1
00:00:00,000 --> 00:00:00,680
I don't know.

2
00:00:00,780 --> 00:00:01,680
We're like two months into it.

3
00:00:01,700 --> 00:00:04,160
I'm like, I'm going to have to set up a new one at some point

4
00:00:04,160 --> 00:00:05,340
because this is going to be outdated.

5
00:00:06,140 --> 00:00:06,380
Really?

6
00:00:06,560 --> 00:00:06,720
Okay.

7
00:00:07,440 --> 00:00:09,100
I think that's the conclusion I'm coming to.

8
00:00:09,140 --> 00:00:10,860
You can switch out the models and stuff like that,

9
00:00:10,880 --> 00:00:16,580
but I think the context memory is the big problem, I think,

10
00:00:16,820 --> 00:00:18,780
and users like me who aren't as technically competent

11
00:00:18,780 --> 00:00:21,580
need to figure out how do we nail that out of the gate.

12
00:00:21,600 --> 00:00:22,380
That's a good problem you're familiar with.

13
00:00:22,400 --> 00:00:23,120
Sounds familiar, yeah.

14
00:00:23,120 --> 00:00:23,360
Right?

15
00:00:23,580 --> 00:00:23,760
Yeah.

16
00:00:25,160 --> 00:00:25,980
That's what I mean.

17
00:00:26,180 --> 00:00:27,440
That feels like a good starting point.

18
00:00:27,440 --> 00:00:29,180
I think

19
00:00:29,180 --> 00:00:32,180
yeah what would be most helpful I think about

20
00:00:32,180 --> 00:00:34,000
just catching up overall

21
00:00:34,000 --> 00:00:35,880
on what's happening in the space and

22
00:00:35,880 --> 00:00:38,320
talking about the pieces you've heard me talk about it for a long time

23
00:00:38,320 --> 00:00:38,600
so

24
00:00:38,600 --> 00:00:42,360
can we talk about the 1031 off site we're at sometimes

25
00:00:42,360 --> 00:00:42,820
yeah yeah

26
00:00:42,820 --> 00:00:46,320
I mean just I get a little embarrassed

27
00:00:46,320 --> 00:00:48,320
getting up there and showing graph stuff

28
00:00:48,320 --> 00:00:50,180
just in front of everyone I see everyone go

29
00:00:50,180 --> 00:00:51,560
oh god Paul another year

30
00:00:51,560 --> 00:00:52,260
the graph guy

31
00:00:52,260 --> 00:00:56,200
and you get a lot of grief

32
00:00:56,200 --> 00:01:02,840
for graphs on online because they've been tried so many times but um i've worked with neo4j since

33
00:01:02,840 --> 00:01:08,200
2010 2011 sometime around then i think they were just starting out uh one of our technical guys

34
00:01:08,200 --> 00:01:15,960
brought it into the company and i hated it because it just crashed all the time so um but now 15

35
00:01:15,960 --> 00:01:20,360
years later it's kind of seeing the light of day so i think you just have to have all these kind of

36
00:01:20,360 --> 00:01:24,960
of primitives in your toolbox and then when the time's right you pull them out

37
00:01:24,960 --> 00:01:29,840
and so we just think that the memory issue you just brought up graph

38
00:01:29,840 --> 00:01:33,800
databases just serve as a great scratch pad for that and it doesn't have to be

39
00:01:33,800 --> 00:01:38,520
in a graph database it can be an obsidian files it's just the whole thing

40
00:01:38,520 --> 00:01:46,820
is relating one thing to another but anyway well I think it'd be worthwhile

41
00:01:46,820 --> 00:01:54,100
wild to go into differentiating like llm's how they work from these these graph this graph

42
00:01:54,100 --> 00:02:01,060
approach because i think again like we're just discussing before we hit record yeah a ton of

43
00:02:01,060 --> 00:02:09,300
capital time and uh effort has been put into llm specifically but i think some would argue your i

44
00:02:09,300 --> 00:02:15,220
think yourself included that llms may not be the the best way to go about this problem yeah i think

45
00:02:15,220 --> 00:02:21,040
people anthropomorphize LLMs a lot you know because it's speaking language to

46
00:02:21,040 --> 00:02:25,420
you because you can talk to it you think that it's actually reasoning and

47
00:02:25,420 --> 00:02:30,400
especially when they call it a reasoning model and it does do an amazing job of

48
00:02:30,400 --> 00:02:35,560
mimicking logic but it does not know why it's saying what it's saying it's just

49
00:02:35,560 --> 00:02:41,720
the statistical output of the next word yeah yeah I mean how do you see it do

50
00:02:41,720 --> 00:02:44,680
you when you think of an LLM do you think of it as do you find yourself

51
00:02:44,680 --> 00:02:48,820
thinking of it as a machine spitting out words or do you think of it as you know

52
00:02:48,820 --> 00:02:51,820
especially when you name a bar or something like that it really starts to

53
00:02:51,820 --> 00:02:56,580
feel like a human or something no I definitely don't think it's human like

54
00:02:56,580 --> 00:03:00,620
anytime I interact with our open call I'm like okay what context do I need to

55
00:03:00,620 --> 00:03:03,700
feed it to make sure that it gives me the right response like that's what I

56
00:03:03,700 --> 00:03:07,900
think most about it is like what do I need to preload this thing with today we

57
00:03:07,900 --> 00:03:10,160
should start with what's everyone running right now like that would be a

58
00:03:10,160 --> 00:03:16,280
good thing you're running a bunch of cool stuff right yeah I mean I'm using

59
00:03:16,280 --> 00:03:25,880
I've of course yeah doing some cloud code but I think the the context issue

60
00:03:25,880 --> 00:03:29,600
is something that everybody's kind of running into they're able to just like

61
00:03:29,600 --> 00:03:35,360
stumble through it or hack their way through it but I think like we're all

62
00:03:35,360 --> 00:03:39,180
gonna feel this need of something better like something that's gonna be more

63
00:03:39,180 --> 00:03:44,940
accurate give us better answers help us do the next thing better so and you're

64
00:03:44,940 --> 00:03:49,200
starting to see it like you open X and there's more and more visualizations of

65
00:03:49,200 --> 00:03:54,540
graphs I feel like the whole like ecosystem is drawn that direction but

66
00:03:54,540 --> 00:03:58,260
yeah those are some of the things I've been messing around with yeah it's you

67
00:03:58,260 --> 00:04:06,900
Marty I mean I for the context like interacting with my open claw bot like

68
00:04:06,900 --> 00:04:11,300
we're doing something i have to be very specific i'm like hey we're gonna write the bitcoin brief

69
00:04:11,300 --> 00:04:16,260
today i've dropped some stuff in a folder that i've named a specific name i'd say the name of

70
00:04:16,260 --> 00:04:20,580
the folder go to my tab the tab is named this like i have to give it like direct direction

71
00:04:20,580 --> 00:04:26,420
and i've heard of people leveraging the obsidian bolts to to sort of solve that context issue i

72
00:04:26,420 --> 00:04:35,300
haven't dove down that but yeah we're using open claw we use claude 4.6 and then um combination of

73
00:04:35,300 --> 00:04:40,180
could we have codex sub agents that we can and we can ping if we need to build something and then

74
00:04:41,140 --> 00:04:46,580
11 labs are the models that we're using now for voice yeah when you told me about your automation

75
00:04:46,580 --> 00:04:50,660
process it was pretty solid about what you do is that stayed the same since then yeah you want to

76
00:04:50,660 --> 00:04:56,660
describe that a little bit or yeah i mean for um like we'll take this podcast this transcript

77
00:04:57,380 --> 00:05:00,900
and then we have prompts that we've been iterating on for literally two years at this point

78
00:05:00,900 --> 00:05:07,140
um in claude that will that very long prompts that will take the transcript and then pull

79
00:05:07,940 --> 00:05:14,260
certain sections quotes and be able to either um directly quote for a section of the newsletter or

80
00:05:15,380 --> 00:05:19,460
find something that we're talking about do deeper research on it to expand on the topic

81
00:05:20,260 --> 00:05:27,380
for another piece of content um and that is for the podcast like post-production

82
00:05:27,380 --> 00:05:34,340
that's one of the things we'll do but other things we have for clips we'll give it the transcript

83
00:05:34,340 --> 00:05:42,180
with time stamps and claude will then use our third party twitter api that we have access to

84
00:05:42,180 --> 00:05:47,140
look at what people are sort of talking about what's trending and then it'll go back to our

85
00:05:47,140 --> 00:05:52,900
transcript see if we talked about something um during that episode that people are really on top

86
00:05:52,900 --> 00:05:57,540
of and then we'll be like all right here's the time stamp of a section of the conversation that

87
00:05:57,540 --> 00:06:02,660
you just had that if you clipped it out and put it on twitter it would it would probably get good

88
00:06:02,660 --> 00:06:07,780
engagement and are using any then for that sort of prop that pipeline anymore or it sounds like

89
00:06:07,780 --> 00:06:13,700
you brought up we bought so we were okay we live coded our own like back-end dashboard that basically

90
00:06:13,700 --> 00:06:18,420
replaced n8n um shout out to ed on our team just put a lot of effort into that that's pretty cool

91
00:06:18,420 --> 00:06:25,740
but uh so when you're done like this episode when you hit finish or record the recordings over does

92
00:06:25,740 --> 00:06:31,980
it automatically kick that over to these workflows once we put the files and Dropbox it'll start and

93
00:06:31,980 --> 00:06:37,560
then yeah you'll be or you can just if we put it in Dropbox you don't want to trigger automatically

94
00:06:37,560 --> 00:06:43,080
in the dashboard that we built and go like all right right around this process that's crazy yeah

95
00:06:43,080 --> 00:06:49,020
that's cool I was talking to Scott who runs you know Scott Foreman he runs a

96
00:06:49,020 --> 00:06:53,520
video lab they do a lot of work the Human Resource Foundation and he's trying

97
00:06:53,520 --> 00:06:57,600
to do a similar automation on the back end so everyone is trying to figure out

98
00:06:57,600 --> 00:07:03,880
this this process yeah but it is like it does feel we're joking you came here in

99
00:07:03,880 --> 00:07:08,820
a way moan it was driving on the wrong wrong side of the road yeah I let you

100
00:07:08,820 --> 00:07:11,880
know I'm like I don't I stopped paying attention to the road when you're in the

101
00:07:11,880 --> 00:07:18,200
back of an uber or waymo and i look the car starts like jittering and i look up the waymo is on the

102
00:07:18,200 --> 00:07:24,040
wrong side of the road with traffic heading towards me because the austin has all these events going

103
00:07:24,040 --> 00:07:29,880
on so there's they were like coning off the streets um so you could tell that somebody took over via

104
00:07:29,880 --> 00:07:35,960
teleob yeah and then finally got me over and then as soon as we were getting over there was a traffic

105
00:07:35,960 --> 00:07:41,240
cop who was it's like trying to direct the waymo and it was just this moment i was like

106
00:07:41,240 --> 00:07:47,400
gosh what the hell are we doing here yeah well that's every human directing a waymo and then

107
00:07:47,400 --> 00:07:53,160
some poor person sitting probably in the philippines operating the game on some playstation

108
00:07:53,160 --> 00:07:59,000
console from 1988 or something like that yeah it's crazy but one thing i was thinking on the way over

109
00:07:59,000 --> 00:08:05,560
was um and we had a team meeting yesterday we were talking about this i think we all

110
00:08:05,560 --> 00:08:11,160
find ourselves thinking about the current state and it's hard to think about the through line

111
00:08:11,160 --> 00:08:16,600
like where things are going to be in a few months um and when i was in this waymo today i was

112
00:08:16,600 --> 00:08:21,880
thinking about when waymo first came out or or self-driving in general people have been talking

113
00:08:21,880 --> 00:08:28,520
about about this for a long time and dismissing it but it's like it's basically here um and of

114
00:08:28,520 --> 00:08:32,660
course there's going to be some kinks along the way but we've come a long way in a short period

115
00:08:32,660 --> 00:08:39,040
of time and it's pretty hard to imagine what a year from now is going to look like because things

116
00:08:39,040 --> 00:08:45,560
are changing so much um but yeah it's exciting it's like an amazing time to be alive i mean we

117
00:08:45,560 --> 00:08:50,160
don't want to say the phrase but you have to say it gradually then suddenly it is and people just

118
00:08:50,160 --> 00:08:54,160
have such a hard time with i have a hard time with it so i mean we're doing coding automation

119
00:08:54,160 --> 00:09:01,180
stuff lots of people are but the platform that brian mentioned um the takeoff on it is so real

120
00:09:01,180 --> 00:09:07,140
it's so crazy i'll do a demo for you later on but i mean we pull up a voice assistant on our

121
00:09:07,140 --> 00:09:13,060
conference calls and you haven't even seen this yet um we nicknamed jamie after uh rogan's assistant

122
00:09:13,060 --> 00:09:17,860
which i don't think anyone's really got on that so i think it's a good nickname um but you

123
00:09:17,860 --> 00:09:23,720
literally just talk and and jamie can read your code it's not just some note-taking assistant

124
00:09:23,720 --> 00:09:26,800
where you tell it to make a note of this

125
00:09:26,800 --> 00:09:28,640
or basically transcript manipulation.

