975
00:50:51,520 --> 00:50:55,200
multiple things running in 
different tabs, but like I, I 

976
00:50:55,200 --> 00:50:57,400
have to keep checking, you know,
I'll cycle through my tabs. 

977
00:50:57,400 --> 00:50:58,680
I'll see. 
Is this one done yet? 

978
00:51:00,040 --> 00:51:02,760
It's it's still a very much kind
of pull process. 

979
00:51:05,440 --> 00:51:10,560
Maybe that's just because I you 
know, there's the supposed quote

980
00:51:10,560 --> 00:51:15,000
from Henry Ford, a Ford Motor, 
who said, you know, if I'd given

981
00:51:15,000 --> 00:51:18,400
people what they wanted, people 
would have wanted, they didn't 

982
00:51:18,400 --> 00:51:22,880
want a car faster horse, right? 
Yeah, it still feels like that a

983
00:51:22,880 --> 00:51:25,440
bit with. 
With a lot of the ways that 

984
00:51:25,440 --> 00:51:28,680
we've that we're working with 
AI, like like we're taking all 

985
00:51:28,680 --> 00:51:31,320
these paradigms that already 
existed and then we're just kind

986
00:51:31,320 --> 00:51:33,720
of shoehorning AI into them. 
I'm working with the terminal 

987
00:51:33,720 --> 00:51:37,040
and tabs, which is what I was 
doing before AI, but I don't 

988
00:51:37,040 --> 00:51:40,120
know, I, I feel like it would 
almost make more sense to have 

989
00:51:40,120 --> 00:51:42,840
some like totally new interface 
where, you know, if I was more 

990
00:51:42,840 --> 00:51:46,000
of a technical PM, which I think
is, is, is what we're becoming, 

991
00:51:46,080 --> 00:51:49,840
you know, you know, like AI 
needs to report to me in some 

992
00:51:49,840 --> 00:51:51,320
way. 
Like, rather than me having to 

993
00:51:51,320 --> 00:51:54,080
go back to the tab and see, see,
like, you know, is this thing 

994
00:51:54,080 --> 00:51:55,800
done yet? 
You know, why can't I have some 

995
00:51:55,800 --> 00:51:59,720
kind of interface where I set 
one thing off, I mean, an inbox 

996
00:51:59,720 --> 00:52:02,000
really. 
It's like, OK, this AI is done. 

997
00:52:02,000 --> 00:52:03,640
It's come in. 
Here's here's the report. 

998
00:52:04,040 --> 00:52:07,120
Here's some stuff to jog my 
memory about what, what exactly 

999
00:52:07,120 --> 00:52:09,280
was in that conversation. 
And I don't know, send that one 

1000
00:52:09,280 --> 00:52:11,440
back off. 
Like, I mean more sort of, yeah,

1001
00:52:12,600 --> 00:52:14,960
I'm more of an executive, you 
know, just just the reporter 

1002
00:52:14,960 --> 00:52:19,400
coming to me with, with 
communicating as well as they 

1003
00:52:19,400 --> 00:52:21,120
can and I send them off on their
way again. 

1004
00:52:22,680 --> 00:52:26,720
Yeah. 
I, I, I think it's also kind of 

1005
00:52:26,720 --> 00:52:29,480
weird in my, in my mind a bit 
that that like there's no new 

1006
00:52:29,480 --> 00:52:33,080
programming language, right? 
Like like we're still, we're 

1007
00:52:33,080 --> 00:52:35,040
still just using all these same 
programming languages that 

1008
00:52:35,040 --> 00:52:40,320
existed anyway. 
Like what like like is there not

1009
00:52:40,320 --> 00:52:42,640
some kind of, is there not the 
sort of like like AI optimal 

1010
00:52:42,640 --> 00:52:45,280
programming language? 
Maybe someone's working on that,

1011
00:52:45,280 --> 00:52:48,000
I don't know, but I thought it 
was just markdown files. 

1012
00:52:49,520 --> 00:52:51,120
Maybe. 
I don't know. 

1013
00:52:51,120 --> 00:52:56,080
I, I, I think in the long run, 
I, I, I think, I, I think 

1014
00:52:56,080 --> 00:52:57,840
eventually there's just going to
be new things. 

1015
00:52:57,840 --> 00:53:00,000
There's going to be some 
visionaries and some people who 

1016
00:53:00,000 --> 00:53:03,520
just come up with very smart, 
very innovative ways that that, 

1017
00:53:04,160 --> 00:53:07,480
you know, once it gets there, 
we'll think like, how did we 

1018
00:53:07,480 --> 00:53:08,960
have? 
It's going to look so obvious in

1019
00:53:08,960 --> 00:53:12,840
retrospect. 
It's like, you know, it's like 

1020
00:53:13,360 --> 00:53:16,760
Brian said, Elixir is the AI. 
Optimal programming language. 

1021
00:53:18,840 --> 00:53:20,840
Certainly Elixir is optimal for 
a lot of things. 

1022
00:53:21,160 --> 00:53:22,920
I'll tell you what gives me 
confidence in Elixir. 

1023
00:53:24,280 --> 00:53:27,200
I think just Elixir has got some
fantastic brains behind it, man,

1024
00:53:27,200 --> 00:53:32,240
Like Jose, Chris McCord that 
they have just, I think they are

1025
00:53:32,400 --> 00:53:33,680
smart guys. 
That's for. 

1026
00:53:33,880 --> 00:53:36,040
Sure. 
Yeah, even. 

1027
00:53:36,080 --> 00:53:38,280
Little things you know, it's 
like now, now you when you 

1028
00:53:38,360 --> 00:53:41,160
generate a new Phoenix app, it 
comes with an agent's MD file, 

1029
00:53:41,160 --> 00:53:43,280
which is pretty obvious, right? 
Like why would you not want an 

1030
00:53:43,280 --> 00:53:45,520
agent's MD file? 
But as far as I'm aware, none of

1031
00:53:45,520 --> 00:53:47,960
the other like like competitor 
frameworks do that yet. 

1032
00:53:48,200 --> 00:53:52,120
And I think it's good to see 
people thinking about this 

1033
00:53:52,120 --> 00:53:56,160
because, yeah, I think it's cool
to see. 

1034
00:53:56,400 --> 00:54:00,800
It's cool to see like. 
Joe Zay and like, Chris being so

1035
00:54:00,960 --> 00:54:04,800
like, forward thinking with AI 
instead of pushing back on it 

1036
00:54:05,000 --> 00:54:08,000
'cause like, yeah, seeing like 
Tide Wave, like Tide wave's 

1037
00:54:08,000 --> 00:54:09,640
amazing. 
I don't know if you've used like

1038
00:54:09,640 --> 00:54:12,920
the browser version of it, but I
played around with it, yeah. 

1039
00:54:13,160 --> 00:54:14,800
I freaking love it. 
It does almost. 

1040
00:54:14,800 --> 00:54:20,320
All my UI, like my live view 
programming and like, 'cause I'm

1041
00:54:20,320 --> 00:54:23,400
not a front end dev. 
So like that's like my 

1042
00:54:23,400 --> 00:54:24,920
restriction. 
I'm very slow. 

1043
00:54:25,280 --> 00:54:27,920
Like I can do front end dev but 
I'm very slow. 

1044
00:54:28,280 --> 00:54:32,080
But like now like I can just 
click on components like tell it

1045
00:54:32,080 --> 00:54:34,720
what I want to do and like 
Tidwiff does it. 

1046
00:54:34,720 --> 00:54:37,280
It's amazing. 
Yeah, A. 

1047
00:54:37,280 --> 00:54:40,160
Big part of figuring out how to 
work with AI effectively, it's 

1048
00:54:40,160 --> 00:54:42,520
closing the loop. 
You know, I in the perfect 

1049
00:54:42,520 --> 00:54:44,880
world, I would just be able to 
one shot everything. 

1050
00:54:44,880 --> 00:54:47,280
I would send the AI off on its 
way and it would come back with,

1051
00:54:47,280 --> 00:54:48,840
you know, everything totally 
done. 

1052
00:54:48,840 --> 00:54:51,560
But you know, every time that 
the AI comes back to and it and 

1053
00:54:51,560 --> 00:54:54,480
it needs a response from me, I 
need to do something manually, 

1054
00:54:54,480 --> 00:54:57,640
give it a screenshot, connect it
to something, paste some text 

1055
00:54:57,640 --> 00:54:59,600
in, answer it's question or 
whatever. 

1056
00:54:59,600 --> 00:55:02,440
Like like every time that it 
needs to pause and have me in 

1057
00:55:02,440 --> 00:55:04,800
the loop that that's like, 
that's something that I want to 

1058
00:55:04,800 --> 00:55:06,440
optimize away. 
You know, right, We're the 

1059
00:55:06,440 --> 00:55:08,600
bottleneck at that point. 
Yeah. 

1060
00:55:08,640 --> 00:55:10,680
You know that that that. 
And, and part of that part of 

1061
00:55:10,680 --> 00:55:13,120
that is closing the loop. 
You know, it, it's just giving 

1062
00:55:13,120 --> 00:55:16,280
the AI the tools it needs so 
that it doesn't have to have my 

1063
00:55:16,280 --> 00:55:17,640
input. 
You know, I feel like every time

1064
00:55:17,640 --> 00:55:21,760
I copy and paste something that,
that that to me is, is, is, is 

1065
00:55:21,760 --> 00:55:24,360
an indicator that things are 
not, are not wired up as well as

1066
00:55:24,360 --> 00:55:26,520
they could be. 
You know, that I, the AI should 

1067
00:55:26,520 --> 00:55:28,480
have some way to access the 
thing that I'm copying. 

1068
00:55:29,400 --> 00:55:33,520
And yeah, you know that that 
Tidewave, I haven't used 

1069
00:55:33,520 --> 00:55:34,880
Tidewave, Tidewave a huge 
amount. 

1070
00:55:34,920 --> 00:55:36,360
I've, I've only really played 
around with it. 

1071
00:55:36,360 --> 00:55:38,720
But like, I mean, it seems to me
that that's the fundamentally 

1072
00:55:38,720 --> 00:55:40,320
the problem that Tidewave is 
trying to solve. 

