1
00:00:07,759 --> 00:00:08,039
Speaker 1: Warren.

2
00:00:08,039 --> 00:00:10,720
Speaker 2: I'm pretty excited for today's episode. Know why is that

3
00:00:11,759 --> 00:00:13,880
because I have a lot of questions about this topic,

4
00:00:14,279 --> 00:00:18,440
Like I've heard that I've heard the phrase MCP so much,

5
00:00:19,480 --> 00:00:21,879
and I just have many questions.

6
00:00:22,280 --> 00:00:23,679
Speaker 3: I mean, at some point you're going to have heard

7
00:00:23,679 --> 00:00:26,719
it too much. And with that, maybe I'll drop a

8
00:00:26,760 --> 00:00:30,199
little fact for the episode. There was a little research

9
00:00:30,280 --> 00:00:33,719
done not too long ago about the adverse impacts of

10
00:00:34,679 --> 00:00:38,159
mentioning artificial intelligence in product and service names, and they

11
00:00:38,159 --> 00:00:42,479
found that it actually significantly decreases consumer trust. Really, I

12
00:00:42,520 --> 00:00:43,320
think that tracks.

13
00:00:43,880 --> 00:00:45,880
Speaker 2: Yeah, it seems reasonable.

14
00:00:46,640 --> 00:00:48,359
Speaker 3: Like you go to Starbucks and you get some coffee

15
00:00:48,359 --> 00:00:50,399
and they're like, now with AI included, are you like,

16
00:00:50,759 --> 00:00:51,719
are you gonna be happy for that?

17
00:00:53,920 --> 00:00:59,039
Speaker 2: Absolutely? Can I get my double vine latte with AI

18
00:00:59,240 --> 00:01:00,000
on a blockchain?

19
00:01:00,560 --> 00:01:02,240
Speaker 3: We'll come with MCP on the side.

20
00:01:02,560 --> 00:01:03,200
Speaker 2: Okay.

21
00:01:03,240 --> 00:01:03,560
Speaker 1: Cool.

22
00:01:04,159 --> 00:01:06,640
Speaker 2: So speaking of which, Gil, you're here to talk to

23
00:01:06,760 --> 00:01:07,560
us about MCP.

24
00:01:08,200 --> 00:01:09,959
Speaker 4: Yeah, I'm very excited too, and I am one of

25
00:01:09,959 --> 00:01:12,640
those people who's actually heard MCP way too many times,

26
00:01:12,719 --> 00:01:14,280
so happy to do that.

27
00:01:14,760 --> 00:01:19,519
Speaker 2: All right, Before we jump into that, give me a

28
00:01:19,560 --> 00:01:20,719
little bit about your background.

29
00:01:21,159 --> 00:01:24,239
Speaker 4: Absolutely, So, as you mentioned, I'm Gil, I am the

30
00:01:24,280 --> 00:01:27,920
co founder and CTO of Merge and Merge is a

31
00:01:27,959 --> 00:01:32,319
platform that offers unified epis to help companies offer integrations

32
00:01:32,519 --> 00:01:35,799
with a ton of different products in any specific vertical

33
00:01:35,840 --> 00:01:39,599
from ticketing, CRM, file storage, and so on, and a

34
00:01:39,599 --> 00:01:41,959
lot more coming in the AI space MCP all of

35
00:01:41,959 --> 00:01:43,400
that as we'll talk about today.

36
00:01:44,040 --> 00:01:45,120
Speaker 1: But before that, I.

37
00:01:45,079 --> 00:01:47,359
Speaker 4: Went to college in New York and then ended up

38
00:01:47,400 --> 00:01:49,920
going straight into tech. So I worked at LinkedIn out

39
00:01:49,959 --> 00:01:54,280
in San Francisco, and then worked at a couple smaller startups,

40
00:01:54,480 --> 00:01:57,519
which ultimately led me to this problem of integration, seeing

41
00:01:57,519 --> 00:02:00,239
how it's just bogging down space, so I decided to.

42
00:02:00,040 --> 00:02:02,200
Speaker 1: Star Merge to tackle that problem. And it's been great.

43
00:02:02,239 --> 00:02:04,879
Speaker 4: It's been about five years now, four years since coming

44
00:02:04,879 --> 00:02:06,640
out of Stealth. We've gone from zero to one hundred and

45
00:02:06,640 --> 00:02:09,599
ten employees. We have almost fifteen thousand free and paying

46
00:02:09,599 --> 00:02:12,520
customers all around the world. We have our three offices

47
00:02:12,560 --> 00:02:14,680
in San Francisco, New York, and Berlin.

48
00:02:15,319 --> 00:02:21,360
Speaker 2: Right on right on. So with four years like that's

49
00:02:21,400 --> 00:02:24,199
a pretty solid record for a startup, you're feeling pretty

50
00:02:24,199 --> 00:02:25,280
confident about this.

51
00:02:26,080 --> 00:02:26,439
Speaker 1: Yeah.

52
00:02:26,479 --> 00:02:28,840
Speaker 4: Absolutely, I think that we're seeing that this is a

53
00:02:28,879 --> 00:02:32,400
bigger problem than we have ever even envisioned. Everyone needs integrations.

54
00:02:32,439 --> 00:02:35,199
The problems only getting worse now with AI. You have

55
00:02:35,280 --> 00:02:38,800
these models that have essentially ingested the full public corpus

56
00:02:38,800 --> 00:02:41,639
of Internet data, and all that's left is private data.

57
00:02:41,680 --> 00:02:42,639
Speaker 1: And that's what Merge does.

58
00:02:42,680 --> 00:02:46,080
Speaker 4: We specifically help companies get access to their customers data.

59
00:02:46,400 --> 00:02:48,520
So we're excited about the problem now, but also where

60
00:02:48,520 --> 00:02:53,159
it's going, both in the traditional API and integration building

61
00:02:53,159 --> 00:02:55,400
space as well as all of the upcoming AI and

62
00:02:55,520 --> 00:02:57,879
MCP driven integrations right on.

63
00:02:58,080 --> 00:03:02,879
Speaker 2: So like is the concept there? Instead of having to

64
00:03:02,919 --> 00:03:05,919
go and figure out the API docs for the fifteen

65
00:03:05,960 --> 00:03:09,360
different services I need to integrate with, I just connect

66
00:03:09,400 --> 00:03:13,919
with Merge and you're like the relay to those services

67
00:03:13,919 --> 00:03:16,479
for me, and I just have to talk to one service.

68
00:03:16,520 --> 00:03:17,400
Speaker 1: That's exactly right.

69
00:03:17,479 --> 00:03:21,520
Speaker 4: So an example here brax Ramp, which are corporate cards.

70
00:03:21,680 --> 00:03:23,719
Their customers have merged and they use us to power

71
00:03:23,759 --> 00:03:27,400
a few use cases. But notably they want to automatically

72
00:03:27,400 --> 00:03:31,560
onboard employees of companies that use their credit cards, automatically

73
00:03:31,599 --> 00:03:33,599
mail them a credit cards to their home address based

74
00:03:33,639 --> 00:03:35,719
on their title. Maybe give them twenty dollars a day

75
00:03:35,719 --> 00:03:39,000
for lunch if they're an engineer, for example, and then

76
00:03:39,159 --> 00:03:41,599
they want to terminate those cards when person leaves the company,

77
00:03:41,599 --> 00:03:44,639
and manually managing all of that is impossible, and so

78
00:03:45,120 --> 00:03:47,199
instead they want to integrate with the HR systems to

79
00:03:47,199 --> 00:03:49,039
be able to pull in all this data. But some

80
00:03:49,080 --> 00:03:52,599
of their customers use Bamboo HR. Gusto namely worked a

81
00:03:52,639 --> 00:03:55,560
SAP and they have to build all of those integrations.

82
00:03:55,599 --> 00:03:58,719
So instead they integrate once with us, and we integrate

83
00:03:58,719 --> 00:04:01,400
and sort of normalize all that data to one format that.

84
00:04:01,360 --> 00:04:02,159
Speaker 1: They integrate with.

85
00:04:02,680 --> 00:04:04,120
Speaker 4: And then again we do that for HR, but we

86
00:04:04,159 --> 00:04:06,759
also do that for a lot of other platforms like ticketings, RM,

87
00:04:06,840 --> 00:04:08,000
file storage and so on.

88
00:04:08,360 --> 00:04:12,840
Speaker 2: Right on, that's cool, that's cool. So let's jump into MCP.

89
00:04:13,360 --> 00:04:17,959
Give me the give me like the layman's version of

90
00:04:18,000 --> 00:04:19,040
what MCP is.

91
00:04:19,879 --> 00:04:22,879
Speaker 4: Yeah, so there's a lot of ways to think about MCP,

92
00:04:23,079 --> 00:04:25,480
but I think the important node here is it's actually

93
00:04:25,480 --> 00:04:28,639
a really simple concept. It was a standard that was

94
00:04:28,720 --> 00:04:32,199
reached similar to other protocols and standards of the Internet

95
00:04:32,240 --> 00:04:35,360
that ultimately aren't so complicated, but solved the need, which

96
00:04:35,399 --> 00:04:38,120
was that there was a major lack of standardization. And

97
00:04:38,160 --> 00:04:41,439
so when we think about the history of APIs building

98
00:04:41,439 --> 00:04:45,160
an integration required you to go to API documentation and

99
00:04:45,240 --> 00:04:47,560
explicitly say I'm going to take data from this point

100
00:04:47,800 --> 00:04:49,399
and I'm going to move it to this point. A

101
00:04:49,399 --> 00:04:52,519
lot of complications there, but overall you're doing that. Now

102
00:04:52,600 --> 00:04:56,000
in the agentic era, you want to expose those API

103
00:04:56,120 --> 00:04:58,879
calls essentially to an agent so that you can actually

104
00:04:58,879 --> 00:05:01,120
give it arms as a to them being these sort

105
00:05:01,160 --> 00:05:03,240
of you know, things to say, Hey, here's how you

106
00:05:03,279 --> 00:05:06,360
solve this. Go log into your Salesforce account, click this button,

107
00:05:06,439 --> 00:05:09,439
click that instead letting the agents actually take those actions

108
00:05:09,480 --> 00:05:12,600
for you. And so MCP is essentially a way to

109
00:05:12,680 --> 00:05:15,839
make it so that those API calls or actions or

110
00:05:15,879 --> 00:05:19,920
tool calls is as you would say, are available and

111
00:05:20,000 --> 00:05:21,600
exposed to the agents in a way that it can

112
00:05:21,639 --> 00:05:24,720
easily understand, and so then it can formulate a workflow

113
00:05:24,720 --> 00:05:28,519
knowing what tools via MCP are available to it and

114
00:05:28,560 --> 00:05:29,639
then make those calls.

115
00:05:29,959 --> 00:05:33,040
Speaker 3: What would stop an agent from being able to integrate

116
00:05:33,120 --> 00:05:36,680
with the existing API docs or one of the standards

117
00:05:36,720 --> 00:05:39,959
like open API Specification or whatever Ataboose is using with

118
00:05:40,040 --> 00:05:43,480
Smithy just consume that and be able to generate the

119
00:05:43,519 --> 00:05:45,079
appropriate calls into the APIs.

120
00:05:45,639 --> 00:05:46,639
Speaker 1: So there's a.

121
00:05:46,639 --> 00:05:48,959
Speaker 4: Few things that are that are differentiators, but overall, I

122
00:05:49,000 --> 00:05:50,759
will say, at a high level, I'm with you. I

123
00:05:50,759 --> 00:05:52,759
think it's it's not even a hot take anymore to

124
00:05:52,759 --> 00:05:55,639
say that MCP doesn't actually do all that much. You

125
00:05:55,720 --> 00:05:59,279
have API documentation that's built for humans to build integrations.