126
00:09:28,840 --> 00:09:30,000
That's what most of these things are.

127
00:09:30,400 --> 00:09:33,240
It can read through all of our graph database,

128
00:09:33,380 --> 00:09:35,300
which is all of our code, all of our chats,

129
00:09:35,400 --> 00:09:38,160
all of our conversations, all the context for the conversation.

130
00:09:38,740 --> 00:09:40,300
So how many times have you been in a meeting where you go,

131
00:09:40,400 --> 00:09:41,660
hey, is that the way it really works?

132
00:09:41,700 --> 00:09:42,400
You don't remember?

133
00:09:43,080 --> 00:09:44,240
Hey, Jamie, can you look that up

134
00:09:44,240 --> 00:09:46,340
and see if the code actually does that?

135
00:09:46,900 --> 00:09:49,160
And then once you can get past that in a meeting,

136
00:09:49,880 --> 00:09:51,800
you just have, it's such a flow state

137
00:09:51,800 --> 00:09:53,400
because you're not like, hey, let's figure this out

138
00:09:53,400 --> 00:09:55,020
and circle back together again.

139
00:09:55,100 --> 00:09:55,280
Yeah.

140
00:09:55,540 --> 00:09:57,020
You just know right away.

141
00:09:57,260 --> 00:09:59,520
Imagine if we were talking about your agent

142
00:09:59,520 --> 00:10:01,520
in this chat right now

143
00:10:01,520 --> 00:10:01,920
and you're like,

144
00:10:01,980 --> 00:10:03,340
yeah, I really wanted to do whatever.

145
00:10:03,720 --> 00:10:04,580
And then by the time we're done

146
00:10:04,580 --> 00:10:05,180
with the conversation,

147
00:10:05,300 --> 00:10:05,860
it's just done.

148
00:10:06,080 --> 00:10:06,380
Fixed it.

149
00:10:06,420 --> 00:10:07,760
I think that's what we all want, right?

150
00:10:07,880 --> 00:10:08,020
Yeah.

151
00:10:08,040 --> 00:10:09,500
I mean, you were demoing it to us

152
00:10:09,500 --> 00:10:10,480
in San Francisco in November

153
00:10:10,480 --> 00:10:12,480
and it was doing it.

154
00:10:12,600 --> 00:10:13,160
Come a long way.

155
00:10:13,680 --> 00:10:13,960
Yeah.

156
00:10:14,160 --> 00:10:16,400
Well, that's the thing as an operator,

157
00:10:17,360 --> 00:10:18,220
integrating these tools.

158
00:10:18,380 --> 00:10:19,360
To your point, Brian,

159
00:10:19,500 --> 00:10:20,660
like what is the through line

160
00:10:20,660 --> 00:10:21,620
to where this is?

161
00:10:21,700 --> 00:10:22,220
Like you don't have to worry.

162
00:10:22,220 --> 00:10:28,040
like we i tell everybody at our team here like hey this stuff's changing rapidly like something's

163
00:10:28,040 --> 00:10:30,580
going to work one week and then something better is going to come out we're gonna have to like

164
00:10:30,580 --> 00:10:36,100
start from scratch to rebuild it and it's just like this constant iteration of like okay build

165
00:10:36,100 --> 00:10:41,500
something quick it works it works better than the thing we built before but oh here's something new

166
00:10:41,500 --> 00:10:45,040
like we got to go do that again so that's what like i'm curious like when are we going to get

167
00:10:45,040 --> 00:10:51,660
to the point where yeah you don't have to keep doing that well there's a book that really impacted

168
00:10:51,660 --> 00:10:56,420
me early is called second machine age have you heard of this before yes i've read that book you

169
00:10:56,420 --> 00:11:01,160
read it it's been a while but yeah it is an old book yeah mit professors i think and they talked

170
00:11:01,160 --> 00:11:08,860
about this idea i mean don't read it but i think the main idea is combinatorial tooling so once you

171
00:11:08,860 --> 00:11:14,840
make tools that help you make other tools this is when you start to see the fast takeoff and so i

172
00:11:14,840 --> 00:11:20,300
think that book made the point of what we're seeing right now is that the software tools to

173
00:11:20,300 --> 00:11:24,960
make software tools is happening so quickly and you can automate workflows

174
00:11:24,960 --> 00:11:29,360
you can automate all the stuff that you're talking about and I think that

175
00:11:29,360 --> 00:11:33,740
the people who figure this out and can get these tools are in a permit you know

176
00:11:33,740 --> 00:11:37,700
we talked about the k-shape economy have you heard of that mm-hmm

177
00:11:37,700 --> 00:11:42,080
talked about that so I was explaining that to someone yesterday it's so true

178
00:11:42,080 --> 00:11:46,860
if which line you're on could sort of determine what it's like forever or for

179
00:11:46,860 --> 00:11:51,860
a very long time if you're on the down slope of that k then you may not ever make it to the up

180
00:11:51,860 --> 00:11:56,360
slope which to me means that we're just at this really critical moment i was talking to the best

181
00:11:56,360 --> 00:12:02,020
man at my wedding last night in georgia he works in construction management and i'm going are you

182
00:12:02,020 --> 00:12:06,580
using any of these tools have you heard of any of this nothing so i sent him a video here's how you

183
00:12:06,580 --> 00:12:14,360
get started so i think that um just the distribution of this tech is just so uneven right now what do

184
00:12:14,360 --> 00:12:17,960
we're in a bubble yeah that's what i was gonna say what do you think the penetration is like

185
00:12:17,960 --> 00:12:23,400
0.1 it's like you go on x and you think everybody's like knows everything and then you go to like a

186
00:12:23,400 --> 00:12:29,080
family gathering you're like nobody knows anything um no i mean that's been my experience it's it's

187
00:12:29,080 --> 00:12:34,760
uh we're sitting here driving around in waymos and experimenting with apple vision pros and

188
00:12:34,760 --> 00:12:41,080
playing around with claw bots and um it's it's pretty scary what i think can happen when you

189
00:12:41,080 --> 00:12:46,660
you diverge people this distinctly it's going to be tough super tough well that's again another

190
00:12:46,660 --> 00:12:51,380
question on the through line that i have like thinking through this social problem like will

191
00:12:51,380 --> 00:12:56,400
it get good enough where yes we're early adopters we're reaping the benefits of this

192
00:12:56,400 --> 00:13:02,660
massively right now but will it get good enough where it doesn't matter somebody like your best

193
00:13:02,660 --> 00:13:07,500
man will be able in a year from now to download it up just do it like that with little learning

194
00:13:07,500 --> 00:13:11,460
curve it'll accelerate so quickly I think it'll be way faster than that I

195
00:13:11,460 --> 00:13:15,740
mean what's your take on the well okay so where are you on the doomer there's

196
00:13:15,740 --> 00:13:19,560
like a divergent point here so on the one hand you could say the k-shaped

197
00:13:19,560 --> 00:13:23,340
graph which is basically to say that people there's the people who know how

198
00:13:23,340 --> 00:13:26,940
to use these tools and the people who don't right on the other hand you could

199
00:13:26,940 --> 00:13:31,000
say all these the tooling is benefiting the creators because the tooling is

200
00:13:31,000 --> 00:13:34,800
getting more accessible and easier so then it lowers the bar to creation I

201
00:13:34,800 --> 00:13:42,320
i hope that's how things come to pass and i think about my kids so i've got four daughters and the

202
00:13:42,320 --> 00:13:48,080
older two are old enough now to where i'm like working with them on some things in fact i'm

203
00:13:48,080 --> 00:13:54,480
working on this bitcoin project with them i want to tell you guys about but um like our local pool

204
00:13:55,040 --> 00:13:59,920
the community pool they have a website where you like go and see the pool schedule and times and

205
00:13:59,920 --> 00:14:06,700
order your swim gear and stuff like that it's a piece of crap website and so my

206
00:14:06,700 --> 00:14:11,320
ten-year-old I was like hey why don't we try to fix this website and the tools

207
00:14:11,320 --> 00:14:15,700
are easy enough for her to sit down at the computer and we just like iterate on

208
00:14:15,700 --> 00:14:21,580
it so if she's able to do these sorts of things she has the intention and the

209
00:14:21,580 --> 00:14:25,360
desire and she has like a point of view on how things should be and she can do

210
00:14:25,360 --> 00:14:32,800
it um and that's today i just imagine a year from now it's going to be everybody with an idea can go

211
00:14:32,800 --> 00:14:37,200
and create their idea i hope that's what the future looks like did you see replets announcement

212
00:14:37,200 --> 00:14:43,280
yesterday agent four yeah i mean i tried it yesterday it's really good i mean it's really

213
00:14:43,280 --> 00:14:49,600
good you can basically just start with a sentence and then it'll build the slides so he completely

214
00:14:49,600 --> 00:14:57,280
expanded what vibe coding or vibe building is is like you get the code you get the decks you get

215
00:14:57,280 --> 00:15:02,640
the images i mean a lot of it is stitching it together in a clever way but you stitch enough

216
00:15:02,640 --> 00:15:08,820
of that stuff together and eight-year-olds can create businesses ten-year-olds i mean what does

217
00:15:08,820 --> 00:15:14,980
this do to well that's i had a 15 year old a 15 year old girl on the show a couple months ago she's

218
00:15:14,980 --> 00:15:20,020
alpha school she's uh oh yeah yeah i saw a clip for that yeah stella she's daughter of a of a good

219
00:15:20,020 --> 00:15:24,980
friend of mine here in austin and they're obviously alpha school is probably tip of the spear of

220
00:15:25,940 --> 00:15:35,540
integrating ai tools into education and it was blowing my mind like how how robust they're the

221
00:15:35,540 --> 00:15:40,420
aperture of like what they let the kids do at the school it has me thinking like so i have three

222
00:15:40,420 --> 00:15:46,580
boys uh the oldest of which is six right now he's in kindergarten and he's just like a catholic

223
00:15:46,580 --> 00:15:51,300
school and i think it's a good school uh kindergarten it's montessori based so they're

224
00:15:51,300 --> 00:15:55,460
not really getting too much in the weeds of math and science and stuff like that but i

225
00:15:55,460 --> 00:16:00,660
really want to like go in there and be like hey we need to like sort of rejigger how you're teaching

226
00:16:00,660 --> 00:16:04,740
these kids with these and you need it in a great ai because this is what they're going to grow up

227
00:16:04,740 --> 00:16:06,200
But they're doing the opposite.