1073
00:55:40,760 --> 00:55:42,880
You know, it gives, it gives 
your, it gives the model access 

1074
00:55:42,880 --> 00:55:47,560
to your, your, your UI in the 
browser, to your logs to, to, 

1075
00:55:47,720 --> 00:55:50,960
you know, to your database. 
Yeah, yeah. 

1076
00:55:50,960 --> 00:55:52,800
It's just trying to connect as 
many things as possible. 

1077
00:55:52,800 --> 00:55:59,560
And yeah, like that, again, that
clear clearly to me, that's the 

1078
00:55:59,560 --> 00:56:00,800
future. 
This this is what we want to be 

1079
00:56:00,800 --> 00:56:02,960
doubling down on, like, like 
just solving those kind of 

1080
00:56:02,960 --> 00:56:04,000
problems, connecting those 
things. 

1081
00:56:04,000 --> 00:56:07,360
So it's great to see that that 
Jose is like going all in on 

1082
00:56:07,360 --> 00:56:09,960
that, going all in on that. 
Yeah. 

1083
00:56:10,160 --> 00:56:11,120
I'm sure he's going to do a 
great. 

1084
00:56:11,120 --> 00:56:12,920
Job of it because he's done a 
great job of everything else 

1085
00:56:12,920 --> 00:56:15,600
he's worked on now the thing 
that like. 

1086
00:56:16,200 --> 00:56:19,440
Is going to make me like I'm 
thinking about this now. 

1087
00:56:19,440 --> 00:56:22,920
Like why isn't there a language 
being developed that is 

1088
00:56:22,920 --> 00:56:26,560
specifically for AI? 
It would just have to be more 

1089
00:56:26,560 --> 00:56:28,600
readable. 
Like I don't even know if it 

1090
00:56:28,600 --> 00:56:31,200
needs to be readable. 
I don't, I don't know, like do 

1091
00:56:31,200 --> 00:56:35,560
we just go to down to bytecode? 
Yeah. 

1092
00:56:35,880 --> 00:56:42,560
Yeah, I don't know. 
Yeah, I, I, I mean, maybe in the

1093
00:56:42,560 --> 00:56:45,960
future we won't it we won't even
need a language of the human 

1094
00:56:45,960 --> 00:56:47,320
that it won't matter, right. 
The. 

1095
00:56:47,880 --> 00:56:49,280
AI will just be writing the 
bytecode. 

1096
00:56:49,800 --> 00:56:52,720
But even if we get there, like 
I, I, I think there's still 

1097
00:56:52,720 --> 00:56:57,400
reasons to be bullish on Elixir,
even even if we get to the point

1098
00:56:57,400 --> 00:56:59,920
where AI is so good that that 
like, you don't need that. 

1099
00:56:59,920 --> 00:57:01,800
No human would ever need to read
the code. 

1100
00:57:01,800 --> 00:57:04,360
And therefore, like, the things 
that we like about Elix's 

1101
00:57:04,360 --> 00:57:07,920
readability are just irrelevant.
Like, I mean, first of all, 

1102
00:57:07,920 --> 00:57:10,800
first of all, like I already 
said, Elix has got a great team 

1103
00:57:10,800 --> 00:57:12,520
behind it. 
And I think, I think they're 

1104
00:57:12,520 --> 00:57:14,880
going to navigate these if, if 
anyone's going to figure it out 

1105
00:57:14,880 --> 00:57:17,280
and navigate these choppy 
waters, I think it's going to be

1106
00:57:17,280 --> 00:57:19,000
them. 
But also, first of all, you've 

1107
00:57:19,000 --> 00:57:22,640
got the beam. 
And yeah, the Beam is just a 

1108
00:57:22,640 --> 00:57:24,960
fantastic piece of technology 
for so many reasons. 

1109
00:57:25,520 --> 00:57:30,320
It just gives you you yeah, it, 
it just, it just gives you 

1110
00:57:30,440 --> 00:57:33,040
things for free and it makes 
them so easy that that in 

1111
00:57:33,040 --> 00:57:35,800
another language, even if AI is 
writing it, there's just so much

1112
00:57:36,000 --> 00:57:37,480
extra stuff you got to wire 
together. 

1113
00:57:37,480 --> 00:57:39,720
And and, you know, it's just 
more complicated. 

1114
00:57:40,160 --> 00:57:41,320
I think. 
Just like what? 

1115
00:57:41,320 --> 00:57:43,360
Like what? 
You get with the actor model or 

1116
00:57:43,360 --> 00:57:47,360
like the supervision trees and 
processes and message passing. 

1117
00:57:47,360 --> 00:57:52,360
It fits perfectly with all the 
AI agent solutions that are 

1118
00:57:52,360 --> 00:57:55,200
trying to be solved. 
Absolutely. 

1119
00:57:55,200 --> 00:57:58,400
Yeah, yeah, yeah. 
I would love to see someone 

1120
00:57:58,400 --> 00:58:01,000
like, like try and build 
something like Open Claw or I, I

1121
00:58:01,000 --> 00:58:04,120
don't know, like Elixir clone. 
Maybe somebody's already done 

1122
00:58:04,120 --> 00:58:06,760
that, but I feel like Elixir 
would be a great language to try

1123
00:58:06,760 --> 00:58:08,320
and build something like that 
in. 

1124
00:58:08,600 --> 00:58:10,560
Yeah, let's see what happens 
there. 

1125
00:58:11,080 --> 00:58:15,560
Another great thing about Elixir
is just the tooling, like, you 

1126
00:58:15,560 --> 00:58:19,200
know, just things like mix where
there's just so many things or 

1127
00:58:19,840 --> 00:58:21,720
so many things that you just get
really easily. 

1128
00:58:21,720 --> 00:58:23,760
There's the standardization 
across projects. 

1129
00:58:24,520 --> 00:58:27,960
I mean, I, I work a lot with 
JavaScript at my day job and 

1130
00:58:28,200 --> 00:58:31,360
many previous jobs. 
And I mean, in the web world, 

1131
00:58:31,360 --> 00:58:35,000
we're all stuck with JavaScript 
for probably for the foreseeable

1132
00:58:35,000 --> 00:58:37,480
future. 
But oh, it just frustrates me. 

1133
00:58:38,200 --> 00:58:40,480
There's just Dude, I hate 
JavaScript. 

1134
00:58:40,480 --> 00:58:44,080
And I can't believe I used to do
it almost daily. 

1135
00:58:44,160 --> 00:58:47,600
I mean, I still write JavaScript
quite a bit like for browser 

1136
00:58:47,600 --> 00:58:52,480
side things that, you know, you 
just can't really do like things

1137
00:58:52,480 --> 00:58:55,480
you don't want to run to the 
server for with a live view app.

1138
00:58:56,080 --> 00:59:01,120
But mix is amazing and a little 
hack that I just learned. 

1139
00:59:01,200 --> 00:59:06,440
Like just did you know that for 
mix test, you can throw the flag

1140
00:59:06,440 --> 00:59:09,800
on their slowest and get back 
the slow like you can do like 

1141
00:59:09,800 --> 00:59:15,320
dash, dash slowest 10 and it'll 
return the 10 slowest tests to 

1142
00:59:15,320 --> 00:59:16,600
you. 
I didn't know that till like a 

1143
00:59:16,600 --> 00:59:18,960
day ago. 
Yeah, I think I. 

1144
00:59:18,960 --> 00:59:21,160
Have seen that, Yeah, exactly. 
It's great. 

1145
00:59:21,160 --> 00:59:23,920
Which is wild like. 
What like because I was trying 

1146
00:59:23,920 --> 00:59:28,240
to optimize like some CI stuff 
and we had slow running tests 

1147
00:59:28,240 --> 00:59:31,720
and that helped me find the 
slowest running tests and then I

1148
00:59:31,720 --> 00:59:33,720
could send Claude. 
I was like all right, help me, 

1149
00:59:33,720 --> 00:59:37,040
you know, fix the. 
Bottlenecks here we want. 

1150
00:59:37,040 --> 00:59:40,360
To speed these tests up and it 
throws a couple ideas at me and 

1151
00:59:40,360 --> 00:59:42,320
I'm like, oh, that's a great 
solution. 

1152
00:59:43,600 --> 00:59:45,640
Yeah, that's the great. 
Thing about it, Elixir, just 

1153
00:59:45,640 --> 00:59:48,480
just this default tooling that's
baked into the language. 

1154
00:59:48,480 --> 00:59:54,840
It's so it's, yeah, there's so 
many like great nifty, like like

1155
00:59:54,840 --> 00:59:57,800
kind of more less use niche 
little features like that, but 

1156
00:59:57,800 --> 00:59:59,400
also just the main stuff that 
you use all the time. 

1157
00:59:59,400 --> 01:00:03,120
It's so it's just excellently 
designed, you know, no one feels

1158
01:00:03,120 --> 01:00:06,080
any need to to make a competitor
to try and do it better. 

1159
01:00:06,080 --> 01:00:07,320
That's a big problem with 
JavaScript. 

1160
01:00:07,320 --> 01:00:09,560
There's just so much churn, so 
much fragmentation. 

1161
01:00:09,560 --> 01:00:11,240
Every project does something 
different, does things 

1162
01:00:11,240 --> 01:00:13,360
differently. 
Dependency management. 

1163
01:00:13,600 --> 01:00:16,800
In JavaScript is an absolute 
nightmare and drives me crazy. 

1164
01:00:17,080 --> 01:00:20,520
Yeah. 
You know, dependency management 

1165
01:00:20,520 --> 01:00:24,160
in Elixir is a really good 
example of something that, yeah,

1166
01:00:24,160 --> 01:00:26,840
where I think it clearly took a 
lot of inspiration from Ruby. 

1167
01:00:26,840 --> 01:00:29,840
Ruby dependency management in 
Ruby with Bundler and Gem, it's 

1168
01:00:29,840 --> 01:00:31,280
great. 
Works really well, Certainly 

1169
01:00:31,280 --> 01:00:32,560
works a lot better than the 
JavaScript. 

1170
01:00:32,560 --> 01:00:36,040
And yeah, yeah, it's very, if 
you know how to do it in Ruby, 

1171
01:00:36,040 --> 01:00:37,840
Elixir does things in a pretty 
similar way. 