126
00:05:59,439 --> 00:06:01,120
Speaker 1: You have open API specs, which.

127
00:06:00,920 --> 00:06:03,439
Speaker 4: Were kind of the next evolution of hey, some static

128
00:06:03,480 --> 00:06:06,439
script can understand my API and build docs or SDKs

129
00:06:06,480 --> 00:06:08,560
out of it. And then finally you have MCP, which

130
00:06:08,600 --> 00:06:12,360
is just another type of rapper that allows agents to interact.

131
00:06:12,560 --> 00:06:15,959
I think the difference is MCP is not stateless like

132
00:06:16,000 --> 00:06:19,319
documentation or like you know, like an open API spect

133
00:06:19,480 --> 00:06:22,240
but instead it's an actual running server that stores credentials,

134
00:06:22,399 --> 00:06:26,519
that manages sessions and can be stateful. So it's it's

135
00:06:26,560 --> 00:06:29,399
not ultimately adding a ton more, but it does unlock

136
00:06:29,439 --> 00:06:33,160
a few additional abilities that are necessary for an agent

137
00:06:33,199 --> 00:06:33,959
to take actions.

138
00:06:34,519 --> 00:06:36,600
Speaker 3: That's I feel like a little scary, like I've I've

139
00:06:36,680 --> 00:06:39,639
designed the perfect service that is stateless in every way

140
00:06:39,680 --> 00:06:41,879
with you know, I've thought a lot about what the

141
00:06:41,959 --> 00:06:44,120
endpoint should be, and in order for an agent to

142
00:06:44,160 --> 00:06:47,160
work with it, we're saying you need to actually forget

143
00:06:47,199 --> 00:06:50,560
everything that you've built so far, make a different doc

144
00:06:50,639 --> 00:06:53,199
that is readable, and also start storing state and do

145
00:06:53,519 --> 00:06:55,680
a lot of extra stuff that you specifically didn't want

146
00:06:55,720 --> 00:06:56,839
in your service to begin with.

147
00:06:58,000 --> 00:07:00,360
Speaker 4: Yeah, it's true, I think that, you know, when you

148
00:07:00,399 --> 00:07:02,480
think about like a statically coded app, though you would

149
00:07:02,480 --> 00:07:04,079
pull from you know, let's say that you want to

150
00:07:04,160 --> 00:07:07,639
update all records, all ticketing system tickets that have Gill

151
00:07:07,680 --> 00:07:09,800
in the title. You have to pull in all those tickets,

152
00:07:09,879 --> 00:07:11,720
then you have to iterate through them and modify them,

153
00:07:11,759 --> 00:07:13,360
and then you need to write them back. So there's

154
00:07:13,360 --> 00:07:17,680
this notion of like state needs to be maintained between calls.

155
00:07:17,839 --> 00:07:19,800
Speaker 1: And an agent. An agent can do that.

156
00:07:19,759 --> 00:07:22,519
Speaker 4: But you can also write tools on the server that actually,

157
00:07:22,519 --> 00:07:24,879
you know, manage a lot of that. So what you're

158
00:07:24,879 --> 00:07:27,639
doing is you're deciding how much complexity you want to

159
00:07:27,680 --> 00:07:29,600
expose to the agent and how much you want to

160
00:07:29,639 --> 00:07:32,720
wrap behind a statically coded tool that the agent can

161
00:07:32,759 --> 00:07:33,240
then call.

162
00:07:33,600 --> 00:07:37,360
Speaker 3: So normally you would push the if your API was

163
00:07:37,759 --> 00:07:42,079
fetch ticket and then update ticket as two independent API calls,

164
00:07:42,120 --> 00:07:44,199
and that's all you offered. You were like, we don't

165
00:07:44,240 --> 00:07:48,199
offer bulk support, and the client would be responsible for

166
00:07:48,240 --> 00:07:51,079
actually pulling each one of those and then updating them individually.

167
00:07:51,079 --> 00:07:52,959
And if we're saying that you're building a proxy in

168
00:07:53,000 --> 00:07:55,959
between there, it's because the API that you're offering to

169
00:07:56,040 --> 00:07:59,240
end users wasn't valuable enough that it wasn't actually solving

170
00:07:59,279 --> 00:08:01,680
the needs that they would frequently have. I don't think

171
00:08:01,800 --> 00:08:04,399
bulk actions are very common, so I think in that

172
00:08:04,439 --> 00:08:06,439
way this is a little interesting to saying, like, well,

173
00:08:06,680 --> 00:08:08,279
now you need to actually start thinking about what the

174
00:08:08,360 --> 00:08:11,279
value of your service is actually offering, because if people

175
00:08:11,319 --> 00:08:13,920
want to do these bulk actions, you may think about

176
00:08:14,000 --> 00:08:16,600
wanting to provide that, and this is where you would

177
00:08:16,639 --> 00:08:20,439
prevent actually provide this logic in a agent based system.

178
00:08:21,120 --> 00:08:24,040
Speaker 4: I mean absolutely, and I even say this about MCP service.

179
00:08:24,079 --> 00:08:27,240
They're not unlocking anything that new. They are fully limited

180
00:08:27,319 --> 00:08:30,600
by the capabilities of the underlying API. Notably this is

181
00:08:30,759 --> 00:08:34,360
MCP for API interactions. You can use MCP to wrap

182
00:08:34,440 --> 00:08:36,600
SEQL calls to a local database, that sort of thing.

183
00:08:37,320 --> 00:08:39,480
But yeah, and one of the big problems you see

184
00:08:39,519 --> 00:08:42,279
in the integration space and have seen for years, is

185
00:08:42,320 --> 00:08:44,960
that you often have to pull full data sets. And

186
00:08:45,000 --> 00:08:47,480
the reason you have to pull full data sets is

187
00:08:47,639 --> 00:08:49,919
that you don't have good search endpoints in these APIs,

188
00:08:50,039 --> 00:08:52,759
So I would say, I would say most inefficient integrations

189
00:08:52,799 --> 00:08:56,679
are built by bad APIs, not necessarily bad consumption patterns

190
00:08:56,879 --> 00:08:57,559
or anything else.

191
00:08:57,759 --> 00:09:00,200
Speaker 3: Oh they're going to say bad engineers, and that I would.

192
00:09:01,200 --> 00:09:06,080
Speaker 4: I mean, yeah, hopefully that one's not going to be

193
00:09:06,279 --> 00:09:09,200
The engineering skill level won't matter as much anymore. We're

194
00:09:09,200 --> 00:09:10,600
working on a new product right now, and I can

195
00:09:10,639 --> 00:09:12,919
tell you from using AI to build it, the skill

196
00:09:12,919 --> 00:09:14,600
gap is closing really really fast.

197
00:09:14,639 --> 00:09:16,720
Speaker 1: It is so good now interesting.

198
00:09:16,799 --> 00:09:21,879
Speaker 2: So so you think the vibe coding is leveling up?

199
00:09:22,120 --> 00:09:23,919
Speaker 1: Oh yeah, Oh yeah, it's scary to say.

200
00:09:24,039 --> 00:09:26,399
Speaker 4: Look, I'm an engineer who's I've been coding for for

201
00:09:26,440 --> 00:09:28,159
almost actually over twenty years now.

202
00:09:28,440 --> 00:09:28,960
Speaker 1: I love it.

203
00:09:29,000 --> 00:09:31,720
Speaker 4: I'm so passionate about building, and it's scary to see

204
00:09:31,720 --> 00:09:34,159
it just writing code that you would have written. But

205
00:09:34,200 --> 00:09:37,000
I also think it's really leveling me up too. We

206
00:09:37,200 --> 00:09:39,879
vibe coded, so me, a product manager, and a software

207
00:09:39,919 --> 00:09:42,840
engineer are building this new product. Our product manager, who

208
00:09:42,879 --> 00:09:45,000
has never written a line of code in his life,

209
00:09:45,240 --> 00:09:47,960
built our entire front end for it. We imported that

210
00:09:48,000 --> 00:09:50,559
into GitHub. Then we took over with Windsurf and started

211
00:09:50,960 --> 00:09:53,919
started you know, actually doing some a little bit more

212
00:09:53,919 --> 00:09:55,679
guided AI coding on top of it.

213
00:09:56,120 --> 00:09:58,159
Speaker 1: But it worked and it was great, and then we

214
00:09:58,159 --> 00:10:00,440
just connected that straight to the back end right on.

215
00:10:00,600 --> 00:10:03,360
Speaker 2: Yeah, I think that's been my experience. Like it's really

216
00:10:03,600 --> 00:10:08,440
easy to get started with AI, but then I think

217
00:10:08,679 --> 00:10:12,799
after like those initial few steps, I think it's still

218
00:10:12,799 --> 00:10:16,720
important to have the technical skills. And I treat AI,

219
00:10:17,440 --> 00:10:19,360
and the way I try to get people to treat

220
00:10:19,360 --> 00:10:22,759
AI is think of it as if you had your

221
00:10:22,799 --> 00:10:26,480
own intern or your own junior engineer, you know, like,

222
00:10:26,600 --> 00:10:30,919
don't treat the AI as a principal architect. Treat it

223
00:10:30,960 --> 00:10:35,960
as an intern, and give it very small scoped tasks

224
00:10:36,039 --> 00:10:38,879
that you can check up on, because it's it does

225
00:10:38,919 --> 00:10:42,440
sometimes get things wrong and just like your intern would,

226
00:10:42,480 --> 00:10:44,960
so you've got to give it a task that you

227
00:10:45,000 --> 00:10:48,120
can follow up and make sure that it's continuing to

228
00:10:48,200 --> 00:10:51,200
build towards the same end goal that you are. And

229
00:10:51,240 --> 00:10:53,159
I think that's where a lot of projects get off

230
00:10:53,200 --> 00:10:56,000
the rails, as they just give AI like this vague

231
00:10:56,120 --> 00:11:01,279
task without the guardrails to keep it from wandering off

232
00:11:01,279 --> 00:11:02,080
and hallucinating.

233
00:11:02,679 --> 00:11:05,480
Speaker 3: I want to probe you on that well, definitely, because

234
00:11:05,759 --> 00:11:09,200
I see a lot of companies are not hiring interns

235
00:11:09,679 --> 00:11:11,919
and have no idea what to do with some interns,

236
00:11:11,919 --> 00:11:16,480
And yet they're coming up and hiring llms out there

237
00:11:16,519 --> 00:11:19,639
in the world to interact with, and I don't have

238
00:11:19,720 --> 00:11:22,759
the faith that they are capable of understanding how to

239
00:11:22,799 --> 00:11:26,320
provide that additional contact. So are you optimistic about where

240
00:11:26,639 --> 00:11:28,519
the quality of software in the world is going?

241
00:11:29,279 --> 00:11:33,159
Speaker 2: I think so, yeah, because like on a large enough

242
00:11:33,200 --> 00:11:35,879
time frame, like this is going to work.

243
00:11:35,759 --> 00:11:37,080
Speaker 1: Itself out, you know.

244
00:11:37,159 --> 00:11:39,320
Speaker 3: Yeah, the death of the universe right around the corner.

245
00:11:39,679 --> 00:11:42,399
Speaker 2: Yeah, absolutely, Like in the big scale of things, it

246
00:11:42,399 --> 00:11:46,200
doesn't really matter. But no, like specifically to your question,

247
00:11:46,279 --> 00:11:51,440
in people like engineers not having the skills to guide

248
00:11:51,480 --> 00:11:54,559
an intern, I would agree with that. I would include

249
00:11:54,559 --> 00:11:58,559
myself in that bucket as well as something I've had

250
00:11:58,600 --> 00:12:04,200
to learn and improve on. And I think that's where

251
00:12:05,200 --> 00:12:08,679
I think that's the skill gap that AI doesn't cover.