228
00:16:06,500 --> 00:16:07,740
The guy I had dinner with last night,

229
00:16:08,080 --> 00:16:12,580
his son's friend is going to one of the top boarding schools in the country

230
00:16:12,580 --> 00:16:16,320
we've all heard of, and they're just so anti-AI.

231
00:16:16,700 --> 00:16:18,220
You cannot use it.

232
00:16:18,300 --> 00:16:19,180
They're not allowing it.

233
00:16:19,240 --> 00:16:19,960
You don't touch it.

234
00:16:20,000 --> 00:16:20,540
It's evil.

235
00:16:21,760 --> 00:16:24,060
Imagine teaching that to kids right now.

236
00:16:25,040 --> 00:16:26,740
I mean, you're just, it's unbelievable.

237
00:16:26,800 --> 00:16:28,840
I'm of two minds.

238
00:16:29,120 --> 00:16:32,400
So, Marty, you haven't met my wife, but Paul's met her many times.

239
00:16:32,400 --> 00:16:33,500
She's awesome.

240
00:16:34,460 --> 00:16:47,790
She on a crusade to reduce the amount of screen time for kids in school We go to public school which I all about I think just giving these kids Chromebooks and having them sit on a computer all day long

241
00:16:47,790 --> 00:16:50,090
is not what young brains should be doing.

242
00:16:50,670 --> 00:16:54,270
On the other hand, I want them to be exposed to these tools

243
00:16:54,270 --> 00:16:57,650
and become capable with this technology because it's so cool.

244
00:16:57,650 --> 00:16:59,150
It's amazing. You can make anything.

245
00:16:59,150 --> 00:17:05,750
so I'm I'm like mixed on this you know how much to expose young people to like

246
00:17:05,750 --> 00:17:11,050
kids to this technology and encourage them to build but without having them

247
00:17:11,050 --> 00:17:14,330
sit in front of a computer screen all day yeah that's why I like the alpha

248
00:17:14,330 --> 00:17:18,830
model it's like two hours of intense like how they do integration then beyond

249
00:17:18,830 --> 00:17:22,430
that's like social skills like group projects stuff like that yeah that's

250
00:17:22,430 --> 00:17:26,590
good especially if we if things happen the way we just described where it's

251
00:17:26,590 --> 00:17:31,390
like people can have a conversation and then you end up with a product that other people can use

252
00:17:31,390 --> 00:17:36,050
that's a lot different than the hacker spend it you know not sleeping spending all night

253
00:17:36,050 --> 00:17:43,630
coding something um i hope that's the the future we find ourselves in you know i i think the the

254
00:17:43,630 --> 00:17:49,690
focus needs to be on like first principles logical thinking like clearly you that's what i've come

255
00:17:49,690 --> 00:17:56,890
to learn again going back to how i interact with my my agent is again thinking about the context

256
00:17:56,890 --> 00:18:00,890
and the direction i'm going to give it so like having to think from first principles okay what's

257
00:18:00,890 --> 00:18:05,370
the sort of logical order of instructions i need to give this thing to get the output that i need

258
00:18:06,170 --> 00:18:14,890
and i think reading comprehension uh is going to be way more important than stem moving forward

259
00:18:14,890 --> 00:18:21,130
unfortunately well you write every day so it forces you to clearly think through steps and i

260
00:18:21,130 --> 00:18:26,170
think having that cape that muscle fully exercised when you write down it's you know the you've heard

261
00:18:26,170 --> 00:18:31,290
this term spectrum and development you're no longer encoding you just you have to clearly articulate

262
00:18:31,290 --> 00:18:36,490
what you want and that's really hard to do when i sit down and try to stack out and the system's

263
00:18:36,490 --> 00:18:41,610
asking you questions well you want to select all and then what if you select all then how do you

264
00:18:41,610 --> 00:18:46,510
unselect and you're just like oh man you're right but forcing you to think through the whole thing

265
00:18:46,510 --> 00:18:51,370
is actually where the mental friction takes place so how do you teach that to the next generation

266
00:18:51,370 --> 00:18:57,190
yeah like how are you i mean you're is right so it's like how old are they and what we say to them

267
00:18:57,190 --> 00:19:02,330
i mean my oldest is he's just in the process of starting to read but i think never too late marty

268
00:19:02,330 --> 00:19:08,670
no but i think reading the classics again like uh getting like i don't know at what age you

269
00:19:08,670 --> 00:19:12,950
introduced of like literally socrates played on like how do you think like how you logically like

270
00:19:12,950 --> 00:19:18,010
aristotle yeah really old stuff yeah i mean i went to a jesuit high school and that uh and actually

271
00:19:18,010 --> 00:19:22,470
like learning latin like we had uh we were we had to take six years of language and two of which had

272
00:19:22,470 --> 00:19:28,290
to be latin and like you get down to like the etymology of of language and uh the whole goal

273
00:19:28,290 --> 00:19:31,790
of our high school was like by the end when you graduate you're gonna be able to write competently

274
00:19:31,790 --> 00:19:36,250
like you're gonna be able to read something and then have an opinion on it and write articulately

275
00:19:36,250 --> 00:19:41,550
about it and i feel very fortunate it set me up like you said i write every day but i think that

276
00:19:41,550 --> 00:19:49,670
that sort of model is going to be more important um moving forward what survives after ai then

277
00:19:49,670 --> 00:19:53,890
i don't know that's the other question like is it going to make us dumber like you you see

278
00:19:53,890 --> 00:20:00,930
people allocating their thinking to the machines and are we going to get find ourselves with the

279
00:20:00,930 --> 00:20:07,070
competency crisis a decade two decades from now where god forbid something happens to to the

280
00:20:07,070 --> 00:20:11,090
machines and then we don't know how to rebuild anything because you have a generation who never

281
00:20:11,090 --> 00:20:19,950
learned the first principles of software or math yeah but but just i i think most of society has

282
00:20:19,950 --> 00:20:25,490
gone that way like when the average person encounters a car engine they're probably i don't

283
00:20:25,490 --> 00:20:29,430
know what the hell yeah what's going on in here and then you meet someone who knows what's happening

284
00:20:29,430 --> 00:20:32,610
and it's extremely impressive.

285
00:20:33,050 --> 00:20:35,510
And that person who knows what's going on can fix it.

286
00:20:35,570 --> 00:20:37,010
They can do a lot more things.

287
00:20:37,210 --> 00:20:40,650
So I hope that that kind of competence gets rewarded.

288
00:20:41,170 --> 00:20:42,890
In the future, I hope these types of people

289
00:20:42,890 --> 00:20:46,010
are better at creating things, at advancing things

290
00:20:46,010 --> 00:20:47,950
and encourages other people to do the same thing.

291
00:20:48,410 --> 00:20:52,110
And I think from like a parenting perspective,

292
00:20:52,110 --> 00:20:55,370
I think that's an important thing to bestow on your kids.

293
00:20:55,790 --> 00:20:58,950
And honestly, my role model is Paul.

294
00:20:58,950 --> 00:21:05,550
yeah i remember the first time do you mind if i told the beaver story sure okay first time i met

295
00:21:05,550 --> 00:21:10,010
paul's younger son um came over to his house first time i went to your house

296
00:21:10,010 --> 00:21:17,950
there was a sous vide happening you know i was like what are you guys cooking like oh beaver tail

297
00:21:17,950 --> 00:21:24,850
like okay interesting and he goes on to tell me that his son just got he was like 10 at the time

298
00:21:24,850 --> 00:21:31,890
yeah 10 10 year old son just got back on a solo hunting trip to he trapped a beaver brought it

299
00:21:31,890 --> 00:21:37,390
back was suviing the beaver tail so he's cooking for the family but then before that he like

300
00:21:37,390 --> 00:21:41,750
scanned the beaver and was making a video game out of the 3d scan of the beaver

301
00:21:41,750 --> 00:21:49,910
what and then over the years i've uh come to you homeschooled your kids yeah for a while

302
00:21:49,910 --> 00:21:55,490
up until uh high school really and i just i think i've learned that you took a very you and nicky

303
00:21:55,490 --> 00:22:02,690
your wife took a very intentional approach to how you brought your kids up and they understand

304
00:22:02,690 --> 00:22:09,010
things deeply and now they're super resilient and capable um so i mean i feel like this is a tangent

305
00:22:09,010 --> 00:22:14,150
but also pretty important thing for people nowadays is to try and teach people these like

306
00:22:14,150 --> 00:22:19,730
yeah i mean i was i mean you guys are in a different phase in terms of upbringing but um

307
00:22:19,730 --> 00:22:22,730
I was right about one thing and wrong about another thing.

308
00:22:22,810 --> 00:22:24,830
I said, don't learn to code.

309
00:22:25,250 --> 00:22:28,410
I don't think that people will be writing these letters long term.

310
00:22:28,770 --> 00:22:33,930
So I wouldn't say that learning to write the letters is unnecessary

311
00:22:33,930 --> 00:22:37,330
because I think that just like learning Latin, you're not writing Latin,

312
00:22:37,690 --> 00:22:41,790
but learning Latin, which I did too, teaches you the structure underneath.

313
00:22:42,530 --> 00:22:47,630
And so I really felt like the job of writing code won't be around long.

314
00:22:47,630 --> 00:22:49,690
That's why I told them when they were that age, just don't.

315
00:22:49,730 --> 00:22:53,730
don't bother worrying about that but i also thought they would never have to drive a car

316
00:22:54,290 --> 00:22:58,290
so um that took way longer than any of us thought you know i rode in a tesla

317
00:22:58,930 --> 00:23:05,730
you know i was driving down the freeway a decade ago and i thought oh we're so close so i think and

318
00:23:05,730 --> 00:23:11,330
now that's basically here so the the question of timing is the one that's the tough part i will say

319
00:23:11,330 --> 00:23:18,930
back to our previous point just that it i know this from firsthand that you can automate basically

320
00:23:18,930 --> 00:23:25,170
two day with no advancements, no change of anything, 80% of every job done on a computer.

321
00:23:25,170 --> 00:23:32,850
So everything that we're all doing here, and 80 is pretty conservative. So go into the most,

322
00:23:32,850 --> 00:23:40,830
you know, esoteric job in some law firm, and you can automate it consistently to a degree that's

323
00:23:40,830 --> 00:23:46,090
better than even doing it. And that's a fact. And whatever we do with that fact, I'm not sure.

324
00:23:46,090 --> 00:23:55,990
But, you know, again, my son's friend who's doing finance at a public university, and I'm looking at his homework, and it's manipulating spreadsheets.

325
00:23:56,330 --> 00:23:58,370
I'm like, oh, my God.

326
00:23:58,430 --> 00:24:04,710
You're paying to have someone test you on editing an Excel spreadsheet?

327
00:24:05,550 --> 00:24:10,490
And I dropped the files in the folder and said, here's this thing called Claude Cowork.

328
00:24:10,610 --> 00:24:11,870
Go to town, you know.

329
00:24:11,870 --> 00:24:17,750
And then I had an Uber driver the other day paying $6,000 to $8,000, something like that.