1172
01:00:39,160 --> 01:00:41,720
Those are the kind of things 
where like even if AI is writing

1173
01:00:41,720 --> 01:00:45,000
all of your code and you don't 
have to read anything, just like

1174
01:00:45,120 --> 01:00:48,400
having that just great 
foundation of great tooling, 

1175
01:00:48,800 --> 01:00:51,720
just great ways to just set up 
the core of your app, the 

1176
01:00:51,720 --> 01:00:54,520
architecture, the dependencies, 
the environments. 

1177
01:00:54,880 --> 01:00:56,560
Environments are another great 
example. 

1178
01:00:56,560 --> 01:00:59,280
You know, you work with Phoenix.
I mean, Rails did it like this 

1179
01:00:59,280 --> 01:01:02,160
as well. 
But yeah, just, I don't know, I,

1180
01:01:02,160 --> 01:01:05,080
I work on JavaScript projects 
where where like it just, it's 

1181
01:01:05,080 --> 01:01:08,120
just so much work to like set up
a test environment or, you know,

1182
01:01:08,600 --> 01:01:12,000
set up staging or, or or get a 
like what? 

1183
01:01:13,800 --> 01:01:17,680
I just, I just say I may have a 
bit of a reputation on sadly, I 

1184
01:01:17,680 --> 01:01:19,680
don't get to use much Elixir at 
my current gig. 

1185
01:01:19,680 --> 01:01:22,720
I may have a bit of a reputation
on the team for for the guy, the

1186
01:01:22,720 --> 01:01:24,280
guy who won't shut up about 
Elixir. 

1187
01:01:25,160 --> 01:01:27,440
Hopefully I can convince someone
on the team to. 

1188
01:01:28,080 --> 01:01:30,400
We'll get there eventually. 
We we've got a lot of projects 

1189
01:01:30,400 --> 01:01:32,640
and we're starting new ones. 
So yeah, you got to bring the 

1190
01:01:32,640 --> 01:01:34,880
Elixir Kool-aid. 
And really push that down their 

1191
01:01:34,880 --> 01:01:36,920
throats, you know, yeah, for 
sure. 

1192
01:01:36,920 --> 01:01:37,800
But you know. 
Things, things like 

1193
01:01:37,800 --> 01:01:39,800
environments, we do all this 
stuff with JavaScript where it's

1194
01:01:39,800 --> 01:01:44,000
just such a pain. 
And I'm like, I, I mean, not, 

1195
01:01:44,000 --> 01:01:45,960
not to keep banging on about 
Elixir, but like, you guys 

1196
01:01:46,000 --> 01:01:47,760
realize this is trivially easy, 
right? 

1197
01:01:47,840 --> 01:01:49,760
Like why does it have to be so 
difficult? 

1198
01:01:50,600 --> 01:01:54,600
You know Elixir, you get it for 
free or to see like all. 

1199
01:01:54,600 --> 01:02:01,240
These LLM companies too, talking
about recreating the BEAM VM 

1200
01:02:01,280 --> 01:02:05,160
essentially in Erlang and like 
guys, all the solutions you're 

1201
01:02:05,160 --> 01:02:09,200
trying to rebuild already exist 
since what, 1986? 

1202
01:02:10,360 --> 01:02:12,040
Yeah, that's a. 
Lot that's the classic quote 

1203
01:02:12,040 --> 01:02:14,760
from Joe Armstrong, one of the 
creators of Erlang. 

1204
01:02:14,760 --> 01:02:18,080
I'm I'm I'm going to butcher it,
but you know, he said Oh yeah, 

1205
01:02:18,160 --> 01:02:21,080
every sufficiently complex 
system contains a a buggy 

1206
01:02:21,080 --> 01:02:22,520
implementation of half of 
Erlang. 

1207
01:02:22,920 --> 01:02:25,920
A buggy important Yeah, I. 
Love that I have heard that one.

1208
01:02:26,200 --> 01:02:27,520
It's good. 
It's true, though. 

1209
01:02:27,560 --> 01:02:30,920
It's just like guys, it's been 
around and it works. 

1210
01:02:31,760 --> 01:02:36,280
Yeah, I. 
Despite being being like a big 

1211
01:02:36,280 --> 01:02:40,320
Elixir guy, I, I, I have never 
written a line of Erlang, you 

1212
01:02:40,320 --> 01:02:43,240
know, or maybe you, you, you 
have to kind of dive into it 

1213
01:02:43,240 --> 01:02:45,320
every now and then because 
there's just certain libraries 

1214
01:02:45,320 --> 01:02:47,720
in I've used. 
Erlang libraries. 

1215
01:02:47,720 --> 01:02:49,960
But yeah, I haven't really 
written Erlang code. 

1216
01:02:50,200 --> 01:02:52,680
Yeah, Erlang's got a weird. 
Syntax. 

1217
01:02:52,680 --> 01:02:54,480
I want it. 
I think it's based on Prologue, 

1218
01:02:54,480 --> 01:02:55,960
which I don't. 
Does anyone write anything in 

1219
01:02:55,960 --> 01:02:59,720
Prologue anymore? 
I, I, I Yeah, but there's a 

1220
01:02:59,720 --> 01:03:01,920
couple maybe. 
That's my Discord server that 

1221
01:03:01,920 --> 01:03:05,840
will write some Erlang and like 
one one of them he was like 

1222
01:03:05,840 --> 01:03:09,920
sharing some AI generated code 
and he's like, it is it can 

1223
01:03:09,920 --> 01:03:12,920
finally write Erlang. 
And he was like, so impressed 

1224
01:03:12,920 --> 01:03:15,360
because I guess like just a few 
months ago, it could barely do 

1225
01:03:15,360 --> 01:03:19,720
like a Hello World app. 
Yeah, I do wonder. 

1226
01:03:19,720 --> 01:03:22,400
Maybe one reason the. 
BEAM never really like like blew

1227
01:03:22,400 --> 01:03:24,760
up and became as big as it 
deserves to be is because Erlang

1228
01:03:24,760 --> 01:03:28,360
is just weird. 
It's like coming from C or Java 

1229
01:03:28,360 --> 01:03:30,760
or maybe these other languages 
that were more popular when 

1230
01:03:30,800 --> 01:03:33,680
Erlang was young. 
Like Erlang is just not, not 

1231
01:03:33,680 --> 01:03:36,560
even not even the BEAM. 
But yeah, it lacked the modern 

1232
01:03:36,560 --> 01:03:39,480
syntax. 
Feel right, like any modern 

1233
01:03:39,480 --> 01:03:41,480
programming language. 
And I think that's why people 

1234
01:03:41,480 --> 01:03:44,640
just didn't like it. 
Yeah. 

1235
01:03:46,840 --> 01:03:49,080
Yeah. 
And then Elixir kind of solved 

1236
01:03:49,080 --> 01:03:54,360
that because Elixir is just. 
Fun like that the the syntax of 

1237
01:03:54,360 --> 01:03:58,520
Elixir made me fall in love with
Elixir before all the other 

1238
01:03:58,520 --> 01:04:02,120
reasons. 
You know, like I enjoyed writing

1239
01:04:02,120 --> 01:04:06,560
it before I understood the power
of like a Gen. server or message

1240
01:04:06,560 --> 01:04:09,520
passing. 
Oh yeah, absolutely. 

1241
01:04:09,520 --> 01:04:11,880
Yeah, like, like, like I said, 
when I first was getting into 

1242
01:04:11,880 --> 01:04:14,040
Elixir, I didn't even, I hadn't 
even heard of any of that. 

1243
01:04:14,040 --> 01:04:18,160
I I got there in the end. 
But, you know, I mean it, it's 

1244
01:04:18,160 --> 01:04:20,480
relatively rare for me to 
actually like need any of that 

1245
01:04:20,480 --> 01:04:23,160
stuff directly. 
But, you know, when working with

1246
01:04:23,160 --> 01:04:25,920
something like Phoenix, you 
know, having worked with Phoenix

1247
01:04:25,920 --> 01:04:27,680
for a while, I've written 
tutorials about it. 

1248
01:04:28,360 --> 01:04:28,960
Yeah. 
Yeah. 

1249
01:04:28,960 --> 01:04:34,280
Like once you understand OTP and
you just you you can dig a bit 

1250
01:04:34,280 --> 01:04:38,160
more into like how Phoenix works
under the hood or even things 

1251
01:04:38,160 --> 01:04:40,320
that are a little bit less under
the hood, like things you things

1252
01:04:40,320 --> 01:04:41,840
you'll actually use yourself. 
Like pops up. 

1253
01:04:42,960 --> 01:04:44,400
Yeah. 
You just see how. 

1254
01:04:44,400 --> 01:04:45,960
Yeah, yeah. 
The beam makes this so easy. 

1255
01:04:45,960 --> 01:04:48,800
The message passing format. 
Yeah, it's great. 

1256
01:04:48,800 --> 01:04:50,920
It it, it's just, it works so 
well. 

1257
01:04:51,240 --> 01:04:52,800
Yeah. 
Well, and then just look, look 

1258
01:04:52,800 --> 01:04:54,640
how well like. 
Live view works too. 

1259
01:04:54,640 --> 01:04:58,960
And that's a an example of just 
using Gen. servers and things, 

1260
01:04:58,960 --> 01:05:01,320
right? 
Like so you're leveraging all 

1261
01:05:01,320 --> 01:05:05,040
this thing, all these things 
just using like the Phoenix 

1262
01:05:05,040 --> 01:05:08,240
library. 
Yeah, yeah, absolutely. 

1263
01:05:08,240 --> 01:05:13,560
Yeah, Yeah, it's pretty cool. 
The, I always tell people, 

1264
01:05:13,560 --> 01:05:16,440
'cause like I would say there's 
like a mix of people. 

1265
01:05:16,440 --> 01:05:20,120
Some people discover Elixir. 
Like I, I feel like you and I 

1266
01:05:20,120 --> 01:05:23,440
have very similar stories. 
Like I was just like writing the

1267
01:05:23,440 --> 01:05:25,760
syntax. 
I heard it was cool. 