252
00:12:08,879 --> 00:12:13,360
So AI can write the technical code, but as an engineer,

253
00:12:15,399 --> 00:12:20,639
your value add into that equation is maintaining the big

254
00:12:20,679 --> 00:12:25,039
picture and breaking that down into concrete, isolated tasks. It

255
00:12:25,080 --> 00:12:31,320
can be distributed to junior engineers, interns, AIS, senior engineers,

256
00:12:31,559 --> 00:12:33,559
whatever is on your team.

257
00:12:34,039 --> 00:12:34,879
Speaker 1: I totally agree with that.

258
00:12:34,919 --> 00:12:36,480
Speaker 4: And I think we see this a lot. We see

259
00:12:36,480 --> 00:12:38,799
a fear of adopting AI from a lot of engineers

260
00:12:38,919 --> 00:12:41,200
or and it's not necessarily a fear of you know, oh,

261
00:12:41,200 --> 00:12:42,799
it's going to take my job, But I think it's

262
00:12:42,840 --> 00:12:45,960
a fear of it building bad software, or it just

263
00:12:46,120 --> 00:12:48,759
changing the way that someone is used to building in general.

264
00:12:49,519 --> 00:12:51,080
And I think the way that I explain this to

265
00:12:51,120 --> 00:12:53,600
my team is, you know, yeah, you're frustrated because it

266
00:12:53,679 --> 00:12:56,919
generated bad code number one, that's time for introspection.

267
00:12:57,039 --> 00:13:00,360
Speaker 1: What did I do wrong? How do I prompt better? Yeah?

268
00:13:00,480 --> 00:13:03,120
Speaker 4: But also, you know, I think that a lot of

269
00:13:03,120 --> 00:13:05,000
people see this as like, Okay, it's spent out really

270
00:13:05,120 --> 00:13:07,639
terrible code, I'm going to waste so much time cleaning

271
00:13:07,720 --> 00:13:09,759
it up. But what I'm not thinking about it's like, Okay,

272
00:13:09,840 --> 00:13:12,279
what if that eight hours buildings by hand? Or I

273
00:13:12,679 --> 00:13:15,320
can spend thirty minutes prompting it to create something, and

274
00:13:15,320 --> 00:13:17,519
then thirty minutes to an hour cleaning up the code,

275
00:13:17,559 --> 00:13:19,279
and then I've spent you know, a fourth of the

276
00:13:19,360 --> 00:13:22,080
time just maybe not how I'm used to building.

277
00:13:23,440 --> 00:13:23,799
Speaker 3: Yeah.

278
00:13:23,840 --> 00:13:26,399
Speaker 2: One of the big takeaways I've had from working with

279
00:13:26,519 --> 00:13:31,080
AI has been applying that to the rest of my life.

280
00:13:31,519 --> 00:13:34,519
Like I give AI a bad prompt, it writes bad code,

281
00:13:34,559 --> 00:13:37,039
and I spend a bunch of time cleaning it up

282
00:13:37,120 --> 00:13:40,200
and then thinking about what should I have said differently?

283
00:13:40,799 --> 00:13:44,600
And then I started looking at like other conversations in

284
00:13:44,639 --> 00:13:47,159
my life, and when I'm talking to humans now, I

285
00:13:47,279 --> 00:13:50,120
use that same process, and I realize, like, a lot

286
00:13:50,159 --> 00:13:52,399
of the pain and suffering I've had in my life

287
00:13:52,440 --> 00:13:55,000
is because I gave another human a bad prompt.

288
00:13:55,799 --> 00:14:02,399
Speaker 3: Yeah, wow, you have relationships built on lies. Right, It's

289
00:14:02,440 --> 00:14:04,120
like I need to go back to this first conversation

290
00:14:04,159 --> 00:14:07,120
I ever had with this person and maybe change what

291
00:14:07,200 --> 00:14:09,600
I said to them, because that that's that set me

292
00:14:09,679 --> 00:14:11,440
up for a success or failure.

293
00:14:12,240 --> 00:14:12,440
Speaker 1: Right.

294
00:14:13,240 --> 00:14:17,200
Speaker 2: Yeah, It's like this whole chain of events that happened

295
00:14:17,240 --> 00:14:20,919
in my life could have been corrected had I given

296
00:14:20,919 --> 00:14:23,240
that person the right prompt to begin with.

297
00:14:24,440 --> 00:14:26,799
Speaker 4: You very well may have just changed my dating life forever.

298
00:14:26,840 --> 00:14:30,639
It wasn't expected outcome of this call, but it really hasn't.

299
00:14:30,639 --> 00:14:35,559
Speaker 3: It just become like you just sick one agentic agent

300
00:14:35,879 --> 00:14:38,399
at some dating service, and it will talk to another

301
00:14:38,759 --> 00:14:41,440
LM out there and then they'll decide collectively whether or

302
00:14:41,519 --> 00:14:42,639
not to start your relationship.

303
00:14:43,440 --> 00:14:45,639
Speaker 4: Yeah, I mean it's once once Tender release is their

304
00:14:45,639 --> 00:14:46,720
official MCP server.

305
00:14:46,759 --> 00:14:47,399
Speaker 1: We're going. I'm in.

306
00:14:48,639 --> 00:14:51,120
Speaker 3: I mean, I assume people are doing this. They're just

307
00:14:51,120 --> 00:14:54,759
scraping the app and you know, uploading data anyway. I mean,

308
00:14:55,000 --> 00:14:56,679
or it's a human in the loop still, like it's

309
00:14:56,679 --> 00:14:59,559
still telling you what to type. Uh, you're just you're

310
00:14:59,639 --> 00:15:02,159
just doing Yeah, Like I don't. I don't think automating

311
00:15:02,200 --> 00:15:03,639
it is going to mean maybe that's what we're missing.

312
00:15:03,679 --> 00:15:05,519
We're missing the automation so people can get on more

313
00:15:05,600 --> 00:15:06,279
dates faster.

314
00:15:07,279 --> 00:15:09,600
Speaker 1: I've been I've been seeing the tiktoks because I guess

315
00:15:09,600 --> 00:15:12,440
the dating apps are going after people who are automating.

316
00:15:12,200 --> 00:15:14,120
Speaker 4: Of some people who set up an actual phone in

317
00:15:14,120 --> 00:15:15,600
a room with a little rubber hand on.

318
00:15:15,559 --> 00:15:18,159
Speaker 1: A screwdriver or on a on a drill that's just swiping.

319
00:15:20,519 --> 00:15:22,600
Speaker 3: I mean, it's a it's a miss market for them.

320
00:15:22,600 --> 00:15:24,519
Like if all of your users are doing something we

321
00:15:24,600 --> 00:15:26,320
know from there's like some great books out there like

322
00:15:26,360 --> 00:15:29,399
Platform Revolution that you should actually, other than trying to

323
00:15:29,559 --> 00:15:32,120
prevent that behavior and punish people for it, realize that

324
00:15:32,240 --> 00:15:35,720
is where the value is being added. And like I

325
00:15:35,720 --> 00:15:38,679
think the swiping left or right is the action which

326
00:15:38,720 --> 00:15:41,759
makes people feel invested in the action they're taking so

327
00:15:41,799 --> 00:15:44,039
that they're more likely to continue. So, you know, give

328
00:15:44,039 --> 00:15:47,039
them that capability, but give them what they want, which

329
00:15:47,080 --> 00:15:50,080
is like maybe a multi select option or or you know,

330
00:15:50,120 --> 00:15:51,840
something like that, and to take it to the next level.

331
00:15:51,919 --> 00:15:54,480
Rather than banning those people, let them have that functionality

332
00:15:54,559 --> 00:15:57,200
like once a day and then you know, up charge

333
00:15:57,279 --> 00:16:00,000
for it for the you know, next one hundred matches.

334
00:16:01,200 --> 00:16:03,559
Speaker 1: Yeah, the AI matchmaker, Yeah right.

335
00:16:03,440 --> 00:16:07,279
Speaker 2: Exactly, well, yeah right, just be open with it, like say, okay,

336
00:16:07,399 --> 00:16:09,919
I see, I see what you're up for here, and

337
00:16:09,960 --> 00:16:12,039
I'm gonna make it easy for you. You just gotta

338
00:16:12,519 --> 00:16:14,960
upgrade your service level here, sign up for the next

339
00:16:14,960 --> 00:16:16,279
plan and I got you dude.

340
00:16:17,039 --> 00:16:18,240
Speaker 1: Mm hm cool.

341
00:16:18,320 --> 00:16:22,960
Speaker 2: So so back to MCP, it's it sounds like it's

342
00:16:23,200 --> 00:16:26,879
very agent focused. Is that an accurate statement?

343
00:16:27,639 --> 00:16:27,919
Speaker 1: Yeah?

344
00:16:27,960 --> 00:16:30,360
Speaker 4: So, so people still ask us this a lot, you know,

345
00:16:30,399 --> 00:16:32,200
when when do we want to use a traditional API

346
00:16:32,279 --> 00:16:34,519
versus MCP? And I think that the way to think

347
00:16:34,519 --> 00:16:37,080
about it is a lot of what we've built is

348
00:16:37,120 --> 00:16:38,720
still the best possible.

349
00:16:38,399 --> 00:16:39,120
Speaker 1: UI or something.

350
00:16:39,200 --> 00:16:41,240
Speaker 4: So if I want to cancel an order is it

351
00:16:41,279 --> 00:16:43,200
easier for me to go to Amazon, go to my

352
00:16:43,320 --> 00:16:45,360
orders and click cancel, or to open up a chat

353
00:16:45,399 --> 00:16:47,759
butt and say hi, the order for this, and it's like, okay,

354
00:16:47,759 --> 00:16:49,679
are you Are you referring to disorder and I'm like yes,

355
00:16:49,720 --> 00:16:52,240
And it's it's just not the best the interface to cancel,

356
00:16:52,279 --> 00:16:55,039
and so for that there's a button, and to have

357
00:16:55,120 --> 00:16:58,080
a button then prompt an agent and say hey, the

358
00:16:58,080 --> 00:17:00,360
customer would like to cancel disorder, and it then has

359
00:17:00,399 --> 00:17:02,200
to decide on a tool, and then it has to

360
00:17:02,200 --> 00:17:03,879
make sure it's calling the right tool and make that call.

361
00:17:04,200 --> 00:17:07,480
One it's just inefficient and really slow. But two, you're

362
00:17:07,519 --> 00:17:09,960
going from a world of a very deterministic action I

363
00:17:10,000 --> 00:17:13,160
click a button that runs this code that cancels this

364
00:17:13,359 --> 00:17:16,319
order to we think that the agent should be able

365
00:17:16,319 --> 00:17:18,720
to figure out the right tool to call, but it

366
00:17:18,799 --> 00:17:20,160
might not get it right every time.

367
00:17:20,599 --> 00:17:23,400
Speaker 1: And again we're spending a lot of money on having

368
00:17:23,440 --> 00:17:24,920
the agent decide what to actually do.

369
00:17:25,000 --> 00:17:28,039
Speaker 4: And so I think in a lot of cases, classic

370
00:17:28,079 --> 00:17:30,799
APIs are still going to be the really valuable one.

371
00:17:31,559 --> 00:17:36,079
MCP is for agentic interfaces, it's for bots, it's for communications.