330
00:24:18,210 --> 00:24:20,430
He saw the Waymos and was like, I have three.

331
00:24:20,530 --> 00:24:21,970
I've been driving for 15 years.

332
00:24:22,490 --> 00:24:26,170
I guess I have three years left before this whole job goes away.

333
00:24:26,730 --> 00:24:33,310
I'm paying $6,000 to $8,000 to learn to write QA code for the class at night.

334
00:24:33,430 --> 00:24:35,970
And I went, oh, buddy, don't do that.

335
00:24:36,110 --> 00:24:36,690
Don't do that.

336
00:24:36,750 --> 00:24:39,610
That's the job that's going away.

337
00:24:40,130 --> 00:24:41,910
Now, I mean, could you turn that into something else?

338
00:24:41,930 --> 00:24:44,610
And he goes, he's sitting there, and he just says, well, what should I do?

339
00:24:44,710 --> 00:24:47,370
And I went, that is a very good question, man.

340
00:24:48,090 --> 00:24:54,670
And I had to go with, you know, what do you love to do in your free time?

341
00:24:54,770 --> 00:24:56,010
How do you turn your hobby into something?

342
00:24:56,010 --> 00:25:00,710
But literally that was when the ride was over, and I just said, it's just stuck with me.

343
00:25:00,790 --> 00:25:01,650
What do you tell people to do?

344
00:25:01,710 --> 00:25:03,230
What do you tell your kids to study?

345
00:25:03,430 --> 00:25:04,890
What do you major in college for?

346
00:25:04,970 --> 00:25:05,810
Do you go to college?

347
00:25:05,810 --> 00:25:12,870
I mean, these are things, you know, my kids don't seem to have plans to go to college, which I don't think is a bad thing.

348
00:25:12,990 --> 00:25:23,070
I just, you know, I think they miss out on some of that critical thinking and socialization, that the college part, the collegiate part of meeting people is super critical.

349
00:25:23,810 --> 00:25:29,110
What they're teaching you, don't use AI to write this essay, is absolute crap, you know.

350
00:25:29,310 --> 00:25:30,590
So how do you separate the two?

351
00:25:30,590 --> 00:25:34,430
and so I think there are you know whatever these schools are popping up

352
00:25:34,430 --> 00:25:39,710
homeschools all that stuff I would say you know go I would advocate for that

353
00:25:39,710 --> 00:25:44,110
strongly I think that you don't want to have you don't want to be cookie cutter

354
00:25:44,110 --> 00:25:48,290
in this neck yeah you don't want to be like everyone else because you're already

355
00:25:48,290 --> 00:25:52,130
in the training data well so are all your friends they're already in the training

356
00:25:52,130 --> 00:25:55,490
data like why do you exist well like talk about gradually then suddenly this

357
00:25:55,490 --> 00:25:59,330
has been around for like we don't have an attendance policy so I could a class

358
00:25:59,330 --> 00:26:03,710
and I just use Khan Academy to teach myself like oh really yeah I basically

359
00:26:03,710 --> 00:26:10,330
took all my Khan Sal Khan I graduated in 2013 so like Sal Khan covered

360
00:26:10,330 --> 00:26:13,670
everything I studied in college I was like all right I'll just teach teach

361
00:26:13,670 --> 00:26:20,090
myself and I were I wound up working during the day and but yeah I mean to

362
00:26:20,090 --> 00:26:22,450
the point gradually and suddenly this has been possible this change has been

363
00:26:22,450 --> 00:26:28,130
emerging slowly and now it's hitting a critical tipping point where yeah it's

364
00:26:28,130 --> 00:26:35,150
it's it's here and to and I really want to dive into this like graph model

365
00:26:35,150 --> 00:26:39,470
versus LLM like sure 80% of the way there are you confident with that graph

366
00:26:39,470 --> 00:26:43,370
because again go back the LLMs that you can tell I was talking to Justin moon

367
00:26:43,370 --> 00:26:47,390
about this like the first thing he does whenever he tests out a new LLM model is

368
00:26:47,390 --> 00:26:50,870
tell me a joke and nine times out of ten the jokes very similar to the

369
00:26:50,870 --> 00:26:55,430
previous model yeah and it's basically like it's using that distribution of okay

370
00:26:55,430 --> 00:26:59,870
this is what a good joke is and you end up in the middle of the bell curve and you get a similar joke

371
00:26:59,870 --> 00:27:06,430
every time um with the graph model it seems like it has way more yeah i mean there's graph pops up

372
00:27:06,430 --> 00:27:10,690
a lot so i mean i don't know if you guys want to dive into the details or not but it's i think i

373
00:27:10,690 --> 00:27:17,210
was practicing explaining this the other day but there's really um four maybe five pieces so should

374
00:27:17,210 --> 00:27:23,390
we go through yeah i think so um and you've seen this slide from i think 2022 is when i wrote this

375
00:27:23,390 --> 00:27:30,950
slide so there's basically um people over in when chat gbt came out it was like six year olds playing

376
00:27:30,950 --> 00:27:37,390
soccer this model is going to do everything because you're so blown away by where you previously were

377
00:27:37,390 --> 00:27:42,150
versus what chat gbt can give you and so everyone was like these models just going to take over the

378
00:27:42,150 --> 00:27:47,110
world remember the thin wrapper thing don't be a thin wrapper because it's going to eat you know

379
00:27:47,110 --> 00:27:52,370
the model's going to eat everything no one talks about that anymore in fact i think open ai bought

380
00:27:52,370 --> 00:27:55,090
a workflow company late last year.

381
00:27:55,690 --> 00:27:56,910
So if the model's going to eat everything,

382
00:27:57,010 --> 00:27:57,890
then why do you have all these extras?

383
00:27:58,090 --> 00:28:00,930
So in the corner, you do have the models,

384
00:28:01,350 --> 00:28:03,670
but I think that they are very good

385
00:28:03,670 --> 00:28:05,330
at interpreting human intent

386
00:28:05,330 --> 00:28:07,530
and then communicating back out

387
00:28:07,530 --> 00:28:09,370
the pieces of logic in words.

388
00:28:09,490 --> 00:28:11,370
So I look at it as like the input-output

389
00:28:11,370 --> 00:28:11,970
to these things.

390
00:28:13,030 --> 00:28:14,830
And you can extract a ton of logic

391
00:28:14,830 --> 00:28:16,130
from the language models

392
00:28:16,130 --> 00:28:18,290
because the sentences contain logic.

393
00:28:19,110 --> 00:28:20,610
So you have the models in one corner,

394
00:28:20,610 --> 00:28:26,030
and then you have the thing that everyone's excited about right now which is agents and loops

395
00:28:26,030 --> 00:28:31,390
but any programmer will tell you is the more technical you are the more you look at this and

396
00:28:31,390 --> 00:28:38,290
you just go well what is an agent doing it's a loop with tools and language and you can do crazy

397
00:28:38,290 --> 00:28:44,070
stuff when this thing just has the right tools it can do crazy good it can do crazy bad but that's

398
00:28:44,070 --> 00:28:50,250
i think of agents as figure this problem out iterate and then get back to me that's what the

399
00:28:50,250 --> 00:28:55,610
agents are really good at. I look at them as like explorers or discoverers. I don't really know what

400
00:28:55,610 --> 00:29:01,350
the problem is. You go and do your thing and then tell me what happened. And those are agents are

401
00:29:01,350 --> 00:29:06,990
incredible. So people call it agent harnesses and stuff like that too. So language models, agents,

402
00:29:07,710 --> 00:29:13,050
and then the piece that people overlook are the workflows. You actually don't want your agent

403
00:29:13,050 --> 00:29:18,710
figuring out how it did something yesterday every single day, right? I don't need it to go and figure

404
00:29:18,710 --> 00:29:24,150
out the way I look at it is if you didn't have Google Maps and you just had your cars driving

405
00:29:24,150 --> 00:29:30,350
around town with no map they will end up at the coffee shop one day but they're just driving and

406
00:29:30,350 --> 00:29:35,370
driving and driving once you know how to get to the coffee shop write it down create a list and

407
00:29:35,370 --> 00:29:42,790
execute that list so the workflows help keep the agents on track so don't you know just go off the

408
00:29:42,790 --> 00:29:47,610
rails so when you know how to do something that's the workflow cron jobs and those are the kind of

409
00:29:47,610 --> 00:29:51,990
things schedulers you'll hear those words so that's that part of it and then

410
00:29:51,990 --> 00:29:55,170
run jobs is you would put that in the workflow bucket yeah that's basically

411
00:29:55,170 --> 00:29:59,250
triggering the workflow okay and just to just make sure I'm tracking so the agent

412
00:29:59,250 --> 00:30:04,290
and the driving around town analogy the agents loop is just go find a coffee

413
00:30:04,290 --> 00:30:07,570
shop and they're just sort of driving aimlessly whereas you give it the

414
00:30:07,570 --> 00:30:10,770
workflow that's kind of like the instructions of how to get exactly to or

415
00:30:10,770 --> 00:30:14,870
the directions to get once you've figured something out got it so input

416
00:30:14,870 --> 00:30:21,190
output with the language models then you've got the agents doing discovery using tools and

417
00:30:21,190 --> 00:30:27,190
in language and then you've got your workflow that actually can do very complex long steps and you

418
00:30:27,190 --> 00:30:30,790
just when you know what you want you generally want it to go you put it into a workflow that's

419
00:30:30,790 --> 00:30:36,070
what you were doing with it yeah or that could be like a cloud skill or something could be anything

420
00:30:36,070 --> 00:30:42,150
yeah anything that's writing down steps skills can be tools though too i'm less familiar with how

421
00:30:42,150 --> 00:30:51,030
cloud does the skills but then the last piece is memory right so um and memory right now is in the

422
00:30:51,030 --> 00:30:57,270
form of files so whenever you talk to your um claw bot or whatever you're calling these things is is

423
00:30:57,270 --> 00:31:02,310
just it's writing down these markdown files and it's reading them so the analogy people use is

424
00:31:02,310 --> 00:31:07,270
that movie um memento where the guy wakes up every morning and doesn't have any memories he has to

425
00:31:07,270 --> 00:31:12,150
read all of his tattoos. So every time the agent goes through the loop it's actually rereading the

426
00:31:12,150 --> 00:31:17,830
files and figuring that out. The problem is is that you have the language model, the agent loop,

427
00:31:17,830 --> 00:31:22,550
and you have text memory. Those combinations results in the thing you were talking about before.

428
00:31:22,550 --> 00:31:28,310
The language model can only have the attention span of so many words. The agent is just spewing

429
00:31:28,310 --> 00:31:32,950
through words like crazy, writing down, reading, and writing down, reading. And then the more

430
00:31:32,950 --> 00:31:37,910
complicated it gets the longer it gets the model can no longer hear all the agent has to say

431
00:31:38,550 --> 00:31:44,710
so then what do you do you have to shrink or truncate or compact those are the words you'll

432
00:31:44,710 --> 00:31:51,270
hear compassion um and it'll just arbitrarily take long sentences and turn them into short ones or

433
00:31:51,830 --> 00:31:58,310
crop them others many different strategies to do that that's when you start to see mistakes so now

434
00:31:58,310 --> 00:32:03,390
you have an agent with Alzheimer's because it's just missing chunks of his memory.

435
00:32:04,030 --> 00:32:05,750
So can you do files?

436
00:32:05,930 --> 00:32:07,550
What everyone's trying to figure out now is,

437
00:32:08,190 --> 00:32:10,230
are files really the best way to do this?