1268
01:05:25,760 --> 01:05:28,320
So I wanted to like check it out
and see how I liked it. 

1269
01:05:28,800 --> 01:05:30,600
And then I got into the 
complicated stuff. 

1270
01:05:30,600 --> 01:05:33,480
When I ran into problems in my 
own projects, I was like, OK, 

1271
01:05:33,800 --> 01:05:37,640
it's time to learn what this OTP
stuff is and like, so like I 

1272
01:05:37,640 --> 01:05:41,200
slowly gravitated towards it. 
But then there's like the 

1273
01:05:41,200 --> 01:05:43,840
reverse, the exact opposite. 
People come for the shiny 

1274
01:05:43,840 --> 01:05:46,480
things, they hear about the 
Genservers and they want to 

1275
01:05:46,480 --> 01:05:49,800
write them right away. 
And I feel like when you when 

1276
01:05:49,800 --> 01:05:55,000
you want to learn Elixir that 
way, you write a lot of you. 

1277
01:05:55,200 --> 01:05:56,600
You abuse. 
Genservers. 

1278
01:05:56,600 --> 01:05:58,880
So because you try to make 
everything stateful, especially 

1279
01:05:58,880 --> 01:06:01,960
if you're coming from like an 
object oriented past, I think 

1280
01:06:01,960 --> 01:06:06,040
you abuse Genservers and kind of
like over create things that you

1281
01:06:06,040 --> 01:06:08,520
don't really. 
Yeah, I, I yeah, I can. 

1282
01:06:08,520 --> 01:06:11,680
Imagine I, I would never tell 
anybody who is new to Elixir to 

1283
01:06:11,680 --> 01:06:15,560
start with OTP and all of that 
like that's that, that that's 

1284
01:06:15,560 --> 01:06:18,160
the last you need to learn it 
eventually. 

1285
01:06:18,160 --> 01:06:21,360
But it it's it's very low down 
the list of the order which you 

1286
01:06:21,360 --> 01:06:23,800
should learn things in. 
I agree, but it's still like 

1287
01:06:23,800 --> 01:06:25,760
that. 
Shiny object that everyone wants

1288
01:06:25,760 --> 01:06:27,600
to touch, right? 
Because it's so cool. 

1289
01:06:29,240 --> 01:06:31,080
I feel like the best 
introduction to OTP. 

1290
01:06:31,080 --> 01:06:32,800
I'm going to teach it like this 
in my courses. 

1291
01:06:32,800 --> 01:06:35,240
Don't don't start by like 
getting into the nitty gritty of

1292
01:06:35,240 --> 01:06:38,400
precisely how how Genservers and
what not use like like pops up 

1293
01:06:38,400 --> 01:06:41,800
is a great example. 
You know, build something with 

1294
01:06:41,800 --> 01:06:44,640
Phoenix, build the web app like 
pops up is a concept that is 

1295
01:06:44,720 --> 01:06:46,960
certainly not unique to Phoenix 
or to OTP. 

1296
01:06:46,960 --> 01:06:50,920
But, you know, don't, don't 
start with with by getting too 

1297
01:06:50,920 --> 01:06:54,680
deep into just these kind of 
esoteric underlying concepts. 

1298
01:06:54,680 --> 01:06:56,440
Let's build something real, you 
know, let's build something 

1299
01:06:56,440 --> 01:06:58,440
where you actually understand 
the value and you can see it 

1300
01:06:58,440 --> 01:06:59,680
being used. 
Yeah. 

1301
01:06:59,680 --> 01:07:02,240
You know, Hey, hey, pub sub. 
So in one of my courses, I'll 

1302
01:07:02,240 --> 01:07:07,160
teach you to build a Slack loan.
So you send a message on on 

1303
01:07:07,160 --> 01:07:10,240
Slack and everybody else who is 
currently in that room needs to 

1304
01:07:10,240 --> 01:07:11,800
see that message appear 
immediately. 

1305
01:07:12,240 --> 01:07:14,680
So yeah, that's the pub sub 
pattern right there. 

1306
01:07:14,680 --> 01:07:17,160
You know, one person sends it, 
it gets fanned out to to all the

1307
01:07:17,160 --> 01:07:20,160
recipients. 
So yeah, let's, let's use 

1308
01:07:20,160 --> 01:07:21,560
Phoenix dot pubsub, see how that
works. 

1309
01:07:21,560 --> 01:07:24,920
And I'll explain to you, yeah, 
what's actually what's actually 

1310
01:07:24,920 --> 01:07:26,480
going on here. 
We're sending messages from one 

1311
01:07:26,480 --> 01:07:28,320
process to the other. 
Oh, now here's a good 

1312
01:07:28,320 --> 01:07:31,080
opportunity to talk about 
processes, OTP Genservers. 

1313
01:07:31,800 --> 01:07:34,400
But yeah, I mean, that's how I 
like to learn. 

1314
01:07:34,400 --> 01:07:36,960
You know, I, I, I'm going to 
remember it and understand it a 

1315
01:07:36,960 --> 01:07:40,040
lot better if I'm learning it in
the context of something that is

1316
01:07:40,040 --> 01:07:42,680
actually like practical and real
and that I'm going to use. 

1317
01:07:43,000 --> 01:07:45,840
And it's it's not, it's not just
the, the kind of theory, you 

1318
01:07:45,840 --> 01:07:48,360
know, I'm going to make it stick
in my head with something real. 

1319
01:07:48,360 --> 01:07:51,200
I always try to like with my. 
Tutorials that I've made in the 

1320
01:07:51,200 --> 01:07:56,000
past, I've always tried to build
something cool along the way and

1321
01:07:56,000 --> 01:07:59,360
as we touch on things, explain 
it enough to use it. 

1322
01:07:59,720 --> 01:08:02,600
Like you don't necessarily have 
to know the whole theory behind 

1323
01:08:02,600 --> 01:08:06,600
why it exists, right? 
Like you want to build something

1324
01:08:06,600 --> 01:08:10,160
and if you have to do the deeper
dive, you'll get there 

1325
01:08:10,160 --> 01:08:11,440
eventually because you'll need 
to. 

1326
01:08:11,440 --> 01:08:14,760
Like you have to when you have 
to understand something more, I 

1327
01:08:14,760 --> 01:08:17,840
feel like you naturally will go 
down that path. 

1328
01:08:19,040 --> 01:08:21,240
Yeah, this is. 
One of the things I enjoy about 

1329
01:08:21,240 --> 01:08:24,960
making courses and teaching, 
it's, it's just an interesting 

1330
01:08:24,960 --> 01:08:30,279
challenge to, to take what I 
know, but put myself in the 

1331
01:08:30,279 --> 01:08:32,520
perspective of someone who 
doesn't and just try and figure 

1332
01:08:32,520 --> 01:08:35,720
out the that what's the, what's 
the optimal order? 

1333
01:08:35,720 --> 01:08:37,240
How, how can I explain 
everything in the. 

1334
01:08:37,479 --> 01:08:41,279
It's in a way it's almost 
similar to programming in that 

1335
01:08:41,279 --> 01:08:44,560
I'm trying to like, I'm trying 
to determine the, the, the the 

1336
01:08:44,560 --> 01:08:47,439
clearest possible instructions. 
Like, you know, what's, what's 

1337
01:08:47,439 --> 01:08:49,840
the flow that I'm going to send 
the reader through to, to. 

1338
01:08:50,680 --> 01:08:53,120
Yeah, yeah. 
Come out at the end with, with 

1339
01:08:53,120 --> 01:09:00,080
like the optimal output. 
Yeah, I, I, I, I mean, like you 

1340
01:09:00,080 --> 01:09:02,600
said, like, like I, I, I like to
do it. 

1341
01:09:03,720 --> 01:09:05,840
Yeah. 
It's just the interweaving of 

1342
01:09:05,840 --> 01:09:09,240
kind of theory and practice. 
Like I, I, I, you know, it's 

1343
01:09:09,240 --> 01:09:10,479
like we're going to build 
something real. 

1344
01:09:10,479 --> 01:09:13,359
And I'm going to explain to you 
the minimum or, or, you know, 

1345
01:09:13,520 --> 01:09:16,439
maybe close to the minimum of 
what you actually need to, like,

1346
01:09:16,560 --> 01:09:18,160
you know, you're going to build 
it, you're going to see how it 

1347
01:09:18,160 --> 01:09:18,800
works. 
It's real. 

1348
01:09:18,800 --> 01:09:21,640
But, you know, we're also going 
to, like, understand what we 

1349
01:09:21,640 --> 01:09:22,560
need to know now. 
So. 

1350
01:09:22,680 --> 01:09:25,880
So yeah, well, I mean, if you 
deep dive into. 

1351
01:09:25,920 --> 01:09:30,600
Theory on everything, every, 
every lesson that you teach 

1352
01:09:30,600 --> 01:09:33,760
would be multiple hours long. 
And it's just like I'm so 

1353
01:09:33,760 --> 01:09:36,560
impatient like me as a person, 
like I want to just build 

1354
01:09:36,560 --> 01:09:41,319
something and get it working. 
So that's what I try to do when 

1355
01:09:41,319 --> 01:09:42,640
I teach it's. 
Like let's build. 

1356
01:09:42,640 --> 01:09:46,120
Something as quickly as we can 
and learn along the way, you 

1357
01:09:46,120 --> 01:09:48,200
know? 
Yeah, I. 

1358
01:09:48,720 --> 01:09:52,479
Mean I certainly the act of 
create nothing has helped me 

1359
01:09:52,479 --> 01:09:55,800
learn Phoenix better than than 
than creating tutorials and 

1360
01:09:55,800 --> 01:09:58,120
teaching people. 
Yeah, teaching, because you got 

1361
01:09:58,120 --> 01:09:59,640
to understand. 
Everything you're teaching, 

1362
01:09:59,640 --> 01:10:01,280
right? 
So you're like, oh, gosh, I know

1363
01:10:01,280 --> 01:10:03,600
how to use this, but do I know 
how to explain this? 

1364
01:10:04,760 --> 01:10:05,920
Yeah, yeah. 
Absolutely. 

1365
01:10:05,920 --> 01:10:06,840
Yeah. 
You know, you've got to. 