372
00:17:36,599 --> 00:17:38,920
It could be a customer facing bot it could be

373
00:17:39,000 --> 00:17:41,640
an internal sorry by bottom referring to it could be

374
00:17:41,680 --> 00:17:44,480
a customer facing agent, it could be an internal agent,

375
00:17:45,559 --> 00:17:48,519
or it could be some form of non exposed agent

376
00:17:48,599 --> 00:17:50,160
that's actually just taking actions.

377
00:17:50,640 --> 00:17:53,240
Speaker 3: I really like this take. I think it's really interesting

378
00:17:53,279 --> 00:17:56,359
and I want to repeat this. Basically, if the thing

379
00:17:56,400 --> 00:17:58,400
that could bring you value in your business or your

380
00:17:58,480 --> 00:18:03,640
API is allowing increased volume or speed for execution, this

381
00:18:03,680 --> 00:18:05,599
could be the right thing to do. But if you

382
00:18:05,599 --> 00:18:07,640
don't want people to cancel your orders, don't make it

383
00:18:07,680 --> 00:18:08,799
easier for them to do that.

384
00:18:09,160 --> 00:18:13,119
Speaker 4: So that's not where it was going.

385
00:18:13,160 --> 00:18:16,000
Speaker 3: But that's why that's what I no, but.

386
00:18:15,960 --> 00:18:16,559
Speaker 1: I think I will.

387
00:18:17,240 --> 00:18:18,920
Speaker 3: It's interesting that you chose Amazon though, Like you know,

388
00:18:18,920 --> 00:18:20,400
if you pick something and it has a really great

389
00:18:20,480 --> 00:18:24,119
user experience, then it doesn't necessarily make sense to automate that.

390
00:18:24,400 --> 00:18:26,960
But as we know out there, there's like every company

391
00:18:27,000 --> 00:18:30,359
sucks at UX. So you know, you think about that,

392
00:18:30,400 --> 00:18:32,000
do you want to invest in you you actually do

393
00:18:32,079 --> 00:18:35,160
want to just throw money at the problem and run

394
00:18:35,200 --> 00:18:39,319
a really expensive service somewhere to manage state for letting

395
00:18:39,359 --> 00:18:41,680
agents manage the user experience.

396
00:18:42,440 --> 00:18:43,160
Speaker 1: Yeah, yeah, I.

397
00:18:43,200 --> 00:18:45,759
Speaker 4: Mean ultimately they're going to that specific example they're going

398
00:18:45,839 --> 00:18:48,640
to do both anyway, But in general, the idea is

399
00:18:48,680 --> 00:18:52,200
like a static action that's button driven, classic API integration

400
00:18:52,480 --> 00:18:53,920
AGENTIC MCP.

401
00:18:54,440 --> 00:18:58,519
Speaker 2: Whenever you're building out an MCP, how do you debug that?

402
00:18:58,720 --> 00:19:01,160
Like in this example, you know where you're talking about.

403
00:19:01,519 --> 00:19:03,319
It's trying to figure out the best course of action

404
00:19:03,440 --> 00:19:05,319
to do what it thinks the user wanted to do.

405
00:19:06,359 --> 00:19:09,119
And then sometimes it's wrong, like what kind of feedback

406
00:19:09,119 --> 00:19:11,480
do you get or what kind of metrics are you

407
00:19:11,559 --> 00:19:13,759
collecting to track that and improve on that?

408
00:19:14,359 --> 00:19:17,279
Speaker 3: It's just a proxy, right or is there something else

409
00:19:17,319 --> 00:19:18,400
magical happening in there?

410
00:19:19,359 --> 00:19:22,079
Speaker 4: Yeah, So how MCP works is effectively you create an

411
00:19:22,200 --> 00:19:25,279
MCP server and that has a function on it called

412
00:19:25,319 --> 00:19:28,079
get tools, and it returns back to the agent that's

413
00:19:28,079 --> 00:19:30,839
calling it, or the MCP client. But effectively the agent

414
00:19:30,920 --> 00:19:33,480
it says, hey, here are the tools available to you.

415
00:19:33,559 --> 00:19:37,440
Create a ticket in asana, modify a ticket, change the

416
00:19:37,480 --> 00:19:40,680
status of a ticket, and effectively, on the MCP server,

417
00:19:40,880 --> 00:19:44,319
you're right, those are just wrapping API calls, and so

418
00:19:44,359 --> 00:19:48,359
when the agent calls get tools, it then actually uses

419
00:19:48,400 --> 00:19:51,440
its LM abilities to decide which tools should be called

420
00:19:51,720 --> 00:19:55,200
based on the human The English description in that MCP

421
00:19:55,359 --> 00:19:58,839
server of what each tool does. So yes, it is

422
00:19:58,880 --> 00:20:01,920
statically coded. The place where there is sort of that

423
00:20:01,920 --> 00:20:04,559
that decision making is on the agent itself about how

424
00:20:04,559 --> 00:20:06,519
it's going to chain all those tool calls together.

425
00:20:07,359 --> 00:20:10,599
Speaker 2: Gotcha, and are you bringing your own LLLM into the

426
00:20:10,680 --> 00:20:14,079
agent or using like publicly available ones.

427
00:20:15,160 --> 00:20:17,480
Speaker 4: So this is totally up to whoever you know. It's

428
00:20:17,519 --> 00:20:20,160
compatible with all agents. The idea is is you know

429
00:20:20,400 --> 00:20:23,079
your agent can make a tool call its most agents,

430
00:20:23,440 --> 00:20:26,599
your agent can make a tool get tools call, and

431
00:20:26,640 --> 00:20:29,440
then it doesn't matter what model it is. It then

432
00:20:29,519 --> 00:20:31,319
determines what tools to call.

433
00:20:31,880 --> 00:20:32,519
Speaker 1: And you see this.

434
00:20:32,519 --> 00:20:36,319
Speaker 4: Also with things like even even windsurf and cursor for coding.

435
00:20:37,039 --> 00:20:38,720
You can see that when you ask it to do something,

436
00:20:38,720 --> 00:20:41,640
it kind of formulates a workflow using the tools available

437
00:20:41,640 --> 00:20:45,240
to it, like edit file, read contents of file, that.

438
00:20:45,240 --> 00:20:45,680
Speaker 1: Sort of thing.

439
00:20:46,119 --> 00:20:48,359
Speaker 3: I think, I think I finally figured it out. It's

440
00:20:48,440 --> 00:20:52,359
that there is just too much functionality in some of

441
00:20:52,400 --> 00:20:55,519
these products and services. And someone thinks that their opinion

442
00:20:56,000 --> 00:20:59,000
opinion version of the world is the most optimal and

443
00:20:59,000 --> 00:21:01,519
effective one, and they've coded that in an MTP server.

444
00:21:01,799 --> 00:21:03,880
So they look at GitHub, they look at GitHub and say,

445
00:21:03,920 --> 00:21:05,559
you know, I only do five actions. I only like

446
00:21:05,640 --> 00:21:08,960
copy code, make commits, push up pull requests. You know,

447
00:21:09,119 --> 00:21:11,000
then you know, once pull requests there as I approve

448
00:21:11,000 --> 00:21:13,519
it and maybe deploy it. And if that's my whole

449
00:21:13,640 --> 00:21:16,480
version of the world, then having the whole access to

450
00:21:16,640 --> 00:21:19,599
all of githubs or git labs or whatever's API is

451
00:21:19,599 --> 00:21:22,640
totally unnecessary. You don't need all those things. So let

452
00:21:22,640 --> 00:21:26,400
me make a personalized MTP server that just understands how

453
00:21:26,440 --> 00:21:28,920
to integrate in this one way, and then I'll expose

454
00:21:28,960 --> 00:21:30,599
that for interacting with the other tools.

455
00:21:30,319 --> 00:21:32,480
Speaker 1: That I have. Oh yeah, yeah, I think that's right.

456
00:21:32,480 --> 00:21:34,079
Speaker 4: And you can also you can extend it, and so

457
00:21:34,160 --> 00:21:36,319
a lot of times people will you know, notice, hey,

458
00:21:36,359 --> 00:21:39,119
whenever it's really important to me to build an agent

459
00:21:39,160 --> 00:21:42,599
where someone can ask it, you know what, what is

460
00:21:43,519 --> 00:21:46,039
what pull requests had the most reviews on it? And

461
00:21:46,119 --> 00:21:49,160
GitHub doesn't necessarily have a way to do that via

462
00:21:49,160 --> 00:21:51,319
their API. And so you could write a tool that's

463
00:21:51,359 --> 00:21:54,599
called you know, get repo with most stars, and that

464
00:21:54,720 --> 00:21:57,000
might make many API requests and do all that and

465
00:21:57,000 --> 00:21:59,920
then just return the repo at the end. So it's

466
00:22:00,119 --> 00:22:02,880
while it is stringing together the API calls. You're effectively

467
00:22:02,880 --> 00:22:05,160
building a tool chain within each tool as well.

468
00:22:06,400 --> 00:22:07,359
Speaker 3: Micro services.

469
00:22:08,359 --> 00:22:11,279
Speaker 2: Yeah, yeah, you can say that I don't know, but

470
00:22:12,759 --> 00:22:16,599
it's micros macro services because we've gotta put AI in

471
00:22:16,640 --> 00:22:18,559
the middle of the word now that it's a new

472
00:22:18,559 --> 00:22:27,200
tool Yeah. So you mentioned that these are not stateless,

473
00:22:27,240 --> 00:22:30,359
So what does the what's the infrastructures look like?

474
00:22:31,039 --> 00:22:32,680
Speaker 1: So it depends on how you end up building it.

475
00:22:32,720 --> 00:22:34,319
Speaker 4: I mean, I mean some people will back it with

476
00:22:34,359 --> 00:22:36,599
just you know, cash, you'll use in memory, you can,

477
00:22:36,640 --> 00:22:39,279
you can connect it to a database. You run into

478
00:22:39,599 --> 00:22:43,200
some risks though, right of of you know, sort of cross.

479
00:22:43,240 --> 00:22:44,920
Speaker 1: Cross user data sharing.

480
00:22:45,720 --> 00:22:48,359
Speaker 4: There's a whole new a whole new suite of potential

481
00:22:48,359 --> 00:22:51,880
security issues with MCP now that that come up, and

482
00:22:51,920 --> 00:22:54,119
it's something that we are we're hard at work at,

483
00:22:54,200 --> 00:22:56,359
not just you know, for us, but for customers as well.

484
00:22:56,359 --> 00:22:58,799
As we built that our new product, but we do

485
00:22:58,920 --> 00:23:00,920
have to also think about it for ourselves in everything

486
00:23:00,920 --> 00:23:01,359
we build.

487
00:23:01,359 --> 00:23:01,640
Speaker 1: Now.

488
00:23:02,079 --> 00:23:04,480
Speaker 3: The thing that really comes to mind a lot here

489
00:23:04,680 --> 00:23:07,359
is think about what, given the state of the world,

490
00:23:07,440 --> 00:23:10,200
what is the optimal way to allow agents to interact

491
00:23:10,279 --> 00:23:12,119
with things? And if you think about where the costs

492
00:23:12,119 --> 00:23:14,880
are it's having the agent do anything, So minimize the

493
00:23:14,960 --> 00:23:18,440
agent workload as much as possible, So limiting the input tokens.