438
00:32:10,690 --> 00:32:14,790
Right now, they're incredible because the models and the agents

439
00:32:14,790 --> 00:32:20,730
have really great file searching capabilities using rep and bash tools.

440
00:32:20,730 --> 00:32:23,870
So they're really finely tuned to do that.

441
00:32:24,570 --> 00:32:27,430
But it's almost like the tail wagging the dog.

442
00:32:27,430 --> 00:32:32,150
hey we're really good at text tools so let's use text for all of our memory it doesn't make sense

443
00:32:32,150 --> 00:32:38,590
so what we've explored for years is using graph databases so we started out with um

444
00:32:38,590 --> 00:32:43,450
micro task payments where we pay human beings to do small tasks using lightning right that was

445
00:32:43,450 --> 00:32:49,170
our company was founded on that so we were taking complex human processes breaking them down into

446
00:32:49,170 --> 00:32:53,990
small pieces and then giving them to different human beings around the world and then reassembling

447
00:32:53,990 --> 00:32:59,950
the answer later which is exactly what you know these language models are good at and all these

448
00:32:59,950 --> 00:33:05,010
systems what you're describing is you're taking your process of of doing a podcast and breaking

449
00:33:05,010 --> 00:33:09,730
it up into small pieces giving them the different models and then reassembling it at the end we were

450
00:33:09,730 --> 00:33:29,780
doing that with humans from day one and so we were very well primed to then say oh we know how to break things up into workflows these skills humans are doing and some are automated and now just more and more is automated So we think that memory will end up in a graph structure

451
00:33:30,140 --> 00:33:31,640
whether it's a graph database or not.

452
00:33:32,260 --> 00:33:34,060
And so that's what we've been working on forever.

453
00:33:34,060 --> 00:33:39,740
And again, you're demoing Hive to us in San Francisco,

454
00:33:39,900 --> 00:33:43,420
like comparing it to how people are using these harnesses,

455
00:33:43,420 --> 00:33:50,120
something like OpenClaw, like how would you describe the benefits of leveraging this graph

456
00:33:50,120 --> 00:33:57,000
database instead of all these text files? Yeah, the graph database can basically construct

457
00:33:57,000 --> 00:34:04,040
using a search the pieces of text you need instead of a file. The reason why we moved,

458
00:34:04,040 --> 00:34:09,700
I mean, when you log into Airbnb, they're not reading a bunch of files to get the listings.

459
00:34:09,700 --> 00:34:11,100
They have a database, right?

460
00:34:11,200 --> 00:34:13,920
Because databases are way more efficient than text files.

461
00:34:14,620 --> 00:34:22,580
So somehow we've gone full circle back to pre-database land and going, we're now in the 70s going, oh, let's just use text files.

462
00:34:23,080 --> 00:34:24,740
So what do you miss with text files?

463
00:34:25,340 --> 00:34:28,340
Relationships, versioning, typing.

464
00:34:28,760 --> 00:34:31,240
It'd be like, I don't want an Excel file.

465
00:34:31,320 --> 00:34:33,240
I just want to put it all into a text file.

466
00:34:33,940 --> 00:34:36,320
You could, but you still lose some stuff.

467
00:34:36,320 --> 00:34:49,620
So that's what we're trying to point out to people is the same things that drove people away from storing all of your information in text files 50 years ago apply to memory today.

468
00:34:50,240 --> 00:34:53,020
So you have to be able to query the pieces that are relevant.

469
00:34:53,380 --> 00:35:03,220
In the background, we do so much using workflows, janitor work to clean up the context so that it can be presented in a way.

470
00:35:03,220 --> 00:35:13,560
This is what Evan from our team has really excelled at, is pre-rendering the context in the way that the model can actually understand it.

471
00:35:14,280 --> 00:35:16,200
So there's a lot of layers.

472
00:35:16,380 --> 00:35:21,460
So you have, say, every recipe you've ever cooked, all the ingredients in text files.

473
00:35:22,060 --> 00:35:25,520
But then you don't know how often you've cooked that recipe, right?

474
00:35:25,740 --> 00:35:27,940
This is a recipe I've cooked 800 times.

475
00:35:28,480 --> 00:35:29,880
Text files aren't great at that, right?

476
00:35:29,880 --> 00:35:31,060
Because they have the recipe.

477
00:35:31,060 --> 00:35:36,580
they have what you ate in a different text file let's say you're tracking your meals

478
00:35:36,580 --> 00:35:42,020
but those two things aren't linked together so how would you do that you you couldn't count and

479
00:35:42,020 --> 00:35:47,880
these things are trivial for databases to do but that's why your open claw um or your agent of any

480
00:35:47,880 --> 00:35:54,200
kind might seem dumb because it can't keep those statistics yeah whereas any database and especially

481
00:35:54,200 --> 00:35:58,680
graph database can do all these things for you yeah does that make sense i mean it does

482
00:35:58,680 --> 00:36:09,500
and I um we have uh our agent has QMD yeah Toby uh Toby yeah from Toby running in his server to

483
00:36:09,500 --> 00:36:15,620
sort of solve some of that context not exactly it's a vector database or a rag yeah um I haven't

484
00:36:15,620 --> 00:36:21,320
loaded QMD but I was just gonna bring up rag um remember you guys invested in the company too

485
00:36:21,320 --> 00:36:28,460
that does rag and so rag for people is just the the process everybody does rag all the time so

486
00:36:28,460 --> 00:36:33,200
So the term RAG is really retrieving stuff and feeding it to the model.

487
00:36:33,200 --> 00:36:38,020
The most popular way of doing RAG for years has been with a vector database.

488
00:36:38,020 --> 00:36:41,580
And no layperson touches vector databases, right?

489
00:36:41,580 --> 00:36:42,640
We've gone over this too.

490
00:36:42,640 --> 00:36:51,260
But it's basically turning text into points in a space, but you can't picture the space

491
00:36:51,260 --> 00:36:55,260
because it's not three-dimensional space, it's multi-dimensional space.

492
00:36:55,260 --> 00:37:01,780
And the example I use is if you do back to the recipe analogy would be if you do a keyword

493
00:37:01,780 --> 00:37:06,880
search for the word avocado, guacamole will not come back.

494
00:37:06,880 --> 00:37:09,580
Those are two totally unrelated spellings.

495
00:37:09,580 --> 00:37:15,260
But a vector database, guacamole and avocado will be very close to each other in this space

496
00:37:15,260 --> 00:37:17,700
because they appear all the time together.

497
00:37:17,700 --> 00:37:18,700
Right?

498
00:37:18,700 --> 00:37:19,700
Yep.

499
00:37:19,700 --> 00:37:25,140
But when you take the word guacamole and take the word avocado,

500
00:37:25,860 --> 00:37:28,760
one is from a farm bureau about avocado farming,

501
00:37:29,380 --> 00:37:32,420
and the other one is a recipe or menu in a Mexican restaurant,

502
00:37:32,880 --> 00:37:36,420
those two sources for those two words are completely unrelated.

503
00:37:36,420 --> 00:37:42,000
But the vector database cannot tell that those two things are unrelated.

504
00:37:42,700 --> 00:37:44,960
They'll just say they appear together very often.

505
00:37:44,960 --> 00:37:50,960
So if you use vector databases for RAG, you get these kind of random combinations.

506
00:37:50,960 --> 00:37:57,960
It's better than keyword search, but it actually orphans the word from the source of the word.

507
00:37:57,960 --> 00:38:06,960
So I don't know if that helps or there's a bunch of different examples, but vector databases peaked like crazy about two years ago.

508
00:38:06,960 --> 00:38:10,960
And they took a precipitous fall because of this problem.

509
00:38:10,960 --> 00:38:12,960
It's orphaned information.

510
00:38:12,960 --> 00:38:18,300
um the best example i've heard and i'm borrowing this from someone else is if you do the berkshire

511
00:38:18,300 --> 00:38:27,080
hathaway annual report and you're searching for 2023 ebitda versus 2024 a vector database

512
00:38:27,080 --> 00:38:33,620
can easily mistake those two things because it's just one character difference and the embedding

513
00:38:33,620 --> 00:38:38,580
might be very close to each other but those two numbers are very different meaning if you're trying

514
00:38:38,580 --> 00:38:43,700
to plug them into a stock picker yeah so that's the kind of thing that you get with vector and

515
00:38:43,700 --> 00:38:50,900
especially vector with code because code is so similar to each other if you search on um if you

516
00:38:50,900 --> 00:38:56,660
use vector databases purely for code search you cast almost like too wide of a net and you get a

517
00:38:56,660 --> 00:39:01,300
bunch of false positives so maybe that's too detailed no i think it's great that helped yeah

518
00:39:01,300 --> 00:39:06,020
but back to the four corners we've been living in this memory here so just to summarize you've got

519
00:39:06,020 --> 00:39:13,300
text memory or you've got um database memory and then once you get into database memory how do you

520
00:39:13,300 --> 00:39:18,580
want to what kind of database do you want to use and vector databases are just we use all three

521
00:39:18,580 --> 00:39:25,060
so it's there's no right or wrong answer um vector is much faster but then when you want to know the

522
00:39:25,060 --> 00:39:31,220
ground truth the graph databases are really good for storing and keeping things that you've learned

523
00:39:31,220 --> 00:39:32,760
that you want to know for sure.

524
00:39:33,280 --> 00:39:35,980
I do care if it's 23 or 24 EBITDA.

525
00:39:36,520 --> 00:39:38,260
Don't just randomly pick one.

526
00:39:38,740 --> 00:39:43,960
So that's where I feel like all roads are leading to graph for us.

527
00:39:44,080 --> 00:39:45,300
It's just the most convenient tool.

528
00:39:45,660 --> 00:39:46,420
So recap.

529
00:39:46,840 --> 00:39:47,040
Sure.

530
00:39:47,240 --> 00:39:48,920
LLM is like the input output.

531
00:39:50,880 --> 00:39:53,460
You've got the agents, which are the loops.

532
00:39:53,620 --> 00:39:54,920
They're out doing stuff.

533
00:39:55,180 --> 00:39:55,620
Discovery.

534
00:39:55,820 --> 00:39:57,200
Think of them like explorers.

535
00:39:57,560 --> 00:39:57,740
Explorers.

536
00:39:58,160 --> 00:39:59,460
That's why OpenClaw is so fun.

537
00:39:59,840 --> 00:40:00,200
You're like.

538
00:40:00,200 --> 00:40:08,640
just go crazy okay and then to direct them you got workflows yeah and then memory is the fourth

539
00:40:08,640 --> 00:40:14,760
point yeah okay um i think there's like we can connect this back to bitcoin pretty soon uh

540
00:40:14,760 --> 00:40:20,440
if you you'll allow me so we're in the it feels to me like we're in this era right now where

541
00:40:20,440 --> 00:40:25,260
it's so fun everybody's trying their own thing like kind of isolated and independent

542
00:40:25,260 --> 00:40:33,640
tokens. Everybody's spending tokens to create their app or whatever it is. But we're redoing

543
00:40:33,640 --> 00:40:39,920
a lot of work. Like millions of people independently are spending tokens to do the same thing.