1366
01:10:09,040 --> 01:10:12,320
Yeah, you've got, you've got to 
know it inside and out. 

1367
01:10:13,240 --> 01:10:15,760
Yeah, Yeah, it's. 
It's hard. 

1368
01:10:15,760 --> 01:10:18,360
I like. 
I definitely. 

1369
01:10:20,240 --> 01:10:23,240
I I get. 
A little burned out with making 

1370
01:10:23,240 --> 01:10:26,720
tutorials. 
I enjoy it but I I think like my

1371
01:10:26,720 --> 01:10:30,200
post editing burns me out and 
then I'm like ah and then I'll 

1372
01:10:30,200 --> 01:10:33,280
just take a few months off and 
not make tutorials and. 

1373
01:10:35,320 --> 01:10:39,640
Yeah, I, I do enjoy it. 
It's rewarding in many ways, but

1374
01:10:39,640 --> 01:10:44,000
it can definitely get tedious. 
Especially things like what? 

1375
01:10:44,040 --> 01:10:46,480
I don't know when a new version 
of Phoenix comes out and I've 

1376
01:10:46,480 --> 01:10:49,000
got to go back and update, 
everything got to be done. 

1377
01:10:49,240 --> 01:10:52,080
I, I, I owe that to my audience.
But yeah, I'm really bad at 

1378
01:10:52,120 --> 01:10:57,000
doing. 
That yeah, yeah, yeah, I 

1379
01:10:57,000 --> 01:11:00,880
probably. 
Don't do it as a speedy enough 

1380
01:11:00,880 --> 01:11:05,960
of a timeline. 
But yeah, I don't even know if 

1381
01:11:05,960 --> 01:11:10,920
I'm going to, I, I don't spend a
huge amount of time on the core 

1382
01:11:10,920 --> 01:11:12,920
stuff these days. 
It's never been my full time 

1383
01:11:12,920 --> 01:11:14,440
gig. 
I mean, it's gone pretty well. 

1384
01:11:15,320 --> 01:11:19,600
It's certainly gone more than 
well enough to like justify. 

1385
01:11:19,600 --> 01:11:22,480
I think the time that I've 
spent, I've put into it, I've 

1386
01:11:22,640 --> 01:11:25,040
been doing it for about. 
I mean, I said I started in the 

1387
01:11:25,040 --> 01:11:26,800
COVID lockdowns. 
That might not be true. 

1388
01:11:27,440 --> 01:11:29,600
I think I like started playing 
around with some stuff and then 

1389
01:11:29,600 --> 01:11:31,520
like, yeah, never really went 
anywhere. 

1390
01:11:31,600 --> 01:11:34,480
Took a couple of years, left it 
to the side for a couple of 

1391
01:11:34,480 --> 01:11:39,760
years and then got back into it.
But yeah, I mean I've made 

1392
01:11:39,760 --> 01:11:44,600
decent money. 
Not enough to, I mean not enough

1393
01:11:44,600 --> 01:11:46,640
to replace a full time software 
engineer's income. 

1394
01:11:46,640 --> 01:11:51,280
But selling courses is hard. 
It's a tough business. 

1395
01:11:51,320 --> 01:11:53,760
But you know the things. 
I think I said this to you at 

1396
01:11:53,840 --> 01:11:58,280
XMEX, like, I mean, I've made 
enough money that I'm, I'm that 

1397
01:11:58,360 --> 01:12:00,800
to be, to be happy with it 
relative to the amount of time I

1398
01:12:00,800 --> 01:12:02,200
put in. 
But also, you know, the things 

1399
01:12:02,200 --> 01:12:04,160
that have come to me from it. 
It's not just financial, you 

1400
01:12:04,160 --> 01:12:06,200
know, it's like, like 
opportunities have come to, to 

1401
01:12:06,200 --> 01:12:07,200
me. 
It's just putting my name out 

1402
01:12:07,200 --> 01:12:09,840
there, putting myself out there,
you know, now, now I'm literally

1403
01:12:09,840 --> 01:12:12,640
the guy who wrote the book and, 
you know, I've got contracts out

1404
01:12:12,640 --> 01:12:15,440
of that. 
I've got, you know, it's, yeah, 

1405
01:12:15,480 --> 01:12:18,160
I, I things, things have come to
me indirectly through it. 

1406
01:12:19,120 --> 01:12:21,480
Yeah, I think that's the biggest
thing like I tried. 

1407
01:12:21,480 --> 01:12:25,400
To sell a like an introductory 
course for a long time and I 

1408
01:12:25,400 --> 01:12:29,360
started to not enjoy it as much.
So I ended up just publishing it

1409
01:12:29,360 --> 01:12:33,440
on YouTube for free. 
And but like to what you're 

1410
01:12:33,440 --> 01:12:37,400
saying is like my free content 
on YouTube and YouTube and this 

1411
01:12:37,400 --> 01:12:41,120
podcast has given me like way 
more opportunities than like the

1412
01:12:41,120 --> 01:12:45,400
financial monetization part 
because one, it like allows me 

1413
01:12:45,400 --> 01:12:46,880
to talk to really awesome 
people. 

1414
01:12:46,880 --> 01:12:50,400
Like I've talked to all the 
cool. 

1415
01:12:50,400 --> 01:12:53,680
People of the Elixir. 
Community on this podcast and 

1416
01:12:53,680 --> 01:12:57,360
that's like that's pretty wild. 
And then like this has 

1417
01:12:57,360 --> 01:13:00,680
definitely gotten me my current 
job and it's gotten me a lot of 

1418
01:13:00,680 --> 01:13:03,400
opportunities. 
Yeah, yeah. 

1419
01:13:03,400 --> 01:13:06,360
Absolutely. 
I I don't know if I'll make 

1420
01:13:06,360 --> 01:13:08,080
another course. 
Partly that's just because I'm 

1421
01:13:08,080 --> 01:13:10,280
super busy. 
You know, it's hard work. 

1422
01:13:10,280 --> 01:13:11,240
It's a lot of work. 
I've. 

1423
01:13:11,240 --> 01:13:12,840
Got a day job, I've got a 
family. 

1424
01:13:13,440 --> 01:13:18,360
But also, I mean, if I was to 
make another course, it would 

1425
01:13:18,360 --> 01:13:20,720
certainly have to be something 
AI related. 

1426
01:13:20,720 --> 01:13:24,920
And I am still really trying to 
figure out like, like how, how 

1427
01:13:24,920 --> 01:13:27,000
does this work? 
You know, like, like how, how do

1428
01:13:27,000 --> 01:13:33,080
we teach agentic engineering? 
I think like, like I said, it's 

1429
01:13:33,080 --> 01:13:35,080
not just because it's not just 
about teaching people how to 

1430
01:13:35,080 --> 01:13:38,320
work with, with the AI itself. 
That's maybe only 20% of it. 

1431
01:13:38,320 --> 01:13:40,400
Like you've got to teach the 
understanding and actually 

1432
01:13:40,400 --> 01:13:43,840
knowing what you're building. 
AI is non deterministic, right? 

1433
01:13:43,840 --> 01:13:46,920
So like my existing course is 
you're you're you're copying the

1434
01:13:46,920 --> 01:13:49,640
code that I give you. 
But like, yeah, you can't 

1435
01:13:49,640 --> 01:13:51,880
control. 
The way you prompt something and

1436
01:13:51,880 --> 01:13:53,640
someone else prompting 
something, right? 

1437
01:13:54,920 --> 01:14:00,320
Yeah. 
And also, it's just, I, this 

1438
01:14:00,320 --> 01:14:02,040
stuff is all moving so God damn 
fast. 

1439
01:14:02,640 --> 01:14:05,840
It is, it takes me, it takes me 
a long time to make a course. 

1440
01:14:06,120 --> 01:14:08,200
And like I I don't want to 
create something that's going to

1441
01:14:08,200 --> 01:14:10,880
be out of date immediately. 
What do you think is going to 

1442
01:14:10,880 --> 01:14:12,560
happen? 
Like, do you think we're going 

1443
01:14:12,560 --> 01:14:20,000
to like, agree on a process and 
a way to use a genic coding like

1444
01:14:20,360 --> 01:14:25,360
soon, or do you think it's gonna
be this weird flow where people 

1445
01:14:25,360 --> 01:14:27,720
come up with new ways of doing 
things, but it's still 

1446
01:14:27,720 --> 01:14:29,320
constantly changing? 
Do you think we're gonna see 

1447
01:14:29,320 --> 01:14:31,160
that for like, quite a few 
years? 

1448
01:14:32,440 --> 01:14:34,320
I don't. 
Want to predict anything 

1449
01:14:34,320 --> 01:14:37,560
confidently in the world of AI? 
None of us have any idea what 

1450
01:14:37,560 --> 01:14:41,360
we're doing. 
I, if I had to guess, I, I think

1451
01:14:41,360 --> 01:14:46,360
we're going to converge somewhat
like right now, there is just so

1452
01:14:46,360 --> 01:14:49,960
many different things, like 
there's so many different 

1453
01:14:49,960 --> 01:14:53,720
models, harnesses, like there's 
all kinds of new tools coming 

1454
01:14:53,720 --> 01:14:56,880
out, MCPS. 
And this is just because it's so

1455
01:14:56,880 --> 01:14:58,480
new. 
None of us know what we're doing

1456
01:14:58,480 --> 01:14:59,720
yet. 
We're all figuring it out. 

1457
01:15:01,160 --> 01:15:02,880
I think if you look at the way 
these kind of things have gone 

1458
01:15:02,880 --> 01:15:05,160
in the past, something new is 
invented. 

1459
01:15:05,720 --> 01:15:10,600
You know, people spend a lot of 
time competing and figuring out 

1460
01:15:10,600 --> 01:15:13,040
where it's going. 
People invent new things and try

1461
01:15:13,040 --> 01:15:15,800
new ideas. 
But, you know, like, over time, 

1462
01:15:16,160 --> 01:15:17,840
some things just emerge 
dominant. 

1463
01:15:17,840 --> 01:15:21,040
And we might not get it down to 
like, one true way of doing 

1464
01:15:21,040 --> 01:15:24,040
things, but, you know, we learn 
our lessons. 