494
00:23:18,480 --> 00:23:20,400
That means that don't force it to read the whole

495
00:23:20,440 --> 00:23:22,799
knowledge base and the whole API spec. You want to

496
00:23:22,839 --> 00:23:26,920
have it collapse into something very opinionated, very specific, with

497
00:23:27,039 --> 00:23:29,519
only the keywords that are necessary, like don't even use

498
00:23:29,559 --> 00:23:32,519
human readable descriptions there, right, like you know, focus just

499
00:23:32,559 --> 00:23:34,839
on those most important keywords. And same goes with the

500
00:23:34,880 --> 00:23:38,680
output stuff. So every single additional output token is also

501
00:23:38,680 --> 00:23:40,559
going to charge you. So you want to capture as

502
00:23:40,640 --> 00:23:42,480
much of the value of the action that you want

503
00:23:42,519 --> 00:23:45,359
to take in this intermediary MCP server.

504
00:23:46,640 --> 00:23:47,920
Speaker 1: Yeah, yeah, I think that's right.

505
00:23:48,119 --> 00:23:50,119
Speaker 4: It's both a security thing as well as it adds

506
00:23:50,160 --> 00:23:51,880
capabilities and makes things more efficient.

507
00:23:52,480 --> 00:23:52,680
Speaker 1: Yeah.

508
00:23:52,880 --> 00:23:55,440
Speaker 3: So yeah, the security thing, I think that's going to

509
00:23:55,519 --> 00:23:57,599
keep coming up for the till the dawn of time.

510
00:23:57,640 --> 00:24:00,400
I mean, because especially with the automation, right, we're using

511
00:24:00,440 --> 00:24:05,079
these LM tools to assist us in making additional services,

512
00:24:05,119 --> 00:24:09,160
additional value provided for our products and businesses and skipping

513
00:24:09,160 --> 00:24:11,599
the part where we're heavily interrogating what comes out of it.

514
00:24:11,960 --> 00:24:13,640
And this is an area where we're going to have

515
00:24:13,720 --> 00:24:16,759
a lot of extra usages going through here. Whereas the

516
00:24:16,839 --> 00:24:20,799
underlying API store is has you know, years maybe decades

517
00:24:20,880 --> 00:24:24,000
built up into how to make this API secure. Now

518
00:24:24,000 --> 00:24:27,960
we're cash data effectively not allowed cross contamination, et cetera.

519
00:24:28,079 --> 00:24:29,839
And now we're adding a layer on top that someone

520
00:24:29,960 --> 00:24:32,960
just throwing things as fast as possible to expose the

521
00:24:33,039 --> 00:24:37,039
data in a way that allows other agents to interact

522
00:24:37,079 --> 00:24:38,799
with it in the way they see as most optimal.

523
00:24:39,559 --> 00:24:41,720
Speaker 1: Yeah, and there's there's a lot of new risks with it.

524
00:24:42,039 --> 00:24:45,240
Speaker 4: And you know, one good example here is is API

525
00:24:45,359 --> 00:24:47,920
token passing in this idea of like, okay, well if

526
00:24:47,920 --> 00:24:50,200
the key lives on the MCP server, maybe that can

527
00:24:50,240 --> 00:24:50,960
never be exposed.

528
00:24:50,960 --> 00:24:52,960
Speaker 1: So you go, you ask the agent like what is

529
00:24:53,000 --> 00:24:53,880
the API token?

530
00:24:53,920 --> 00:24:55,559
Speaker 4: It's going to say, I don't have access to that,

531
00:24:55,680 --> 00:24:57,440
or I've been I've been programmed not to tell you.

532
00:24:59,240 --> 00:25:01,799
But let's say that this is an integration where it's

533
00:25:01,799 --> 00:25:04,400
connecting to say Salesforce, where everyone has a different sub domain,

534
00:25:04,440 --> 00:25:06,559
and part of your setup is what is your sub domain?

535
00:25:06,759 --> 00:25:08,039
And you find a way to just pass in an

536
00:25:08,160 --> 00:25:11,640
entirely different URL and basically overad what it's doing and

537
00:25:11,680 --> 00:25:13,640
have it make an API call and that URL as

538
00:25:13,680 --> 00:25:15,680
a private server that you're hosting, and you're getting that

539
00:25:15,720 --> 00:25:16,960
apike sent right to you.

540
00:25:17,480 --> 00:25:19,240
Speaker 1: So we're seeing a big new class.

541
00:25:19,279 --> 00:25:21,799
Speaker 4: That's just one example, and there's going to be a

542
00:25:21,839 --> 00:25:23,279
whole new set of rules, a whole new set of

543
00:25:23,319 --> 00:25:25,680
like linting that's going to have to catch this, but

544
00:25:25,759 --> 00:25:30,799
also intermediate security services for MCP specifically, and yeah, again,

545
00:25:30,839 --> 00:25:32,079
that's that's somewhere we're going.

546
00:25:33,640 --> 00:25:36,359
Speaker 2: Yeah, so there's a whole layer of impersonation in here

547
00:25:36,440 --> 00:25:41,240
because you've got the MCP service that responds to requests

548
00:25:41,279 --> 00:25:46,920
from multiple people, but each one of those individuals probably

549
00:25:47,000 --> 00:25:50,200
has different permissions, and so the MCP service has to

550
00:25:50,920 --> 00:25:55,200
impersonate the request based on what credentials that person has,

551
00:25:55,240 --> 00:25:59,039
like to use your example earlier of canceling ramp cards

552
00:25:59,039 --> 00:26:02,480
for employees. It has to determine, you know, does this

553
00:26:02,519 --> 00:26:06,400
person have the ability to cancel cards at all, and

554
00:26:06,480 --> 00:26:07,720
if so, which cards.

555
00:26:08,720 --> 00:26:11,559
Speaker 3: Yeah, I mean, I think realistically you're talking a little

556
00:26:11,559 --> 00:26:14,400
bit about the confused deputy problem here, where you're passing

557
00:26:14,440 --> 00:26:16,960
off the request to a privileged agent which is going

558
00:26:17,039 --> 00:26:20,160
to have requests from multiple different users and customers and

559
00:26:20,279 --> 00:26:23,680
it needs to do effective authentication and authorization. You can't

560
00:26:23,720 --> 00:26:26,759
just pass all the data along and expect the underlying

561
00:26:26,759 --> 00:26:29,000
API to do the right thing, and the individual agent

562
00:26:29,079 --> 00:26:31,759
might have its own identity or the actions that it's taking,

563
00:26:31,839 --> 00:26:34,680
And like, we've already solved these problems in the world.

564
00:26:34,920 --> 00:26:36,960
It's just that I think fundamentally, the people that are

565
00:26:37,000 --> 00:26:41,240
most closely working on the software and the control inside

566
00:26:41,319 --> 00:26:45,799
MCP services haven't thought about these things as deeply as

567
00:26:45,839 --> 00:26:49,039
they are. Like, my whole domain is just app seck,

568
00:26:49,240 --> 00:26:52,319
Like I only think about authentication and authorization all day long.

569
00:26:52,440 --> 00:26:55,680
That's pretty much my only job. And there's still way

570
00:26:55,680 --> 00:26:57,799
more in security than that, and I can get And

571
00:26:57,839 --> 00:27:00,160
I'm giving a talk literally in a couple of weeks

572
00:27:00,319 --> 00:27:02,839
like what the heck is off, like just explaining to

573
00:27:02,880 --> 00:27:05,559
people what these concepts are, and I'm getting a lot

574
00:27:05,559 --> 00:27:07,839
of feedback like, oh, mcps a thing, like how do

575
00:27:07,880 --> 00:27:10,160
we do this? It's even more important now, And I'm

576
00:27:10,160 --> 00:27:11,200
like that's really surprising.

577
00:27:11,720 --> 00:27:15,599
Speaker 4: Yeah, it begs the question it does does OFF even

578
00:27:15,640 --> 00:27:17,799
belong in the MCP s back itself or is everyone

579
00:27:17,839 --> 00:27:19,279
going to keep just building it on their own because

580
00:27:19,359 --> 00:27:21,440
right now there's not much in there about it, how

581
00:27:21,440 --> 00:27:24,240
to actually implement it, how to actually handle it other

582
00:27:24,359 --> 00:27:27,480
than saying, you know, you probably should should keep off

583
00:27:27,559 --> 00:27:28,960
on the MCP server itself.

584
00:27:29,240 --> 00:27:31,079
Speaker 3: I think, yeah, I mean it's a really good point.

585
00:27:31,160 --> 00:27:33,720
I think the biggest problem is that you're not just

586
00:27:33,799 --> 00:27:38,039
building this proxy layer that statelessly passes along request. And

587
00:27:38,039 --> 00:27:40,359
this has been a huge problem with proxies to begin with,

588
00:27:40,480 --> 00:27:43,079
and now people are taking it further and saying, oh,

589
00:27:43,200 --> 00:27:45,119
we know what we want to do, We've already figured

590
00:27:45,119 --> 00:27:47,519
it out. We don't want to explode the number of

591
00:27:47,519 --> 00:27:50,319
input tokens or pass the context back to the caller,

592
00:27:50,400 --> 00:27:53,559
so they have to iteraly call the endpoints. We want

593
00:27:53,599 --> 00:27:55,680
to handle all that in scope here, which means, as

594
00:27:55,720 --> 00:27:59,000
you pointed out, managing state, and that becomes the risk

595
00:27:59,079 --> 00:28:01,359
really doing that effect actively and correctly so you don't

596
00:28:01,440 --> 00:28:03,079
end up with this crossing. And there are tons of

597
00:28:03,480 --> 00:28:06,880
security vulnerabilities cvees that get published every single month about

598
00:28:06,920 --> 00:28:09,920
how one customer was able to access a different customer's

599
00:28:10,000 --> 00:28:12,799
data publicly on the Internet, and like, those things all

600
00:28:12,839 --> 00:28:17,200
happen without MVP servers, and now people aren't even thinking about, oh,

601
00:28:17,359 --> 00:28:20,480
how do we actually even handle off here? So yeah,

602
00:28:20,720 --> 00:28:22,720
We're definitely going to get into a problem very quickly.

603
00:28:23,359 --> 00:28:24,880
Speaker 4: Yeah, and the problem, you know, one of the things

604
00:28:24,920 --> 00:28:26,640
we're seeing as we build out this new product is

605
00:28:26,880 --> 00:28:31,519
specific requirements from people around credential sharing among people who

606
00:28:31,559 --> 00:28:35,319
talk to the agents, but varying who shares which credentials

607
00:28:35,319 --> 00:28:36,200
for different services.

608
00:28:36,240 --> 00:28:37,160
Speaker 1: So, you know, you have an.

609
00:28:37,079 --> 00:28:40,039
Speaker 4: Agent that has access to both a ticketing system and salesforce,

610
00:28:40,319 --> 00:28:42,160
and everyone who talks to it should be able to

611
00:28:42,200 --> 00:28:44,720
use the shared ticketing credentials, but only the sales team

612
00:28:44,759 --> 00:28:47,440
should be able to actually communicate with salesforce, and then

613
00:28:47,599 --> 00:28:50,519
only super admins should be able to use the HR

614
00:28:50,960 --> 00:28:53,359
you know, credentials. So we're seeing this a lot, and

615
00:28:53,359 --> 00:28:56,160
it's going to be interesting to see how people approach it.

616
00:28:55,720 --> 00:28:58,480
Speaker 3: That's like our whole product right there, Like that little

617
00:28:58,559 --> 00:29:00,720
aspect you described is like the whole thing that we do,

618
00:29:01,279 --> 00:29:04,960
uh for internal resources because I mean, it's really unfortunate

619
00:29:05,000 --> 00:29:08,920
that we live in the world of credential sharing, but yeah,

620
00:29:09,039 --> 00:29:13,000
it still happens because companies charge for uh, you know,

621
00:29:13,160 --> 00:29:15,000
SSO and first class options.

622
00:29:15,440 --> 00:29:15,640
Speaker 1: Yeah.