544
00:40:39,920 --> 00:40:44,500
Now what if that thing that they spent tokens on, let's say they spent 100 tokens to do to

545
00:40:44,500 --> 00:40:52,640
create a transcript of the latest TFTC episode. You already do that. You spend your 100 tokens

546
00:40:52,640 --> 00:40:58,040
doing that you put it on an accessible graph with an l402 in front of it and

547
00:40:58,040 --> 00:41:03,380
charge 10 tokens to get that transcription that's where I think all

548
00:41:03,380 --> 00:41:06,680
of this independent work that we're all doing which is a lot of fun for

549
00:41:06,680 --> 00:41:11,180
everybody I think it starts to connect with each other that is an incredible

550
00:41:11,180 --> 00:41:15,380
point because one of the things we'll do is we'll turn like if I listen to a

551
00:41:15,380 --> 00:41:19,740
podcast like hey this was a great podcast I think we should make Twitter

552
00:41:19,740 --> 00:41:26,420
aware aware of it and or acts aware of it and a lot of people don't want the

553
00:41:26,420 --> 00:41:30,240
clip they prefer to read it and so you just transmute the the content from

554
00:41:30,240 --> 00:41:37,740
audio video and ultimately a transcript into like a long yep a long tweet but we

555
00:41:37,740 --> 00:41:42,600
are going to YouTube sort of spending tokens to download the transcript

556
00:41:42,600 --> 00:41:45,720
running it through our prompt but your point if you had a marketplace where it's

557
00:41:45,720 --> 00:41:51,420
like hey who already produced this transcripts I'll pay you 10 sats for it

558
00:41:51,420 --> 00:41:55,500
like a token arbitrage I think there's like this big thing that's this is this

559
00:41:55,500 --> 00:42:00,300
is this is very big yeah yeah it's big and I think I take a look at it yeah

560
00:42:00,300 --> 00:42:07,560
also yeah so this whole concept that and I think this is an exciting moment for

561
00:42:07,560 --> 00:42:14,580
just lightning micropayments all the stuff that we worked on for years mm-hmm

562
00:42:14,580 --> 00:42:16,620
I can record for you, yeah.

563
00:42:22,260 --> 00:42:24,060
You want me to clap really loud?

564
00:42:24,280 --> 00:42:25,760
Just tell me when you can.

565
00:42:27,600 --> 00:42:28,160
Now.

566
00:42:29,080 --> 00:42:29,420
All right.

567
00:42:29,680 --> 00:42:35,200
So this is a graph database, and we have workflows.

568
00:42:36,560 --> 00:42:38,440
Well, first, we have a topic.

569
00:42:38,560 --> 00:42:40,620
So think of this as subreddit.

570
00:42:40,620 --> 00:42:44,560
And this also leads to the question of what will remain.

571
00:42:44,740 --> 00:42:47,440
How do you make money in a post-AI world?

572
00:42:47,600 --> 00:42:48,360
It's not easy.

573
00:42:48,520 --> 00:42:50,060
I don't have a great answer for that question.

574
00:42:50,820 --> 00:42:55,660
So I think short answer is software goes to very cheap.

575
00:42:55,900 --> 00:42:57,100
Everyone's going to have software.

576
00:42:58,000 --> 00:43:02,100
Software engineers will perflate for a period, but five years, really?

577
00:43:02,520 --> 00:43:03,740
I mean, maybe not.

578
00:43:04,460 --> 00:43:06,600
So what do you end up with is marketplaces,

579
00:43:06,860 --> 00:43:10,220
and not just marketplaces, but marketplaces for agents.

580
00:43:10,620 --> 00:43:17,320
And so your agents, if you instruct it to save money, get my job done, but don't burn all my tokens.

581
00:43:18,680 --> 00:43:28,980
So if you're downloading a transcript or you're downloading a podcast and transcribing and someone else is, you're just – everyone is wasting money and you're just burning electricity and tokens for no reason.

582
00:43:28,980 --> 00:43:37,180
So if there's a source, if you instruct your agent, go to this place, and they tend to have transcripts.

583
00:43:37,380 --> 00:43:42,940
Search and then get the transcript back, and it'll be cheaper than you doing it yourself, cheaper, faster, better.

584
00:43:43,560 --> 00:43:49,520
And so these content graphs can be built by agents or human beings.

585
00:43:50,440 --> 00:43:54,980
And then what you can see here is you have –

586
00:43:54,980 --> 00:44:00,800
I'm using a shout out to Callie's, uh, claw interface.

587
00:44:00,800 --> 00:44:01,740
So I signed up for this.

588
00:44:01,740 --> 00:44:02,220
It's great.

589
00:44:02,800 --> 00:44:09,300
And you can see here, all I have to do is say search this graph or let's say.

590
00:44:09,300 --> 00:44:27,980
so clawy has an lsat that i gave it and it's using lightning to go out and retrieve data back

591
00:44:27,980 --> 00:44:34,340
from the graph and paying 10 sats to get that data and then the person who uploaded the transcript

592
00:44:34,340 --> 00:44:41,960
you hopefully is getting paid some of that money back for providing that data so if you play this

593
00:44:41,960 --> 00:44:47,640
out and there it is coming back so this is everything the graph has on this thing so

594
00:44:47,640 --> 00:44:55,660
this is scraping all tweets podcasts um reddit you can put anything you want into the graph

595
00:44:55,660 --> 00:45:01,120
but in order to make money back from the graph by putting it up there you have to stake a little

596
00:45:01,120 --> 00:45:08,000
money otherwise you get garbage so everyone is talking about how do you have an internet when

597
00:45:08,000 --> 00:45:16,000
content goes to zero the cost of content goes to zero you're just going to get spam right so um if

598
00:45:16,000 --> 00:45:24,020
creating profiles is free and creating content is free then how the heck is your agent going to find

599
00:45:24,020 --> 00:45:29,620
real information or not and so you'll have to have these wall gardens think of them as sub

600
00:45:29,620 --> 00:45:35,800
a protected subreddit it could be on k-pop on fly fishing on skiing on machine learning whatever

601
00:45:35,800 --> 00:45:42,180
the topic is and then you have a community of people and agents building this shared memory

602
00:45:42,180 --> 00:45:48,620
together and then if you instruct your retrieval agent to just get it in the most efficient way

603
00:45:48,620 --> 00:45:54,420
then they'll say yeah it'll cost me 10 cents of tokens to get this or i can spend two cents here

604
00:45:54,420 --> 00:46:02,820
but then you if you upload tftc now you'll be making sats for years to come every time someone

605
00:46:02,820 --> 00:46:08,560
retrieves it you get it back this is what um i talked about with adam curry five years ago this

606
00:46:08,560 --> 00:46:15,280
was and the way i told him i said hey there's going to be podcasting 2.0 but really that's a

607
00:46:15,280 --> 00:46:23,680
stop in the journey this is the final destination of the journey it's all content it's um

608
00:46:23,680 --> 00:46:29,000
gated so that the creators make their money without having to have advertising

609
00:46:29,000 --> 00:46:34,080
and this just wasn't going to happen I didn't believe with human beings so we

610
00:46:34,080 --> 00:46:38,220
built the first prototype of that podcasting 2.0 and streaming sass with

611
00:46:38,220 --> 00:46:45,100
you and I loved it this is you can feel how this could work but also human

612
00:46:45,100 --> 00:46:49,480
beings are going to change their habits but now you don't have to convince human

613
00:46:49,480 --> 00:46:55,980
beings change their habits the agents figure it out on their own so people like us who aren't

614
00:46:55,980 --> 00:47:01,380
great at marketing myself i'm speaking about myself this could be the golden era you just

615
00:47:01,380 --> 00:47:08,420
have to build something better and the agents find it and then once you have this rolling and your

616
00:47:08,420 --> 00:47:15,020
graph becomes the most popular graph on that topic honestly it's like open it's like owning

617
00:47:15,020 --> 00:47:21,020
beach farm property if you keep the best graph going it's the most thorough high quality reputation

618
00:47:21,020 --> 00:47:26,320
all that stuff the agents will keep coming back if you're down for a week agents are going to look

619
00:47:26,320 --> 00:47:31,400
for someone else this is the full information meritocracy but there's pay to put it up there

620
00:47:31,400 --> 00:47:36,600
but you earn it back it's like an investment and then you pay to retrieve you have to pay your way

621
00:47:36,600 --> 00:47:45,340
so this really is the value for value for the agent world anyway and but this is where like

622
00:47:45,340 --> 00:47:49,840
the human in the loop is like the curator at the tastemaker exactly right and that's going to be

623
00:47:49,840 --> 00:47:53,520
again going back to what i was saying earlier like understanding logic and like what people need and

624
00:47:53,520 --> 00:47:58,960
thinking about humanity and the age of machines like what is our edge it's funny to think about i

625
00:47:58,960 --> 00:48:03,460
was one of the many listeners of that episode with you guys talking about podcasting 2.0

626
00:48:03,460 --> 00:48:08,120
and being involved with the Bitcoin and Lightning community for a long time.

627
00:48:08,440 --> 00:48:11,520
We've been talking about machine-to-machine payments forever.

628
00:48:11,520 --> 00:48:16,180
But I think everybody, if we're being honest, when we were talking about it,

629
00:48:16,220 --> 00:48:20,340
we were thinking about a refrigerator is going to pay a Roomba or something like that.

630
00:48:20,500 --> 00:48:22,200
It was like an IoT use case.

631
00:48:22,840 --> 00:48:25,660
But really, what it's come to be is agent.

632
00:48:26,140 --> 00:48:28,100
Machine-to-machine is agentic payments.

633
00:48:28,840 --> 00:48:32,160
And so I think this graph concept, what we're talking about here,

634
00:48:32,160 --> 00:48:37,780
for me it was really eye-opening to it just makes sense like these agents are going to be going

635
00:48:37,780 --> 00:48:42,260
around on the internet trying to get information or skills or things to complete a job they're

636
00:48:42,260 --> 00:48:49,040
going to be spending tokens and they're economically rational actors the agents so if it costs them

637
00:48:49,040 --> 00:48:53,340
100 tokens to do the thing or they can get it for 10 tokens they're just going to go get it for 10

638
00:48:53,340 --> 00:48:59,120
tokens but the as paul was saying the cool opportunity is for people to cultivate these

639
00:48:59,120 --> 00:49:04,160
graphs and make sure their their graph is really good and has the highest quality content so the

640
00:49:04,160 --> 00:49:10,320
agents keep coming back to them um so yeah beachfront property is pretty if i were to do

641
00:49:10,320 --> 00:49:14,320
anything i've ever to say to my kids what to do i would say build these graphs on things that you

642
00:49:14,320 --> 00:49:21,280
know about my son's super into restoring toyota land cruisers and he was on uh claude yesterday

643
00:49:21,280 --> 00:49:28,000
or two days ago building a toyota maintenance app a web app i mean he's just going you can just

644
00:49:28,000 --> 00:49:34,800
build this and i said yeah but anyone can build that so what is actually useful collecting every

645
00:49:34,800 --> 00:49:41,760
single tip and trick about how to do this building a community and then making it available to humans

646
00:49:41,760 --> 00:49:47,040
and making it available to agents if you have those two things and you have the lead this is

647
00:49:47,040 --> 00:49:53,920
the k-shaped thing if i try to start the second subreddit on open claw no one's going to come to

648
00:49:53,920 --> 00:50:06,270
in my subreddit because there already one and it already huge So there a real advantage to be first mover It kind of like TFTC is this for Bitcoin for humans

649
00:50:06,750 --> 00:50:09,370
Right, like the humans who want to learn about Bitcoin

650
00:50:09,370 --> 00:50:11,330
come to TFTC to learn about it.

651
00:50:11,330 --> 00:50:13,790
What is the TFTC for agents?

652
00:50:13,790 --> 00:50:15,850
It's probably like this graph thing

653
00:50:15,850 --> 00:50:17,430
that's like easy to traverse,

654
00:50:17,430 --> 00:50:20,650
easy to do an economic trade to get the information.