1465
01:15:24,040 --> 01:15:27,240
You know, some things clearly 
work better than others. 

1466
01:15:27,280 --> 01:15:32,000
And I think AI will probably go 
that way in that like, yeah, I, 

1467
01:15:32,000 --> 01:15:35,560
I think things will converge on 
fewer ways and methodologies 

1468
01:15:35,560 --> 01:15:39,360
than there are now. 
But I mean, who knows, really? 

1469
01:15:39,360 --> 01:15:41,920
I'm, I'm just, I'm just trying 
to ride, ride this wave like 

1470
01:15:41,920 --> 01:15:43,920
everybody else. 
Yeah, same. 

1471
01:15:44,760 --> 01:15:45,960
Same. 
Yeah, Really. 

1472
01:15:46,320 --> 01:15:48,440
Really. 
Early and I think these. 

1473
01:15:48,760 --> 01:15:51,440
These kind of conversations are 
really important too, because 

1474
01:15:51,800 --> 01:15:55,840
like to hear how other people 
are using it, like helps you 

1475
01:15:55,840 --> 01:16:00,200
change your daily workflows and 
like, or maybe not, maybe you're

1476
01:16:00,200 --> 01:16:01,920
like, wow, I'm glad I don't do 
it that way. 

1477
01:16:02,400 --> 01:16:05,000
But like, I think we can, we can
all learn from it. 

1478
01:16:05,000 --> 01:16:09,040
Like the the guys that I work 
with at Collects, we're always 

1479
01:16:09,040 --> 01:16:13,400
sharing like new ways of 
implementing AI into just 

1480
01:16:13,840 --> 01:16:18,160
working on features or tracking 
down century airs and things 

1481
01:16:18,160 --> 01:16:21,360
like that, you know? 
Yeah, yeah, absolutely. 

1482
01:16:21,360 --> 01:16:22,640
That was. 
One of the great things about 

1483
01:16:22,640 --> 01:16:26,240
Gauntlet, you know, it's three 
weeks remote, 7 weeks in person,

1484
01:16:26,240 --> 01:16:30,480
but yeah, yeah, just in a room 
all day every day with smart 

1485
01:16:30,480 --> 01:16:33,240
people who are all trying to 
figure it out like you are. 

1486
01:16:33,240 --> 01:16:35,880
And everybody is trying 
different things. 

1487
01:16:35,880 --> 01:16:38,520
They've got a perspective that 
you don't and everybody is just 

1488
01:16:38,520 --> 01:16:40,960
sharing what they've learned. 
And yeah, just such a fantastic 

1489
01:16:40,960 --> 01:16:43,960
environment. 
And you know, now I'm now I'm 

1490
01:16:43,960 --> 01:16:48,440
contracting, I'm working with 
AIAI first team where the 

1491
01:16:48,440 --> 01:16:50,760
philosophy is very much the 
gauntlet mentality. 

1492
01:16:52,080 --> 01:16:56,360
We are doing everything we can 
to try and use AI as much as 

1493
01:16:56,360 --> 01:16:58,920
possible and making a lot of 
mistakes along the way. 

1494
01:16:58,920 --> 01:17:01,440
But, you know, we're, we're 
sharing those lessons and 

1495
01:17:01,440 --> 01:17:07,160
learning because I think 
everyone is, is, is got the same

1496
01:17:07,160 --> 01:17:09,280
mindset that I've already talked
about, which is, which is that 

1497
01:17:09,280 --> 01:17:11,400
like, yeah, we don't know where 
it's going. 

1498
01:17:11,400 --> 01:17:12,520
Yeah, yeah. 
There's still a lot of 

1499
01:17:12,520 --> 01:17:14,680
uncertainty. 
But like, like, it's just so 

1500
01:17:14,680 --> 01:17:20,120
obvious to me that that that 
like, well, I don't know how 

1501
01:17:20,120 --> 01:17:22,480
long it's going to take. 
Maybe it'll take 10 years, maybe

1502
01:17:22,480 --> 01:17:25,440
it'll take six months. 
But like, clearly, like AI is 

1503
01:17:25,440 --> 01:17:27,600
going to be doing much more than
it is now. 

1504
01:17:28,280 --> 01:17:29,560
Like, we've just got to get 
there. 

1505
01:17:29,560 --> 01:17:32,760
Like, like the person who gets 
there first is just going to 

1506
01:17:32,760 --> 01:17:34,360
absolutely crush the 
competition. 

1507
01:17:34,360 --> 01:17:36,400
So. 
Yeah, we, we. 

1508
01:17:36,560 --> 01:17:38,880
We we can risk. 
Some mistakes along the way just

1509
01:17:38,880 --> 01:17:42,640
for that speed advantage. 
What are your feelings along? 

1510
01:17:42,680 --> 01:17:46,880
Like just giving access to your 
code base to just, you know, 

1511
01:17:46,880 --> 01:17:49,800
third parties essentially like 
all these like Claw, like 

1512
01:17:50,040 --> 01:17:52,960
Anthropic and a big 
conversation. 

1513
01:17:52,960 --> 01:17:54,800
Like, you know. 
All these Gray beards that are 

1514
01:17:54,800 --> 01:17:58,680
really anti AIA. 
Lot of their reasoning is like, 

1515
01:17:58,680 --> 01:18:02,800
well I'm not just going to give 
all my code and access to these 

1516
01:18:02,800 --> 01:18:04,920
companies to use it however they
please. 

1517
01:18:05,880 --> 01:18:07,600
Oh great. 
Question. 

1518
01:18:07,600 --> 01:18:09,080
This is something I think about 
as well. 

1519
01:18:09,080 --> 01:18:10,800
I mean, at the end of the day, 
I'm doing it. 

1520
01:18:10,800 --> 01:18:13,360
So like, yeah, same. 
And it's something, I think. 

1521
01:18:13,360 --> 01:18:15,000
About too, but no matter what I 
say. 

1522
01:18:15,960 --> 01:18:17,520
Because there's two there's. 
Two aspects to this right? 

1523
01:18:17,520 --> 01:18:21,720
Because there's, there's the 
sort of privacy angle in that, 

1524
01:18:21,720 --> 01:18:25,160
like I'm sending a lot of things
off to some third party server, 

1525
01:18:25,520 --> 01:18:27,640
which previously would just have
been on my machine. 

1526
01:18:27,640 --> 01:18:31,920
And yeah, you know, like, like 
I'm not publishing it publicly, 

1527
01:18:31,920 --> 01:18:35,040
but yeah, I mean, there, there 
are reasons to be wary of that. 

1528
01:18:35,040 --> 01:18:39,240
The the other side of it is just
more security and permissions 

1529
01:18:39,240 --> 01:18:43,400
aspect. 
Like if I'm just running cloud 

1530
01:18:43,400 --> 01:18:46,600
code blindly and it's got access
to a lot of things on my system,

1531
01:18:47,760 --> 01:18:49,760
yeah. 
I mean there are things built in

1532
01:18:49,800 --> 01:18:53,680
it'll ask like A&V files alone. 
Too like a lot of keys and 

1533
01:18:53,720 --> 01:18:55,520
things like. 
Do people think of like? 

1534
01:18:55,880 --> 01:18:59,080
Are you always thinking about 
like, not having a copy of your 

1535
01:18:59,080 --> 01:19:02,480
production keys in your code 
base locally? 

1536
01:19:02,880 --> 01:19:05,880
Like these are things that we 
have to be, like, really aware 

1537
01:19:05,880 --> 01:19:07,000
of. 
I feel like now. 

1538
01:19:08,000 --> 01:19:08,720
Yeah. 
It's. 

1539
01:19:10,040 --> 01:19:13,360
I mean, a month or two ago, just
there was this, this 

1540
01:19:13,800 --> 01:19:16,760
unbelievable amount of hype 
about Open Claw. 

1541
01:19:17,080 --> 01:19:19,600
You you couldn't, I couldn't 
open Twitter without seeing 1000

1542
01:19:19,600 --> 01:19:22,680
posts about it. 
Yeah, yeah, like immediately. 

1543
01:19:23,000 --> 01:19:24,520
And, and, you know, Open Claw is
still big. 

1544
01:19:24,520 --> 01:19:27,600
People are still using it. 
But yeah, you know, there were 

1545
01:19:27,600 --> 01:19:30,240
just so many stories around of 
like, like people who didn't 

1546
01:19:30,240 --> 01:19:31,360
really know what they were 
doing. 

1547
01:19:31,360 --> 01:19:33,760
And yeah, yeah, less technical 
people who were like, oh, wow, 

1548
01:19:33,840 --> 01:19:34,800
this is cool. 
Check it out. 

1549
01:19:34,800 --> 01:19:37,040
This AI can do anything. 
I'll just install it and run it.

1550
01:19:37,480 --> 01:19:39,400
And then like, I just. 
And it just needs all. 

1551
01:19:39,560 --> 01:19:41,360
It just needs these. 
API keys. 

1552
01:19:42,640 --> 01:19:44,160
Yeah, there was a blog post. 
I read it. 

1553
01:19:44,160 --> 01:19:46,760
Was some some white hat, white 
hat guy. 

1554
01:19:46,760 --> 01:19:50,000
He like, I haven't even used 
open claw that much, but there 

1555
01:19:50,000 --> 01:19:52,000
there, you know, there's like 
skill repositories. 

1556
01:19:52,000 --> 01:19:55,080
You can you can just anybody can
upload a skill and then your, 

1557
01:19:55,080 --> 01:19:58,240
your open claw can download it. 
And yeah, some some guy it was 

1558
01:19:58,240 --> 01:20:00,600
one of the top rated skills. 
I think he even used some like 

1559
01:20:00,600 --> 01:20:03,120
hacky technique to just 
artificially get it to the top. 

1560
01:20:03,120 --> 01:20:06,440
So a lot of people would 
download it And yeah, it was a 

1561
01:20:06,440 --> 01:20:09,200
skill that just you downloaded 
it. 

1562
01:20:09,440 --> 01:20:12,120
I forget the exact details, but 
it it, it just did something 

1563
01:20:12,120 --> 01:20:14,080
very insecure. 
You know it it just like. 