623
00:29:16,400 --> 00:29:21,160
Speaker 2: So, Warren, are are you at offerers creating an off

624
00:29:21,200 --> 00:29:22,960
front end for MCP services.

625
00:29:23,880 --> 00:29:25,759
Speaker 3: I don't even know how to answer that question. You know,

626
00:29:25,839 --> 00:29:28,279
it's like, what's the difference between off for MCP and not?

627
00:29:28,400 --> 00:29:30,839
I mean, there's literally no difference. Like whatever we offer

628
00:29:31,039 --> 00:29:34,400
is like it's the same. There really is no difference

629
00:29:34,440 --> 00:29:36,599
between what we offer. Like I was joking to my

630
00:29:36,759 --> 00:29:40,039
CEO who almost said that she was going to have

631
00:29:40,039 --> 00:29:41,599
a mental breakdown when I told her that we could

632
00:29:41,640 --> 00:29:46,079
start offering off for AI because realistically, there is no difference,

633
00:29:46,160 --> 00:29:49,200
like we offer other off like the it doesn't matter

634
00:29:49,279 --> 00:29:52,759
difference marketing marketing.

635
00:29:56,400 --> 00:29:58,839
Speaker 4: Your one little rapper file that makes it interact nicely

636
00:29:58,880 --> 00:29:59,920
with a lot of the AI library.

637
00:30:00,079 --> 00:30:01,279
Speaker 1: Maybe that it's.

638
00:30:01,799 --> 00:30:03,759
Speaker 3: Actually not even that, it's literally just what shows up

639
00:30:03,759 --> 00:30:06,160
on the marketing page like that. That's really what's important,

640
00:30:06,200 --> 00:30:08,440
because you know, you can put whatever you want on

641
00:30:08,480 --> 00:30:10,599
the in the knowledge based on the documentation, and people

642
00:30:10,599 --> 00:30:13,079
will pick up and run with it, you know, focusing

643
00:30:13,160 --> 00:30:15,559
on the problem and the vocabulary that they're using to

644
00:30:15,559 --> 00:30:18,000
solve it, right, you know, if it's not your company,

645
00:30:18,039 --> 00:30:19,559
but like I say, one of your customers are like, oh,

646
00:30:19,720 --> 00:30:22,599
how do we keep the credential ownership separate? Or while

647
00:30:22,680 --> 00:30:24,759
using some sort of agent. You know, now we have

648
00:30:24,799 --> 00:30:26,839
to match on all those terms to hit SEO so

649
00:30:26,920 --> 00:30:29,359
that shows up in search results in one of the

650
00:30:29,400 --> 00:30:32,519
search providers or through one of the lamps six months

651
00:30:32,519 --> 00:30:34,359
from now. So it has to match on that and

652
00:30:34,359 --> 00:30:36,880
then also produce you know, relevant code with variable names

653
00:30:36,880 --> 00:30:40,400
that look like it's appropriate. So yeah, I mean, we

654
00:30:40,400 --> 00:30:43,039
don't have any any plans to throw up an MCB server,

655
00:30:43,079 --> 00:30:44,680
but if one of our customers was like, you know,

656
00:30:45,279 --> 00:30:46,960
we have this use case and we're willing to pay

657
00:30:47,000 --> 00:30:50,559
you some money for it, yeah, I mean done.

658
00:30:54,880 --> 00:30:59,359
Speaker 2: Sort of the scaling issues you see with MCP services.

659
00:30:59,720 --> 00:31:02,480
Speaker 1: Yeah, so we've seen a few. I think I think.

660
00:31:02,319 --> 00:31:05,079
Speaker 4: One is is like managing I hate to say this again,

661
00:31:05,119 --> 00:31:06,920
but managing off and I think I think it's because

662
00:31:07,000 --> 00:31:09,799
companies just aren't really thinking about that at scale, how

663
00:31:10,079 --> 00:31:12,240
like groups in differentiation of who has access.

664
00:31:12,240 --> 00:31:12,960
Speaker 1: So that's one of them.

665
00:31:13,680 --> 00:31:15,640
Speaker 4: I think another one is that people are using AI

666
00:31:15,839 --> 00:31:18,279
to generate a lot of MCP servers right now, and

667
00:31:18,319 --> 00:31:21,200
I think that at scale it brings on again more

668
00:31:21,240 --> 00:31:24,119
security risks, but also scaling risks, right you're not necessarily

669
00:31:24,680 --> 00:31:27,559
hitting the end points very efficiently. We actually we see

670
00:31:27,599 --> 00:31:30,839
another big problem which is MCP servers and MCP in

671
00:31:30,839 --> 00:31:33,319
general love to do linear scans of APIs.

672
00:31:33,400 --> 00:31:35,279
Speaker 1: So if you have an MCP server that.

673
00:31:35,640 --> 00:31:38,079
Speaker 4: Lets you, you know, fetch tickets and then you can

674
00:31:38,079 --> 00:31:41,319
pass a page number and you say to your LLM like,

675
00:31:41,440 --> 00:31:44,240
get me all tickets that have Gill in the title,

676
00:31:44,640 --> 00:31:46,559
it is going to do a linear scan of that API,

677
00:31:46,839 --> 00:31:49,160
and that will will crash your MCP server. It's going

678
00:31:49,200 --> 00:31:51,000
to kill the agent. It's going to cost you a ton.

679
00:31:51,759 --> 00:31:53,119
So that's another one we've seen.

680
00:31:55,200 --> 00:31:58,079
Speaker 3: I mean, because like, as an experienced engineer, it's like, oh,

681
00:31:58,119 --> 00:32:00,640
this is I'm actually paging through this for like ten minutes.

682
00:32:00,799 --> 00:32:02,960
What is going on is it's stuck, you know, something

683
00:32:03,000 --> 00:32:04,960
else going on. And you may look at your code

684
00:32:05,000 --> 00:32:06,839
and be like, oh, there's actually a lot of items here,

685
00:32:06,960 --> 00:32:08,759
or each one of these calls is like really expensive,

686
00:32:08,799 --> 00:32:11,039
or I'm getting throttled, and you'd go and investigate that.

687
00:32:11,119 --> 00:32:13,079
But if you're not paying attention and you don't even

688
00:32:13,119 --> 00:32:15,440
really understand what's going on with that MTP server or

689
00:32:15,559 --> 00:32:18,759
any product, and you get something automatically generated, it's not

690
00:32:18,799 --> 00:32:21,880
paying attention to that. It doesn't care. It doesn't care

691
00:32:21,880 --> 00:32:22,680
along it runs right.

692
00:32:23,440 --> 00:32:26,000
Speaker 4: That's a good point, and it knows how to exponentially

693
00:32:26,039 --> 00:32:28,119
back off, so it will take its time and really

694
00:32:28,200 --> 00:32:28,960
pull everything.

695
00:32:31,279 --> 00:32:34,400
Speaker 2: Yeah, I can see a scenario where this is like

696
00:32:34,480 --> 00:32:36,640
after rollout and people get used to it and it's

697
00:32:36,640 --> 00:32:41,000
fully adopted. There's like a layer of knowledge abstraction where

698
00:32:42,200 --> 00:32:47,200
someone troubleshooting it or interacting with it doesn't even know

699
00:32:47,599 --> 00:32:51,640
to go look at the API end points that it's

700
00:32:51,680 --> 00:32:53,759
calling to find out that you've been getting hit with

701
00:32:53,839 --> 00:32:57,799
overage charges for the last twelve hours because the service

702
00:32:57,880 --> 00:32:59,240
is just banging away on it.

703
00:33:00,000 --> 00:33:02,640
Speaker 3: I think there's another another huge complexity there, like some

704
00:33:02,680 --> 00:33:04,799
of those routes may be paid and other ones may

705
00:33:04,839 --> 00:33:07,440
be free, or like instead of using the search endpoint,

706
00:33:07,519 --> 00:33:11,119
it's using some sort of native pat genation or you know,

707
00:33:11,279 --> 00:33:13,319
if you if you if the MCP server isn't built

708
00:33:13,319 --> 00:33:16,200
by the company who's running the API, they know what

709
00:33:16,240 --> 00:33:19,559
their internal complexity is and should be focusing on that,

710
00:33:19,599 --> 00:33:22,359
whereas someone else may not understand or really don't even care.

711
00:33:22,680 --> 00:33:24,960
You know where the limitations are for using that server,

712
00:33:25,240 --> 00:33:27,119
and like, if you come and talk to us, we'll say, oh, yeah,

713
00:33:27,119 --> 00:33:29,319
you know what the thing you're doing. You want to

714
00:33:29,400 --> 00:33:32,000
use endpoints A and B. This will be the fastest,

715
00:33:32,039 --> 00:33:35,279
the cheapest option for you, most cash option, et cetera.

716
00:33:35,680 --> 00:33:38,039
And if you're just dynamically generating something, it may look

717
00:33:38,119 --> 00:33:41,039
like different endpoints could be appropriate. And if you're not

718
00:33:41,079 --> 00:33:43,519
paying attention, that's what'll get used and that could be

719
00:33:43,960 --> 00:33:47,079
the worst scenario for both the caller and also uh

720
00:33:47,480 --> 00:33:48,759
the service provider in the end.

721
00:33:49,640 --> 00:33:53,039
Speaker 4: Yeah, and ultimately that's why we pretty firmly believe that

722
00:33:53,039 --> 00:33:54,799
that you still need to think full sets of data

723
00:33:54,880 --> 00:33:56,720
for a lot of types of integration too. You know,

724
00:33:56,759 --> 00:33:59,200
if there's it's the lack of a search endpoint that

725
00:33:59,240 --> 00:34:02,079
causes that. And if you have no search endpoint and

726
00:34:02,200 --> 00:34:05,559
you you're building your business right, your product team doesn't

727
00:34:05,559 --> 00:34:07,279
want to hear no, we can't sink the full data

728
00:34:07,279 --> 00:34:08,760
is that they want to hear, like, here's how we're

729
00:34:08,760 --> 00:34:11,199
going to solve this problem. I mean, you have to

730
00:34:11,239 --> 00:34:13,239
do that, And so MCP becomes a little useless there.

731
00:34:14,280 --> 00:34:18,440
Speaker 2: So with with this being a stateful service, is that

732
00:34:18,800 --> 00:34:23,079
a concept that exists of saying, hey, go grab this

733
00:34:23,199 --> 00:34:26,519
API data and if you need it again for anything else,

734
00:34:26,639 --> 00:34:31,039
this this payload is valid for the next six hours

735
00:34:31,119 --> 00:34:33,719
or for the next two days or whatever, and then

736
00:34:33,760 --> 00:34:35,440
it just cashes that and reuses that.

737
00:34:36,119 --> 00:34:39,280
Speaker 4: Yeah, I mean, ultimately, your your tool implementation can do anything.

738
00:34:39,280 --> 00:34:41,079
And so you know, if you have a tool that's

739
00:34:41,159 --> 00:34:44,639
like you know, synk this data and then it cashes

740
00:34:44,679 --> 00:34:46,960
for six hours, it can under the hood be hitting

741
00:34:47,000 --> 00:34:50,000
some internal cash, whether that's like a memcash, REDDI or

742
00:34:50,159 --> 00:34:52,960
a database, and then it knows who it's who's talking

743
00:34:53,000 --> 00:34:54,360
to the server, so it knows who to look it

744
00:34:54,440 --> 00:34:54,760
up for.

745
00:34:55,239 --> 00:34:55,400
Speaker 1: Yeah.