655
00:50:20,650 --> 00:50:24,770
Like agents are probably not listening to ad reads, right?

656
00:50:24,770 --> 00:50:26,290
they're just like going to get the information.

657
00:50:26,290 --> 00:50:27,130
Yeah.

658
00:50:28,290 --> 00:50:32,310
So yeah, the information curation as like a job,

659
00:50:32,310 --> 00:50:35,190
I think it's gonna be very rewarding

660
00:50:35,190 --> 00:50:36,630
because people will focus on the things

661
00:50:36,630 --> 00:50:38,070
they're naturally interested in,

662
00:50:38,070 --> 00:50:41,050
whether it's Land Cruisers or Bitcoin.

663
00:50:41,050 --> 00:50:41,890
Yeah.

664
00:50:41,890 --> 00:50:44,510
Hopefully nobody from Zero Hedge is listening to this,

665
00:50:44,510 --> 00:50:46,930
but I have like a Zero Hedge Pro subscription

666
00:50:46,930 --> 00:50:49,090
and with that you get access to all the banking reports

667
00:50:49,090 --> 00:50:50,090
that they have.

668
00:50:50,090 --> 00:50:50,930
You just get access to-

669
00:50:50,930 --> 00:50:52,530
Zero Hedge's agent is listening to this.

670
00:50:52,530 --> 00:50:57,330
it's like i've been like so i've been running cron jobs and all the banking reports that get

671
00:50:57,330 --> 00:51:01,810
uploaded to the zero hedge pro subscribers so like i've got insight like with gold and city

672
00:51:01,810 --> 00:51:06,770
like all the european banks are saying like every day and like i just wake up to a report from

673
00:51:07,650 --> 00:51:12,690
martin the sophisticated marty agent right that uh it's like what are the banks thinking today

674
00:51:12,690 --> 00:51:17,570
he's like here's what they're all saying well people will do this with play it forward and i'm

675
00:51:17,570 --> 00:51:22,390
I'm saying that I would, but people will upload articles from behind a paywall, right?

676
00:51:22,630 --> 00:51:25,310
Same way a friend will print it to a PDF and send it to you.

677
00:51:25,910 --> 00:51:28,690
Agents will figure all this stuff out whether you want them to or not.

678
00:51:29,110 --> 00:51:29,210
Yeah.

679
00:51:29,390 --> 00:51:29,670
You know?

680
00:51:30,090 --> 00:51:36,310
So the next step, I think, so build graphs for agents with L402s in front.

681
00:51:37,410 --> 00:51:44,270
One shot, open claw, once I pasted the LSAT, knew what to do with a 402 response from the web server,

682
00:51:44,270 --> 00:51:47,270
which is, do you have a LSAT?

683
00:51:47,270 --> 00:51:48,730
Do you have this string of characters?

684
00:51:48,910 --> 00:51:50,590
I paste it in, it goes.

685
00:51:51,430 --> 00:51:54,050
The next step will be, can it buy one on its own and all that stuff.

686
00:51:54,190 --> 00:52:01,510
I think there's actually a real business for selling L4O2s where you take Bitcoin and turn it into an L4O2 in case your wallet doesn't support it.

687
00:52:02,070 --> 00:52:07,070
But I think the big next step is your personal graph.

688
00:52:07,490 --> 00:52:15,170
So this is your shared machine learning graph or your fly fishing graph.

689
00:52:15,170 --> 00:52:17,290
and think of it like a magazine.

690
00:52:17,730 --> 00:52:19,150
In the old days where you would buy

691
00:52:19,150 --> 00:52:21,290
a fly fishing magazine and read about what's going on,

692
00:52:21,370 --> 00:52:22,290
you know, the physical magazine.

693
00:52:22,830 --> 00:52:25,550
So this is like a collection point for all that information

694
00:52:25,550 --> 00:52:27,310
for enthusiasts on any topic,

695
00:52:27,650 --> 00:52:29,970
a business topic, personal topic, whatever it is.

696
00:52:31,210 --> 00:52:34,150
And then I get so sick of going to YouTube

697
00:52:34,150 --> 00:52:35,450
and to Twitter to try to keep up

698
00:52:35,450 --> 00:52:36,870
with anything I'm interested in.

699
00:52:37,550 --> 00:52:38,670
Surfing, we both surf.

700
00:52:39,710 --> 00:52:43,090
I want my personal graph to know what I'm interested in,

701
00:52:43,090 --> 00:52:47,050
give it a budget and just go get me the best surf clips from Instagram.

702
00:52:47,050 --> 00:52:48,850
I don't want to go to Instagram and YouTube.

703
00:52:48,850 --> 00:52:52,210
I just, I want five minutes of great surf clips every day.

704
00:52:52,710 --> 00:52:55,830
And I want, you know, surf clips from where my friends live.

705
00:52:55,890 --> 00:52:58,010
So I see what they're, you know, that kind of thing.

706
00:52:58,510 --> 00:53:01,190
So you'll store that on your personal graph.

707
00:53:01,350 --> 00:53:04,890
And so your graph will be talking to your agent.

708
00:53:04,890 --> 00:53:09,010
We'll have your graph and talk to the shared graph and exchanging value back

709
00:53:09,010 --> 00:53:12,670
and forth. And then it'll know what you know.

710
00:53:12,670 --> 00:53:19,630
So applying this to say the example I use is like the Joe Rogan podcast because he's a ton of interest that I like.

711
00:53:20,150 --> 00:53:20,630
I'm a bow hunter.

712
00:53:21,450 --> 00:53:23,690
I love bow hunting stories from Joe Rogan.

713
00:53:23,910 --> 00:53:26,670
There are certain ones he tells too many times.

714
00:53:27,530 --> 00:53:33,110
So my graph knows that I've heard this bow hunting story.

715
00:53:33,110 --> 00:53:40,770
so when i listen to joe rogan's podcast i can have it shrink that section of the story

716
00:53:40,770 --> 00:53:44,890
down to hey he's telling that same story again that you've already heard 50 times

717
00:53:44,890 --> 00:53:51,790
so i can just set to skip it just like i can skip ad reads right well so as you're describing this

718
00:53:51,790 --> 00:53:56,730
i'm thinking because like you brought up surfing and uh i just want to get the pinch my salt clips

719
00:53:56,730 --> 00:54:02,370
i think uh yes i think those guys are hilarious funny it's like but i don't want to depend on the

720
00:54:02,370 --> 00:54:07,890
instagram algo to like surface it to me or me to have to go exactly find it via search uh pinch

721
00:54:07,890 --> 00:54:14,390
myself how's it funny like a top tier comedian is running a surf podcast yeah this is like so funny

722
00:54:14,390 --> 00:54:19,570
and it's uh but it's like to your point like you can get it and this is like another thing that's

723
00:54:19,570 --> 00:54:24,470
been a big topic is like how do you get away from the algos exactly and you just create your own by

724
00:54:24,470 --> 00:54:31,650
curating what you want that's your that's some sort of personal agent and it's some sort of

725
00:54:31,650 --> 00:54:37,770
graph or memory i think of it as graph um is what people try to do with obsidian remember rome

726
00:54:37,770 --> 00:54:44,350
this is what you know really intense people want how to run their life but what if this were more

727
00:54:44,350 --> 00:54:50,370
passive and what if this were just as you're listening to things it's collecting what you

728
00:54:50,370 --> 00:54:55,130
know what you don't know um there's i think it's called math academy or something i don't know if

729
00:54:55,130 --> 00:55:00,550
you saw yeah yeah yeah third grader doing calculus or speed running six years of math

730
00:55:00,550 --> 00:55:05,690
all by doing a skills graph where it knows the skill depends on this skill.

731
00:55:05,990 --> 00:55:08,970
So if you want to learn, there's a Google Maps for your brain.

732
00:55:08,970 --> 00:55:10,490
It's like a spiral vector, right?

733
00:55:10,810 --> 00:55:11,110
Exactly.

734
00:55:11,790 --> 00:55:12,930
That's a graph database.

735
00:55:14,050 --> 00:55:21,890
And so if you have my math, my math graph would be a tiny fraction of their math graph.

736
00:55:22,050 --> 00:55:27,890
But I have this node colored, and it knows by referencing the shared math graph

737
00:55:27,890 --> 00:55:31,050
that the next step for me would be to learn this concept.

738
00:55:31,710 --> 00:55:34,810
So I have five minutes, and I've told my personal agent

739
00:55:34,810 --> 00:55:36,550
while I'm waiting for this plane,

740
00:55:36,870 --> 00:55:37,810
I want to learn a little math.

741
00:55:37,910 --> 00:55:39,010
Give me a math thing.

742
00:55:39,470 --> 00:55:41,650
Well, it knows that I'm sitting at a terminal,

743
00:55:42,030 --> 00:55:44,510
and so I can actually watch a short video.

744
00:55:44,890 --> 00:55:50,370
So it could construct a lesson using the video and content

745
00:55:50,370 --> 00:55:52,890
and just play me a video in the style that I like,

746
00:55:53,010 --> 00:55:55,570
the playback speed, the talking speed, the accent I want,

747
00:55:55,890 --> 00:55:57,210
male, female, whatever you want.

748
00:55:57,210 --> 00:56:03,950
just generate the content and feed it to me and the only way to get that is you have to have know

749
00:56:03,950 --> 00:56:09,790
what you know your agent has to know and store that memory and then you have to have shared you

750
00:56:09,790 --> 00:56:15,130
know someone has figured out all these math skills i don't want to do that right it wouldn't make any

751
00:56:15,130 --> 00:56:21,450
sense and so those two interactions i think for the first time we're seeing this open claw phenomena

752
00:56:21,450 --> 00:56:27,950
really open up the idea that people want this control they want their version of this which

753
00:56:27,950 --> 00:56:33,910
i think is super encouraging yeah well bringing it back to machine payable web like i bought a 21

754
00:56:33,910 --> 00:56:39,010
co computer back in 2015 uh when pelagi first launched it and we're talking about machine

755
00:56:39,010 --> 00:56:46,570
payable web via on-chain bitcoin an idea that was great too early um brian you're on the board

756
00:56:46,570 --> 00:56:50,630
of lightning labs like obviously l402 has been around for years now and it's been funny

757
00:56:50,630 --> 00:56:52,830
with the emergence of OpenClone,

758
00:56:52,910 --> 00:56:54,370
everybody talking about the agentic economy.

759
00:56:54,490 --> 00:56:55,810
Like, oh, now we have X402

760
00:56:55,810 --> 00:56:58,270
and all these other payment protocols for agents.

761
00:56:58,590 --> 00:57:00,090
And as we were talking about,

762
00:57:00,170 --> 00:57:03,590
I've had Cali, Justin Moon, Matt Corallo on

763
00:57:03,590 --> 00:57:07,530
in the last month talking about agentic payments specifically.

764
00:57:07,770 --> 00:57:10,470
And I think massive opportunity for Bitcoin,

765
00:57:10,590 --> 00:57:11,730
but I think it's still unclear to me

766
00:57:11,730 --> 00:57:13,910
how this two-sided market emerges.