1564
01:20:14,720 --> 01:20:18,000
Pinged some public URL in a 
harmless way. 

1565
01:20:18,000 --> 01:20:21,000
But you know, if that had been, 
if that had been a Black Hat 

1566
01:20:21,000 --> 01:20:25,320
guy, then yeah, he, he was doing
it out of charity. 

1567
01:20:25,320 --> 01:20:27,000
You know, you pinged this and 
then it would send a message 

1568
01:20:27,000 --> 01:20:29,240
back saying, hey, you shouldn't 
have downloaded this skill, you 

1569
01:20:29,240 --> 01:20:31,280
know, you know, you just gave me
access to everything. 

1570
01:20:31,320 --> 01:20:34,200
If I'd wanted to, I could, I 
could have stolen all your data.

1571
01:20:34,200 --> 01:20:40,080
And yeah, he got like a lot of. 
Yeah, I think this is yeah, that

1572
01:20:40,240 --> 01:20:44,120
I heard a lot like I. 
I have like glanced at Openclaw.

1573
01:20:44,120 --> 01:20:46,360
I have not really played around 
with it, but they have like a 

1574
01:20:46,360 --> 01:20:48,680
marketplace for skills 
essentially, right? 

1575
01:20:50,720 --> 01:20:54,800
Yeah, Yeah, yeah, there's I 
download it when everybody was. 

1576
01:20:54,800 --> 01:20:57,600
Talking about it. 
And like there's like 30%. 

1577
01:20:57,600 --> 01:21:00,920
Of them are like malicious and 
dangerous or something. 

1578
01:21:00,920 --> 01:21:05,200
Maybe it's not that high, but it
was a pretty big number, yeah. 

1579
01:21:05,200 --> 01:21:08,640
I I can believe it. 
I mean, to answer your earlier 

1580
01:21:08,640 --> 01:21:10,920
question, I, I don't know how 
we're going to, we're going to 

1581
01:21:10,920 --> 01:21:12,160
solve this. 
It's a huge problem. 

1582
01:21:12,360 --> 01:21:17,200
I mean, you see it in, in vibe 
coding as well, like we've been 

1583
01:21:17,200 --> 01:21:20,400
talking about, about the risks 
of, you know, building things 

1584
01:21:20,400 --> 01:21:22,160
without a great understanding of
what you're building. 

1585
01:21:22,160 --> 01:21:24,200
And that's, you know, something 
I've seen. 

1586
01:21:24,520 --> 01:21:26,840
Yeah. 
Like that's a real danger. 

1587
01:21:26,840 --> 01:21:30,000
Like security is difficult and 
security, security requires a 

1588
01:21:30,000 --> 01:21:32,480
lot of knowledge. 
Security breaches happen a lot 

1589
01:21:32,480 --> 01:21:35,440
of the time because you don't 
know what you don't know because

1590
01:21:35,440 --> 01:21:39,040
you don't have that experience. 
Like like there's just a lot of 

1591
01:21:39,040 --> 01:21:42,320
things that are non obvious. 
You know, if you're a totally 

1592
01:21:42,320 --> 01:21:44,680
new programmer and no one, no 
one has ever told you that, 

1593
01:21:44,680 --> 01:21:47,920
like, I don't know, like no one 
has told you, Oh, no, you 

1594
01:21:47,920 --> 01:21:52,600
shouldn't output untrust. 
You shouldn't render untrusted 

1595
01:21:52,600 --> 01:21:56,840
user input on, on on your page 
because that exposes you to XSS.

1596
01:21:56,840 --> 01:22:00,080
Like, you got to sanitize. 
You got to sanitize the HTML. 

1597
01:22:00,080 --> 01:22:02,080
You know, that's the kind of the
thing where where like, yeah, 

1598
01:22:02,120 --> 01:22:04,240
yeah, yeah, that's an important 
security concern and you've got 

1599
01:22:04,240 --> 01:22:06,880
to be wary of it. 
But like, if no one's ever told 

1600
01:22:06,880 --> 01:22:09,440
you that, you don't know. 
And yeah, it would be very easy 

1601
01:22:09,440 --> 01:22:11,520
to just build something and you 
think, oh, yeah, it works. 

1602
01:22:11,560 --> 01:22:13,440
And and it never occurs to you 
that there's a problem with 

1603
01:22:13,440 --> 01:22:14,840
this. 
You know, you see this in things

1604
01:22:14,840 --> 01:22:18,760
like just architectural choices,
setting things up where like 

1605
01:22:18,920 --> 01:22:22,240
you're not, you don't have the 
right or beginners mistakes. 

1606
01:22:22,240 --> 01:22:25,320
Just like like everything looks 
like it works, but you know, 

1607
01:22:25,320 --> 01:22:27,520
just things aren't authenticated
correctly. 

1608
01:22:27,520 --> 01:22:30,040
Maybe the back end is sending 
more data than it should. 

1609
01:22:30,120 --> 01:22:31,880
You're exposing PII, that kind 
of thing. 

1610
01:22:31,880 --> 01:22:34,520
These are just really easy 
mistakes to make when when you 

1611
01:22:35,560 --> 01:22:38,560
I. 
Remember, AI makes that super. 

1612
01:22:38,560 --> 01:22:40,120
Dangerous. 
And I don't know what the 

1613
01:22:40,120 --> 01:22:42,560
solution is. 
I think it comes back to what I 

1614
01:22:42,560 --> 01:22:45,440
said about, about, about new 
paradigms like, like, like we, 

1615
01:22:45,520 --> 01:22:47,800
we, we've got to invent just 
just like new ways of 

1616
01:22:47,800 --> 01:22:52,760
architecting things, new ways of
new ways of working that weren't

1617
01:22:52,760 --> 01:22:56,200
necessary before. 
But yeah, because. 

1618
01:22:56,480 --> 01:22:58,680
The. 
Old way it, it just doesn't gel 

1619
01:22:58,680 --> 01:23:02,480
perfectly with these new tools. 
Yeah, yeah. 

1620
01:23:02,960 --> 01:23:07,920
And like, it's like, yeah, we 
learned before AI right through 

1621
01:23:07,920 --> 01:23:11,560
mistakes a lot of times. 
And like, I didn't know like how

1622
01:23:11,560 --> 01:23:14,720
dangerous like man in the middle
attacks were until I experienced

1623
01:23:14,720 --> 01:23:18,360
one myself. 
Or like giving too much data in 

1624
01:23:18,360 --> 01:23:23,480
a Restful API response and like 
just allowing things to be 

1625
01:23:23,480 --> 01:23:26,840
manipulated and you don't really
learn these things. 

1626
01:23:26,840 --> 01:23:31,640
And yeah, now if any of this 
stuff just gets thrown into your

1627
01:23:31,640 --> 01:23:34,200
code base and you don't even 
know it's there, could be a 

1628
01:23:34,200 --> 01:23:36,400
problem. 
Yeah, I. 

1629
01:23:36,400 --> 01:23:37,640
Think we'll figure it out in the
end. 

1630
01:23:37,640 --> 01:23:39,840
Everybody's got different risk 
profiles, you know, if you're 

1631
01:23:39,840 --> 01:23:42,440
working for sure or something 
where you've got super high. 

1632
01:23:42,480 --> 01:23:42,880
Yeah. 
Yeah. 

1633
01:23:44,440 --> 01:23:46,280
Not all security incidents are 
equal. 

1634
01:23:46,280 --> 01:23:48,080
High, you know, the the cost is 
missing. 

1635
01:23:49,000 --> 01:23:51,320
Yeah. 
Yeah, some companies and some 

1636
01:23:51,320 --> 01:23:54,160
industries, just just the cost, 
the cost of failure is far 

1637
01:23:54,160 --> 01:23:55,880
higher. 
And so there's just a lot more 

1638
01:23:55,880 --> 01:23:57,920
incentive to be cautious. 
And so I can totally understand 

1639
01:23:57,920 --> 01:24:02,080
why those guys are slow walking.
AII mean that's. 

1640
01:24:03,120 --> 01:24:04,920
Yeah, I don't know. 
I guess we'll see. 

1641
01:24:05,680 --> 01:24:08,720
Yeah, I don't know. 
I feel like it's just the Wild 

1642
01:24:08,720 --> 01:24:11,200
West right now and things will 
settle down. 

1643
01:24:12,600 --> 01:24:15,280
Yeah, I I think the. 
Industry will probably advance 

1644
01:24:15,280 --> 01:24:20,600
one disaster at a time. 
I mean, that's the best. 

1645
01:24:20,600 --> 01:24:23,720
Way to learn, right? 
You need the mistakes to happen 

1646
01:24:23,720 --> 01:24:29,040
and then everyone adjusts. 
Yeah, you know what I said about

1647
01:24:29,040 --> 01:24:30,000
new? 
Paradigms, right? 

1648
01:24:30,000 --> 01:24:35,720
It's like computers, computers 
used to be ugly like computers 

1649
01:24:35,720 --> 01:24:38,840
used to be be beige boxes and 
like no one ever questioned it 

1650
01:24:38,840 --> 01:24:41,360
and and it was Apple and Steve 
Jobs and the iPhone. 

1651
01:24:41,360 --> 01:24:43,160
Steve Jobs came along and said, 
you know, it doesn't have to be 

1652
01:24:43,160 --> 01:24:44,800
like this. 
Why, why can't your phone be 

1653
01:24:44,800 --> 01:24:47,160
sleek and beautiful and your 
laptop as well? 

1654
01:24:47,320 --> 01:24:50,800
And, and, and it's one of those 
things it just, it seems so 

1655
01:24:50,800 --> 01:24:52,760
obvious in retrospect. 
It's like, why, why didn't 

1656
01:24:52,760 --> 01:24:55,200
anyone think of that? 
Of course, you know, of course 

1657
01:24:55,200 --> 01:24:57,240
computers should be aesthetic as
well as functional. 

1658
01:24:57,240 --> 01:24:59,520
But, you know, you know, Steve 
Jobs was a visionary in that 

1659
01:24:59,520 --> 01:25:02,160
way. 
He, he and yeah, I, I think it's

1660
01:25:02,160 --> 01:25:05,760
going to be like that with, 
with, with vibe coding, not vibe

1661
01:25:05,760 --> 01:25:08,120
coding, just agentic engineering
and AI. 