746
00:34:55,440 --> 00:34:57,280
Speaker 3: I mean there's like a lot of different ways to

747
00:34:57,280 --> 00:34:59,559
handle cases, right. I think there's like two hard problems

748
00:34:59,559 --> 00:35:03,039
and computer science. So I mean you can, right, you

749
00:35:03,079 --> 00:35:06,000
can do read through cases, write through cashes. You know

750
00:35:06,000 --> 00:35:09,039
how you're managing it Basically, you're basically saying that the

751
00:35:09,079 --> 00:35:12,000
API as it's written isn't effective for the actions you

752
00:35:12,000 --> 00:35:14,079
want to take, and if one of those is like

753
00:35:14,800 --> 00:35:17,519
very bulk data related, you have no choice but to

754
00:35:17,599 --> 00:35:20,719
handle all of the interactions and then also get sinks

755
00:35:20,880 --> 00:35:23,199
back from the source, like constantly pulling it to get

756
00:35:23,199 --> 00:35:27,119
any updates there are. I imagine over time we're going

757
00:35:27,159 --> 00:35:31,599
to see primary providers of that data offer better strategies

758
00:35:31,599 --> 00:35:34,760
for interacting with them because third parties that spin up

759
00:35:34,760 --> 00:35:38,800
these MTP servers they're not benefiting anyone directly. I mean,

760
00:35:38,800 --> 00:35:42,039
it's the value is there clearly if the customers say, hey,

761
00:35:42,079 --> 00:35:44,280
you know, your API end points aren't giving us what

762
00:35:44,280 --> 00:35:47,480
we want and we need something better. So there's a

763
00:35:47,599 --> 00:35:50,800
drive for it. But I think over time, realizing what

764
00:35:50,840 --> 00:35:53,599
those traders are will have to be changed into whoever

765
00:35:53,639 --> 00:35:54,920
owns the data fundamentally.

766
00:35:55,880 --> 00:35:58,280
Speaker 4: Yeah, the question is whether you know, especially among the enterprise,

767
00:35:58,320 --> 00:36:01,159
you're going to see companies wanting to release the MCP servers.

768
00:36:01,199 --> 00:36:04,079
I think that you know, there's the world's always gone

769
00:36:04,079 --> 00:36:06,360
back and forth on whether API should be fully public,

770
00:36:06,360 --> 00:36:07,280
what data should.

771
00:36:07,000 --> 00:36:07,840
Speaker 1: Be exposed, and whatnot.

772
00:36:07,880 --> 00:36:09,480
Speaker 4: And you know, one of the things we're seeing among

773
00:36:09,559 --> 00:36:13,079
enterprise is fhear around MCP the idea both of unintended

774
00:36:13,119 --> 00:36:15,840
actions being taken, but also of data extraction.

775
00:36:16,400 --> 00:36:19,519
Speaker 3: Yeah. Well, I mean that's an interesting point because there's

776
00:36:19,599 --> 00:36:22,079
a I live in La la land, and here's my

777
00:36:22,159 --> 00:36:24,199
perspective of how the reality works, and this is how

778
00:36:24,239 --> 00:36:27,400
we make money and the mature, grown up approach of

779
00:36:27,599 --> 00:36:30,360
our users are actually doing this thing, and we should

780
00:36:30,360 --> 00:36:32,360
figure out how to encourage them to do the right

781
00:36:32,400 --> 00:36:34,400
thing so they don't accidentally do the wrong thing and

782
00:36:34,440 --> 00:36:36,559
the companies that realize that will be the ones that

783
00:36:36,639 --> 00:36:38,280
capture the value in the long term, and the ones

784
00:36:38,280 --> 00:36:40,519
that don't we'll just end up failing because people will

785
00:36:40,519 --> 00:36:42,719
stop using their services because they will no longer want

786
00:36:42,719 --> 00:36:45,559
to use the interface that's being provided to them.

787
00:36:45,960 --> 00:36:46,519
Speaker 1: Yeah, I think.

788
00:36:46,559 --> 00:36:48,840
Speaker 4: I think a notable example of that is Salesforce having

789
00:36:48,920 --> 00:36:52,000
a very open ecosystem and because of that, other platforms

790
00:36:52,000 --> 00:36:53,880
built on top of them. And now when you use

791
00:36:53,920 --> 00:36:56,840
Salesforce and you haven't integrated everywhere, you can't chur and

792
00:36:56,840 --> 00:36:59,320
that data is the core system of record for all

793
00:36:59,400 --> 00:37:03,000
the services your good market stack uses. But I'm not

794
00:37:03,039 --> 00:37:04,840
going to name names here, but there are a lot

795
00:37:04,840 --> 00:37:07,480
of enterprise players that are very locked down about their data,

796
00:37:07,760 --> 00:37:11,679
and actually mid market players as well, and somehow they've

797
00:37:11,760 --> 00:37:14,639
maintained a solid user base. I wouldn't say they are

798
00:37:14,639 --> 00:37:16,400
the fastest growing companies anymore, though.

799
00:37:16,920 --> 00:37:21,079
Speaker 3: I hope this causes a turnaround here because historically the

800
00:37:21,360 --> 00:37:23,840
products that have the most data, which end up being

801
00:37:23,840 --> 00:37:26,519
things like CRMs and data platforms, have had the worst

802
00:37:26,559 --> 00:37:29,760
APIs and the ones with the most Yeah, I'm sorry

803
00:37:29,800 --> 00:37:33,079
I had to come out and say that, but we're all.

804
00:37:32,920 --> 00:37:36,039
Speaker 2: Thinking it so yeah.

805
00:37:35,000 --> 00:37:37,280
Speaker 3: Yeah, sorry to all our previous guests from like the

806
00:37:37,360 --> 00:37:39,400
last four or five weeks to all of it on

807
00:37:39,440 --> 00:37:43,639
the data side, just from experience here. But if they're

808
00:37:43,679 --> 00:37:46,559
all the ones with the worst APIs and now we

809
00:37:46,679 --> 00:37:49,159
all care about that data as fundamentally as we can

810
00:37:49,280 --> 00:37:52,079
to get at it, and those APIs don't exist, the

811
00:37:52,119 --> 00:37:55,320
ones that survive will be the companies that actually invest

812
00:37:55,440 --> 00:37:57,440
in better APIs for accessing their data.

813
00:37:58,119 --> 00:37:59,920
Speaker 1: I agree, I completely agree.

814
00:38:00,400 --> 00:38:01,920
Speaker 3: So I just put it on the blockchain, I think

815
00:38:02,000 --> 00:38:02,760
is what wells thinking.

816
00:38:03,400 --> 00:38:05,400
Speaker 1: Absolutely very efficient.

817
00:38:06,000 --> 00:38:09,039
Speaker 3: Yep, well it's public, you know, problem solved.

818
00:38:09,880 --> 00:38:14,920
Speaker 2: Yeah, to share database, We're just sharing, just share damn it.

819
00:38:17,360 --> 00:38:21,159
So now this is scaled in complexity quite a bit.

820
00:38:21,639 --> 00:38:24,679
Just in the last thirty ish minutes we've been talking

821
00:38:24,679 --> 00:38:27,280
about it because we started saying, okay, you have this

822
00:38:27,360 --> 00:38:29,199
agent and it just makes it easier to talk to

823
00:38:30,039 --> 00:38:34,800
APIs for you. But now we're talking about you know,

824
00:38:35,000 --> 00:38:41,880
off issues and scaling and like security of the data.

825
00:38:42,960 --> 00:38:46,639
Like so for someone who's thinking, man, I saw this

826
00:38:46,840 --> 00:38:49,320
MCP term, I should go build one of these things,

827
00:38:49,360 --> 00:38:52,920
Like what's your recommendation for the top things they need

828
00:38:52,960 --> 00:38:53,880
to be thinking about.

829
00:38:54,519 --> 00:38:56,360
Speaker 4: I mean, I think number one right now. It's really

830
00:38:56,360 --> 00:39:00,639
good marketing. It's the topic and so and so releasing one.

831
00:39:00,679 --> 00:39:01,199
Speaker 1: It's out there.

832
00:39:01,239 --> 00:39:03,119
Speaker 4: I think, I think there are things to consider, like

833
00:39:03,920 --> 00:39:06,760
your MCP servers only as good as the underlying tools

834
00:39:06,880 --> 00:39:09,800
or you know, in other terms, the underlying API that

835
00:39:09,880 --> 00:39:12,400
it's talking to. I think you need to think through

836
00:39:12,440 --> 00:39:14,599
real world use cases and you need to test it.

837
00:39:15,000 --> 00:39:16,079
Speaker 1: And actually one of the things.

838
00:39:15,920 --> 00:39:18,440
Speaker 4: We talked about earlier was how to evaluate this, and

839
00:39:18,920 --> 00:39:20,599
I think it's you know, it's similar to other AI

840
00:39:20,679 --> 00:39:23,079
evaluation methods where where you're feeding it sort of like

841
00:39:23,119 --> 00:39:27,440
mock prompts and data and you're using evaluators to decide like,

842
00:39:27,559 --> 00:39:29,960
is the output eighty percent of the time close enough

843
00:39:29,960 --> 00:39:32,119
to what I needed to be that sort of thing,

844
00:39:32,159 --> 00:39:35,519
is the right tool getting called based on certain prompts?

845
00:39:36,719 --> 00:39:38,519
So I think I think making sure that you build

846
00:39:38,519 --> 00:39:40,840
good tools, making sure they're well documented, and making sure

847
00:39:40,840 --> 00:39:42,239
that you actually test them is important.

848
00:39:42,760 --> 00:39:45,079
Speaker 3: Yeah, it sounds like, you know, if I rephrase that,

849
00:39:45,119 --> 00:39:48,199
you've got to make sure that you understand what your

850
00:39:48,280 --> 00:39:50,519
users are actually asking for, you know, because there's like

851
00:39:50,559 --> 00:39:52,039
an infinite number of things that can be done with

852
00:39:52,079 --> 00:39:54,000
an API. So if you don't know what that is.

853
00:39:54,000 --> 00:39:56,440
You don't know what functionality to actually throw in your

854
00:39:56,760 --> 00:39:59,119
MCP server to begin with. You can't just like have

855
00:39:59,239 --> 00:40:01,599
it spun up and have just have it work. It

856
00:40:01,679 --> 00:40:03,199
needs to actually do something right.

857
00:40:04,199 --> 00:40:05,000
Speaker 1: That's exactly right.

858
00:40:05,039 --> 00:40:07,760
Speaker 4: And if you know your your customers are very likely

859
00:40:07,800 --> 00:40:10,639
to come in and say create an invoice. But whenever

860
00:40:10,679 --> 00:40:12,480
they do that, the way they word it is like

861
00:40:12,800 --> 00:40:15,639
create a I don't know, create a spend report. So whatever,

862
00:40:16,320 --> 00:40:18,079
you you know that you want to put that in

863
00:40:18,119 --> 00:40:20,320
the documentation for the MCP server when someone asks for

864
00:40:20,360 --> 00:40:22,480
creating a spend report, this is the tool to call.

865
00:40:22,800 --> 00:40:24,480
Speaker 3: I think you want to like add a line item

866
00:40:24,519 --> 00:40:27,599
to the accounts payable list.

867
00:40:27,679 --> 00:40:29,519
Speaker 1: You know, Yeah, there you go.

868
00:40:29,599 --> 00:40:32,599
Speaker 3: That's that's what's important. Getting that language right, okay, and

869
00:40:32,880 --> 00:40:33,880
words are hard.

870
00:40:34,400 --> 00:40:35,079
Speaker 1: Words are hard.