767
00:57:14,750 --> 00:57:17,170
Like, and how the,

768
00:57:17,390 --> 00:57:19,730
I think demand from Bitcoiners is there

769
00:57:19,730 --> 00:57:25,150
in terms of we have bitcoin we're willing to pay for this stuff with sats over the lightning network

770
00:57:25,150 --> 00:57:31,810
or whatever it may be ultimately the the path to make the payment um but i think that the chicken

771
00:57:31,810 --> 00:57:39,330
and egg is like how do we make sure like enough people have like payment gates that are leveraging

772
00:57:39,330 --> 00:57:46,150
this tech yeah a lot to say on this topic so i remember um when elizabeth and lalu first were

773
00:57:46,150 --> 00:57:51,450
telling me about l402s i was like what are you talking about this was back in the refrigerator

774
00:57:51,450 --> 00:57:58,050
paying the Roomba machine to machine concept um but turns out they were very prescient on this

775
00:57:58,050 --> 00:58:04,810
point so you talked with matt in your last podcast about the concept of 402s we don't want to

776
00:58:04,810 --> 00:58:13,990
replay that but um having a system where you can trade value for resources where agents can trade

777
00:58:13,990 --> 00:58:17,910
value for resources is going to be a very important part to realize this future that we just described.

778
00:58:19,770 --> 00:58:25,310
I think that there's still a lot of work that needs to happen to make it easier to get up and

779
00:58:25,310 --> 00:58:30,330
running. Like you said, the three of us were Bitcoiners. We've made transactions in the

780
00:58:30,330 --> 00:58:36,770
Lightning Network. We run nodes, etc. That's still pretty hard to go from zero to one right now. It's

781
00:58:36,770 --> 00:58:44,570
pretty complicated. I would encourage people in this ecosystem, and I know Lightning Labs is

782
00:58:44,570 --> 00:58:50,990
working on this right now, to make it as easy as possible for the next Vibe Coder to integrate L402

783
00:58:50,990 --> 00:58:57,410
payments. So that's not just about putting an L402 gateway in front of some API. They need to be

784
00:58:57,410 --> 00:59:01,630
able to spin up a node. They need to be able to manage their channel balances and then liquidity.

785
00:59:02,330 --> 00:59:05,710
I still think there's quite a bit of friction there. I think that's getting solved very quickly,

786
00:59:05,710 --> 00:59:10,910
I'm excited for some announcements coming up pretty soon that are gonna lower those barriers.

787
00:59:10,910 --> 00:59:18,950
But I think if we can make it easier, there's a team called Money Dev Kit, Nick,

788
00:59:18,950 --> 00:59:23,750
who's those guys are doing a lot of good work on this front. But if we can just like lower the

789
00:59:23,750 --> 00:59:28,570
barrier for developer to get going on this, I think well that's just gonna be a massive

790
00:59:28,570 --> 00:59:34,230
accelerant for all these things. And the vision that we're talking about with the these interconnected

791
00:59:34,230 --> 00:59:39,710
knowledge graphs that that's where we're going but to get there faster we need to make it easier to

792
00:59:39,710 --> 00:59:46,250
just get started in the first place yeah i mean my our clanker he's running a phoenix d server

793
00:59:46,250 --> 00:59:53,230
and it was i mean it it was fascinating also that process because to your point like the

794
00:59:53,230 --> 01:00:01,310
lightning node management stuff for some reason i had the ldk docks up on my on my screen and i was

795
01:00:01,310 --> 01:00:05,310
all right take these docs and like see if you can spin up an ldk node and it completely like we

796
01:00:05,310 --> 01:00:10,190
banged our head for an hour and it completely um completely failed and i remember i was like all

797
01:00:10,190 --> 01:00:13,790
right let's think through this logically like instead of me giving it something while i have

798
01:00:13,790 --> 01:00:17,950
it do research like what would work best for your server setup and like what do you think

799
01:00:17,950 --> 01:00:23,390
you could actually do and then like went did some research came back it's like hey this phoenix d

800
01:00:23,390 --> 01:00:29,470
server looks looks good like i think i can give enough space on our disk like it makes sense it

801
01:00:29,470 --> 01:00:33,630
seems like it's easy enough to set up i was like all right go set it up and then set up it's like

802
01:00:33,630 --> 01:00:39,070
okay it's it's here it's running but we don't have any bitcoin in it it's awesome you need to open up

803
01:00:39,070 --> 01:00:43,790
a channel um i was like okay like what's the best way to do that he's like oh i found this bolts tool

804
01:00:43,790 --> 01:00:49,310
where you can send on chain bitcoin then it will submarine swap and automatically open up a lightning

805
01:00:49,310 --> 01:00:54,750
channel with async this is the stuff we need yeah it's got to be like less yeah it was like all right

806
01:00:54,750 --> 01:00:59,230
all right get an on-chain address from bolts and tell me where to send it and i was like boom

807
01:00:59,230 --> 01:01:05,230
and then it was up and running yep this is where i i just pictured the cat opening cash app sending

808
01:01:05,230 --> 01:01:10,350
bitcoin getting a channel yeah you know you asked us yesterday how did you get the inbound liquidity

809
01:01:10,350 --> 01:01:14,750
first question yeah so no one's gonna want to deal with any of that you're just gonna have to have

810
01:01:14,750 --> 01:01:19,870
it work your agent will have to say do you have this app do you have any bitcoin in it do you have

811
01:01:19,870 --> 01:01:25,470
any money go buy some now turn it into a channel yeah if you want to run one yourself i want to just

812
01:01:25,470 --> 01:01:32,110
like so my dream is to be able to let me back up a second so first some a few shout outs like shout

813
01:01:32,110 --> 01:01:38,510
out people like uh graham at voltage jesse at amboss the team at alby like a lot of people have

814
01:01:38,510 --> 01:01:42,990
been working on the key bits of infrastructure to make this easier we just got to keep going on that

815
01:01:42,990 --> 01:01:51,390
front I can't wait for the moment where someone they're making like viral apps so they see

816
01:01:51,390 --> 01:01:56,730
something that happens on X and then they make an app to respond to that thing they saw so I have

817
01:01:56,730 --> 01:02:03,390
an example of this that I worked on which is do you guys remember seeing that that tweet of there's

818
01:02:03,390 --> 01:02:07,950
a speed reading video it's like you can read something really fast if your eye focuses on

819
01:02:07,950 --> 01:02:14,210
the center letter. So I was like, okay, how fast can I make an app to do this? And a bunch of people

820
01:02:14,210 --> 01:02:19,750
did this too, but I wanted to like integrate L402 payments. So the website is called speedread.fit.

821
01:02:19,850 --> 01:02:24,230
So you can see it in the, in the wild, but I want to be able to take that idea and have like a

822
01:02:24,230 --> 01:02:29,390
production app ready with L402 payment gateways with the right liquidity and everything that's

823
01:02:29,390 --> 01:02:35,610
necessary in under an hour. I think that we're pretty close to that. It took me like a day to

824
01:02:35,610 --> 01:02:41,510
get it live because of some of the friction but I think if like if we can

825
01:02:41,510 --> 01:02:45,110
head in that direction where people have ideas and they can quickly convert

826
01:02:45,110 --> 01:02:49,590
those ideas to cool things that people can use we're just gonna see an

827
01:02:49,590 --> 01:02:55,550
explosion of innovation yeah I wish we had two more hours I know you got to go

828
01:02:55,550 --> 01:02:59,790
soon yeah I think yeah that was great I think I guess just final thoughts on

829
01:02:59,790 --> 01:03:06,590
all this the where we are uh in terms of approaching the suddenly moment of ai like where

830
01:03:06,590 --> 01:03:11,810
this will go by the end of the year and how bitcoin payments actually become an integral

831
01:03:11,810 --> 01:03:17,670
part of this agentic economy i mean everything takes longer than i think so you know my timelines

832
01:03:17,670 --> 01:03:26,170
are uh are always off but um it's everything that we planned on where speech and money are combined

833
01:03:26,170 --> 01:03:31,670
into one protocol and that's what we started with Sphinx and we continue to work on we pulled

834
01:03:31,670 --> 01:03:38,230
everything back knowing we needed to re-architect everything but speech and money become more

835
01:03:38,230 --> 01:03:44,030
important than ever because when content is infinite then you have to gate it somehow so

836
01:03:44,030 --> 01:03:49,950
people start to look for that I think that we'll have this concept of staking is going to come up

837
01:03:49,950 --> 01:03:56,270
a ton micro payments even nick zabo i think recently retweeted something that was counter

838
01:03:56,270 --> 01:04:01,150
to his original criticism of micro payments where people will never dedicate their brain space to

839
01:04:01,150 --> 01:04:05,870
micro payments and now the era of you don't have to dedicate your brain space to micropayments

840
01:04:05,870 --> 01:04:13,630
allocate it to the age finally here right so lightning l402s staking marketplaces

841
01:04:13,630 --> 01:04:21,030
graph databases that act as shared agent memory you can see how this just speed runs

842
01:04:21,030 --> 01:04:28,010
this is not stuff that we have to cook up every piece exists right now we're in the assembly phase

843
01:04:28,010 --> 01:04:34,130
and it's built on i think you know we all agree this is the best money out there it's global

844
01:04:34,130 --> 01:04:41,350
um and it doesn't put money into someone's pocket who made up that money right and so

845
01:04:41,350 --> 01:04:47,870
I think that the agents will be doing their owners a favor by operating on a Bitcoin standard.

846
01:04:48,050 --> 01:04:52,130
Yeah. I think especially for the listeners of this podcast who are Bitcoiners by default,

847
01:04:52,290 --> 01:04:58,830
it's like, why put your energy into this? Well, you're advancing the core principles of Bitcoin

848
01:04:58,830 --> 01:05:04,990
if you help build, because now we're all kind of contributing to the ecosystem of Bitcoin.

849
01:05:05,370 --> 01:05:10,110
The more sats are flowing around the internet, the more Bitcoin is successful and the more we're

850
01:05:10,110 --> 01:05:15,330
all achieving the bigger picture goal that we care about as people who believe in the power of Bitcoin

851
01:05:15,330 --> 01:05:21,010
and the reason for its existence. So I think there's like this really neat underlying motivation

852
01:05:21,010 --> 01:05:26,210
that everybody in this community has, which is not tied to corporate profits. It's tied to the

853
01:05:26,210 --> 01:05:31,290
mission of Satoshi, which makes it, I think when people are mission driven like that, like even if

854
01:05:31,290 --> 01:05:36,050
it's sort of deep down in the psyche, it makes it more interesting, more fun. People are more

855
01:05:36,050 --> 01:05:40,950
motivated to to see things through so you know like we've been saying it's a

856
01:05:40,950 --> 01:05:46,250
really fun time to be out there and I think everybody should be trying things

857
01:05:46,250 --> 01:05:51,350
trying to make you have an idea try to make it make it happen it's never been

858
01:05:51,350 --> 01:05:56,390
easier I need to set up a graph yeah we'll help you it'd be fun yeah and I'll

859
01:05:56,390 --> 01:06:01,610
just reiterate what I said with with Matt and I think the one core advantage

860
01:06:01,610 --> 01:06:05,910
that all this Bitcoin tech has compared to other payments mechanisms is the

861
01:06:05,910 --> 01:06:10,730
interoperability and i think that it's not it doesn't mean it's a foregone conclusion it's

862
01:06:10,730 --> 01:06:15,830
going to win but i think it is a massive edge that people need to lean into more it's you don't have

863
01:06:15,830 --> 01:06:21,010
to the agents will find the pathways that work the best and bitcoin and lightning have the best

864
01:06:21,010 --> 01:06:27,730
pathways it's just purely factually correct you know and so um now's the time and bitcoiners like

865
01:06:27,730 --> 01:06:32,430
myself aren't good at marketing so you don't have to agents will just find you so yeah we can't wait

866
01:06:32,430 --> 01:06:37,770
five years and let's not brian thank you for coming on let's harvest yeah thanks for having

867
01:06:37,770 --> 01:06:39,550
thank you thanks marty peace and love freaks