1662
01:25:08,160 --> 01:25:10,000
Although those people are going 
to come along and they're going 

1663
01:25:10,000 --> 01:25:12,720
to, they're going to say, hey, 
why don't we do it in this new 

1664
01:25:12,720 --> 01:25:15,000
way that actually solves those 
problems with security or 

1665
01:25:15,000 --> 01:25:17,400
whatnot. 
And you're going to sit down in 

1666
01:25:17,400 --> 01:25:20,000
front of it and you're going to 
think like, damn, this is so 

1667
01:25:20,000 --> 01:25:22,160
much better. 
Like, why didn't, why didn't we 

1668
01:25:22,160 --> 01:25:23,760
all see this? 
But, yeah, I, I think that 

1669
01:25:24,120 --> 01:25:25,600
that's often how these things 
advance. 

1670
01:25:25,960 --> 01:25:30,200
Yeah. 
And I, I make no claim of being 

1671
01:25:30,200 --> 01:25:32,560
that visionary that's I can 
dream. 

1672
01:25:33,200 --> 01:25:36,800
Hey look, I have. 
Always said I'm like, I'm smart 

1673
01:25:36,800 --> 01:25:39,440
enough to use smart people's 
tech, right? 

1674
01:25:39,480 --> 01:25:41,440
And that's about it. 
Like, I don't think I'm going 

1675
01:25:41,440 --> 01:25:43,720
to. 
I'll just keep talking to people

1676
01:25:43,720 --> 01:25:47,800
that create really cool things 
on on my podcast and that's, and

1677
01:25:47,800 --> 01:25:51,320
then I'll use them and I'll 
build cool products and that's 

1678
01:25:51,320 --> 01:25:54,160
as far as that's, that's as 
smart as I get. 

1679
01:25:55,800 --> 01:25:57,800
Yeah, this is a. 
Point I hear people make a lot 

1680
01:25:57,800 --> 01:26:02,520
that, you know, AI is so 
knowledgeable and it's so smart 

1681
01:26:02,520 --> 01:26:04,200
and you can ask it a question 
about anything. 

1682
01:26:04,200 --> 01:26:08,800
And yet like AI is actually 
given how, how quote UN quote 

1683
01:26:08,800 --> 01:26:12,480
smart it is, It's it's not as 
innovative as like you might 

1684
01:26:12,480 --> 01:26:14,720
expect. 
Like like where are the like 

1685
01:26:14,720 --> 01:26:18,640
huge paradigm busting kind of 
advances in knowledge that you 

1686
01:26:18,640 --> 01:26:21,760
know, where, where AI just kind 
of connected to things that 

1687
01:26:21,760 --> 01:26:25,280
hadn't been connected before. 
We haven't seen that yet, have 

1688
01:26:25,280 --> 01:26:27,000
we? 
Yeah, and and you know, maybe 

1689
01:26:27,000 --> 01:26:29,880
that's just an. 
Artefact of how current AI works

1690
01:26:29,880 --> 01:26:34,560
in in that it's just. 
It's just it, it's not trying to

1691
01:26:34,560 --> 01:26:37,000
invent new things, it's trying 
to predict, it's trying to 

1692
01:26:37,000 --> 01:26:39,560
predict the next token based on 
like, like the existing corpus 

1693
01:26:39,560 --> 01:26:41,040
of knowledge and everything it's
seen. 

1694
01:26:41,040 --> 01:26:43,320
So you know, it's, it's somewhat
backwards looking. 

1695
01:26:43,600 --> 01:26:45,080
It's just really good at 
pattern. 

1696
01:26:45,080 --> 01:26:50,360
Matching essentially right, like
or pattern recognition, yeah. 

1697
01:26:52,560 --> 01:26:56,680
Yeah, I was thinking of that 
because I was wondering why, 

1698
01:26:56,720 --> 01:26:58,760
why? 
Why isn't AI that visionary? 

1699
01:26:58,760 --> 01:27:01,320
Maybe AI can solve the problems 
with AII. 

1700
01:27:01,320 --> 01:27:03,520
Don't know. 
Someone much smarter than me at 

1701
01:27:03,520 --> 01:27:05,920
Open It. 
I've seen some cool things. 

1702
01:27:05,920 --> 01:27:13,480
Like in medical, right, where it
can like look at like DNA and 

1703
01:27:13,480 --> 01:27:16,560
things like that and find 
patterns in like, you know, in 

1704
01:27:16,720 --> 01:27:19,320
billions and billions of data 
points. 

1705
01:27:19,800 --> 01:27:23,600
But at the same time, it's just 
looking through that. 

1706
01:27:23,600 --> 01:27:25,520
Data. 
Quicker than a human could, 

1707
01:27:25,520 --> 01:27:32,640
Right? 
So it's not actually discovering

1708
01:27:32,640 --> 01:27:34,600
things, it's just finding 
patterns. 

1709
01:27:34,680 --> 01:27:41,200
Faster than a human can, yeah. 
I guess like connecting to a lot

1710
01:27:41,200 --> 01:27:43,240
of a lot of discovery and 
invention, it's not something 

1711
01:27:43,240 --> 01:27:46,280
that's wildly beyond what we 
already know. 

1712
01:27:46,280 --> 01:27:49,320
It's often just like connecting 
to disparate things that that, 

1713
01:27:49,360 --> 01:27:51,080
yeah, that's true. 
Not been connected before. 

1714
01:27:51,720 --> 01:27:55,240
Yeah. 
So yeah, I don't know, maybe. 

1715
01:27:55,240 --> 01:27:57,320
We'll be seeing more of that 
soon. 

1716
01:27:57,320 --> 01:28:00,160
That would be cool, right? 
Yeah, I'm sure. 

1717
01:28:00,160 --> 01:28:02,840
Someone much smarter than me at 
Anthropic or something like that

1718
01:28:02,840 --> 01:28:07,520
is working on that. 
Probably where's Where's Thomas 

1719
01:28:07,520 --> 01:28:13,160
Miller at? 
But yeah. 

1720
01:28:14,000 --> 01:28:16,720
This has been a lot of fun, man.
I think we can we can wrap it 

1721
01:28:16,720 --> 01:28:18,600
up. 
We've been chatting for an hour 

1722
01:28:18,600 --> 01:28:23,280
and a half. 
Do you have any advice for 

1723
01:28:23,280 --> 01:28:28,720
people out there that might be 
hesitant to use AI or agenic 

1724
01:28:28,720 --> 01:28:33,000
workflows in their current 
day-to-day or anything? 

1725
01:28:33,000 --> 01:28:36,360
For maybe even like. 
Junior developers like any 

1726
01:28:36,360 --> 01:28:38,960
advice there on how to 
accelerate your career? 

1727
01:28:40,520 --> 01:28:45,680
Ask a lot of questions. 
Be as curious as you can as you 

1728
01:28:45,680 --> 01:28:48,640
can. 
Force yourself to be like every 

1729
01:28:49,240 --> 01:28:51,720
try and understand every last 
little thing that you can. 

1730
01:28:51,720 --> 01:28:53,720
If there's ever anything you're 
not sure about. 

1731
01:28:53,720 --> 01:28:56,160
Why does this code do this? 
What does that tool do? 

1732
01:28:56,160 --> 01:28:58,480
What's the best way here? 
Ask the AI. 

1733
01:28:58,480 --> 01:29:02,600
And yeah, yeah, just you've got 
to use your own judgement. 

1734
01:29:02,600 --> 01:29:04,920
But at the same time, like AI is
a great tutor. 

1735
01:29:05,520 --> 01:29:08,800
And I guess the other thing I 
would say is, yeah, like I 

1736
01:29:08,800 --> 01:29:11,920
understand the skepticism. 
I was there myself for a long 

1737
01:29:11,920 --> 01:29:13,560
time. 
But yeah, I I think part of it 

1738
01:29:13,560 --> 01:29:14,880
is you've just got to push 
through. 

1739
01:29:15,600 --> 01:29:18,320
Yeah, if it, if you're 
experienced, it might, it might 

1740
01:29:18,320 --> 01:29:21,480
slow you down at first, but try 
something new, you know, try, 

1741
01:29:21,480 --> 01:29:24,240
try and build something, I don't
know, a side project, something 

1742
01:29:24,240 --> 01:29:26,600
that, that even if you're 
experienced, maybe just 

1743
01:29:26,600 --> 01:29:29,640
something that's just a totally 
different stack or, or, or some 

1744
01:29:29,640 --> 01:29:33,280
kind of apathet that is very far
outside what you do know. 

1745
01:29:33,280 --> 01:29:35,280
And, and just see how far AI can
take you. 

1746
01:29:35,280 --> 01:29:38,920
You know, like I'm just say, 
say, OK, well, I will try and 

1747
01:29:38,920 --> 01:29:42,520
build this thing entirely with 
AI and just not use my old 

1748
01:29:42,520 --> 01:29:44,200
habits. 
I'll force myself to just do it 

1749
01:29:44,200 --> 01:29:45,880
all with AI. 
And then even if you don't end 

1750
01:29:45,880 --> 01:29:48,600
up building something fantastic,
like you're going to learn so 

1751
01:29:48,600 --> 01:29:51,160
much along the way that yeah, 
just just just put the time in. 

1752
01:29:51,160 --> 01:29:54,120
You've got to put in the reps. 
Yeah, I agree. 

1753
01:29:54,560 --> 01:29:57,960
And the best way to learn is 
building your own thing, not 

1754
01:29:57,960 --> 01:30:01,400
following tutorials. 
And I mean, take that tutorial 

1755
01:30:01,400 --> 01:30:04,040
knowledge, apply it to your own 
project. 

1756
01:30:05,280 --> 01:30:06,200
Yeah. 
Absolutely. 

1757
01:30:07,000 --> 01:30:08,480
All right, awesome. 
Thanks, George. 

1758
01:30:08,480 --> 01:30:10,320
And. 
Everyone that has tuned in, 

1759
01:30:10,320 --> 01:30:12,440
thank you so much and have a 
good weekend guys. 

1760
01:30:14,720 --> 01:30:15,320
Testing.