871
00:40:38,079 --> 00:40:39,800
Speaker 3: I mean you could you can in your MCP server,

872
00:40:39,920 --> 00:40:42,599
receive the request, pass it to another l M to

873
00:40:42,639 --> 00:40:45,320
get it translated into you know what makes sense, and

874
00:40:45,320 --> 00:40:48,079
then actually execute. But I highly don't recommend that.

875
00:40:48,760 --> 00:40:49,679
Speaker 1: Yeah, yeah, you know.

876
00:40:51,239 --> 00:40:55,079
Speaker 3: It's going to multiply your your security issues by quite

877
00:40:55,079 --> 00:40:55,679
a large amount.

878
00:40:56,960 --> 00:40:57,719
Speaker 1: That is so true.

879
00:40:57,800 --> 00:41:01,440
Speaker 4: Yeah, we we definitely believe in that separate we give

880
00:41:01,480 --> 00:41:02,000
you the tools.

881
00:41:02,199 --> 00:41:03,320
Speaker 1: You do what you want to do with it.

882
00:41:05,880 --> 00:41:08,039
Speaker 2: Here's the knife, cut off whatever you'd like. We're not

883
00:41:08,079 --> 00:41:09,800
responsible for the medical bill.

884
00:41:10,559 --> 00:41:11,559
Speaker 1: Yeah.

885
00:41:11,679 --> 00:41:16,519
Speaker 2: Have you seen scenarios of MCP services instead of talking

886
00:41:16,559 --> 00:41:19,840
to APIs or talking to other MCP services.

887
00:41:20,719 --> 00:41:21,719
Speaker 1: Oh that's interesting.

888
00:41:22,400 --> 00:41:25,639
Speaker 4: So I mean I guess realistically you could have an

889
00:41:25,719 --> 00:41:28,559
MCP service ask another service for tools, or call other

890
00:41:28,840 --> 00:41:34,119
call other tools. I'm not sure there's a great use case.

891
00:41:34,159 --> 00:41:36,519
Maybe I would have to think three use cases there.

892
00:41:36,559 --> 00:41:38,559
So I haven't seen any specific instances.

893
00:41:38,599 --> 00:41:38,719
Speaker 1: You know.

894
00:41:38,760 --> 00:41:41,519
Speaker 4: What I've really seen is is MCP servers. You know,

895
00:41:41,559 --> 00:41:45,360
they can return static data, they can query databases and

896
00:41:45,760 --> 00:41:49,119
like query databases directly or call APIs. So I definitely

897
00:41:49,159 --> 00:41:51,719
have seen where you have tools that are formulating SQL

898
00:41:51,760 --> 00:41:55,039
queries and so you could you know, like have an

899
00:41:55,079 --> 00:41:59,480
agent that you know, equeries a database using English language

900
00:41:59,519 --> 00:42:03,000
and that English language causes another you know tool, call

901
00:42:03,159 --> 00:42:06,039
to another MCP server that can translate to SQL for example.

902
00:42:06,199 --> 00:42:08,599
And I guess that's one one example, but I think

903
00:42:08,639 --> 00:42:11,599
you'll probably see more like a to a level communications

904
00:42:11,639 --> 00:42:14,880
as opposed to like agent to MCP to MCP.

905
00:42:15,440 --> 00:42:17,880
Speaker 3: Yeah, I mean, I guess there would have to be fundamentally.

906
00:42:18,039 --> 00:42:19,800
There could be a case, right, you call an API

907
00:42:19,880 --> 00:42:22,960
today and it does something with an l M, and

908
00:42:23,159 --> 00:42:24,719
in the future you call an API and that can

909
00:42:24,760 --> 00:42:27,079
call a different API and instead of doing that, it

910
00:42:27,119 --> 00:42:30,719
could use natural language there, although natural language is terrible.

911
00:42:30,960 --> 00:42:33,159
Like if you know how to if one one software

912
00:42:33,159 --> 00:42:34,960
developer rites in a code to call a different API,

913
00:42:35,400 --> 00:42:38,599
like use the first class API notions that are available there,

914
00:42:38,639 --> 00:42:40,119
you know you always want that code and that that

915
00:42:40,159 --> 00:42:43,079
would be software development there. And the other thing is

916
00:42:43,119 --> 00:42:45,159
that like the cost, like you want to pass that

917
00:42:45,199 --> 00:42:47,519
back to the caller as fast as possible and push

918
00:42:47,599 --> 00:42:50,199
them with the costs there. Otherwise you're running two agents

919
00:42:50,199 --> 00:42:52,800
at the same time, so you know whatever determinations need

920
00:42:52,840 --> 00:42:55,639
to be made. Plus there's also this like i'm gonna

921
00:42:55,679 --> 00:43:00,000
say security again, like delegation, like who owns that recquid.

922
00:43:00,440 --> 00:43:02,440
You probably don't want to build your service in a

923
00:43:02,480 --> 00:43:06,599
way which allows users to interact with a third party

924
00:43:06,599 --> 00:43:08,280
solution because that means they have to give you the

925
00:43:08,320 --> 00:43:11,119
credentials to do that, and then you're managing your customer's

926
00:43:11,159 --> 00:43:13,920
credentials to other third party systems. And if you're doing that,

927
00:43:13,960 --> 00:43:16,920
maybe you should like, talk to Gil and see if

928
00:43:17,119 --> 00:43:18,760
this is the use case that there, you know, because

929
00:43:18,760 --> 00:43:22,719
that really sounds like merge. Yeah, so you know, pass

930
00:43:22,760 --> 00:43:24,679
the data back to the caller and let them deal

931
00:43:24,719 --> 00:43:26,559
with the complexity and the cost of calling out to

932
00:43:26,599 --> 00:43:27,960
that second system.

933
00:43:28,559 --> 00:43:30,440
Speaker 4: I thought you were about to say, if you're doing that,

934
00:43:30,480 --> 00:43:32,320
you're wrong, and I was like, well, that's our whole business.

935
00:43:32,360 --> 00:43:32,599
Speaker 1: So I.

936
00:43:35,920 --> 00:43:39,400
Speaker 3: Try very hard not to say things in episode that

937
00:43:39,440 --> 00:43:40,960
contradict what the guest is saying.

938
00:43:41,760 --> 00:43:44,360
Speaker 1: So I'll keep an eye on your Twitter later.

939
00:43:45,960 --> 00:43:47,639
Speaker 3: Oh well, I deleted that a long time ago.

940
00:43:50,880 --> 00:43:51,960
Speaker 1: All right, I'm safe then.

941
00:43:52,440 --> 00:43:54,840
Speaker 2: Yeah. Yeah, we have an unwritten rule not to insult

942
00:43:54,880 --> 00:43:56,480
our guests till after the recording's over.

943
00:43:59,440 --> 00:44:04,360
Speaker 1: Man, all right, mutual to two Way Street them, I.

944
00:44:04,239 --> 00:44:05,760
Speaker 3: Like, I like you brought up Yeah. I mean, you

945
00:44:06,079 --> 00:44:08,320
have plenty of opportunities here, and I'm sure most of

946
00:44:08,320 --> 00:44:11,280
the audience are just waiting for those punches. Gil. So

947
00:44:11,360 --> 00:44:14,280
you know, if if myself or Will is there for you, you know,

948
00:44:14,360 --> 00:44:16,480
feel free to you know, come at us with with

949
00:44:16,559 --> 00:44:20,159
full force. No no, uh, no worries. You did mention

950
00:44:20,239 --> 00:44:22,760
a to Ah, and I do want to ask about

951
00:44:22,760 --> 00:44:24,920
this because you meant agent to agent, I think, and

952
00:44:25,079 --> 00:44:28,679
not A to A as the protocol that GDP released

953
00:44:28,679 --> 00:44:30,519
to do MCP right.

954
00:44:32,000 --> 00:44:35,000
Speaker 4: So well, so the the the g CP A to

955
00:44:35,079 --> 00:44:37,960
A is a true agent to agent that is sort

956
00:44:37,960 --> 00:44:39,559
of compatible with MCP.

957
00:44:39,880 --> 00:44:41,719
Speaker 1: It works well, and it's it's a good concept.

958
00:44:42,280 --> 00:44:44,400
Speaker 4: It's just one one of the many protocols that are

959
00:44:44,400 --> 00:44:46,480
coming out for agent agent communications right now.

960
00:44:47,239 --> 00:44:48,039
Speaker 1: It's just newer.

961
00:44:48,079 --> 00:44:50,480
Speaker 4: It's a it's a newer concept entirely, but I think

962
00:44:50,519 --> 00:44:52,679
it's similar to MCP and that it's solving a problem

963
00:44:52,679 --> 00:44:54,719
that everyone was just solving in a million different ways.

964
00:44:54,760 --> 00:44:57,280
Speaker 1: Right with MCP, people new agents.

965
00:44:56,960 --> 00:44:59,679
Speaker 4: Wanted to have remote tool calling as as an option

966
00:44:59,760 --> 00:45:02,000
and CP was a protocol that form to solve that,

967
00:45:02,239 --> 00:45:04,079
and a day is similar. It's you know, agents need

968
00:45:04,119 --> 00:45:05,519
to talk to each other. How do we have just

969
00:45:05,599 --> 00:45:07,760
like what's the most simple way to make that happen?

970
00:45:08,119 --> 00:45:09,760
Speaker 3: Well, if you're if you're talking about that, I have

971
00:45:09,760 --> 00:45:12,360
to ask, have you seen the one where someone devised

972
00:45:12,719 --> 00:45:15,800
the idea of oh it's so expensive to send request

973
00:45:15,960 --> 00:45:19,719
using Bluetooth or USB or Wi Fi. Let's just send

974
00:45:19,719 --> 00:45:22,599
an audio signal over the air from one device to

975
00:45:22,639 --> 00:45:24,719
another one and have like, you know, your phone will

976
00:45:24,719 --> 00:45:26,719
make a sound and your computer will pick it up

977
00:45:26,960 --> 00:45:29,920
on its you know, speaker or microphone and listen to it.

978
00:45:31,039 --> 00:45:31,280
Speaker 1: Yeah.

979
00:45:31,400 --> 00:45:34,280
Speaker 4: I think that's how like people in the old days

980
00:45:34,280 --> 00:45:37,199
you do things on iPhone when when Apple hadn't exposed

981
00:45:37,199 --> 00:45:39,480
all of the SDKs or the APIs yet.

982
00:45:39,679 --> 00:45:42,000
Speaker 3: I like how you put that, because yes, I mean,

983
00:45:42,039 --> 00:45:43,800
this was how you did it in the old days

984
00:45:43,800 --> 00:45:46,719
before we invented technology that did it correctly, and now'll

985
00:45:46,719 --> 00:45:48,760
people like, well, I have an agent over here and

986
00:45:48,800 --> 00:45:51,000
an agent over here, how do I get them to communicate?

987
00:45:52,400 --> 00:45:54,639
Speaker 2: There are a lot like a fax.

988
00:45:55,000 --> 00:45:57,320
Speaker 1: I was just like.

989
00:45:57,760 --> 00:46:01,320
Speaker 3: Dial up go. Well, I like, there it goes, you know,

990
00:46:02,039 --> 00:46:05,199
it just DSL you know, making those things over broadband Yep.

991
00:46:05,960 --> 00:46:08,199
Speaker 4: The stripe reader that was plugged into the headphone slot,

992
00:46:08,280 --> 00:46:10,880
it did the same thing, converted to a microphone signal

993
00:46:10,880 --> 00:46:11,920
from your credit card number.

994
00:46:13,599 --> 00:46:14,239
Speaker 1: Yeah, I think so.

995
00:46:14,440 --> 00:46:18,280
Speaker 4: Yeah, don't quote me on that, but I'm like nearly

996
00:46:18,320 --> 00:46:19,639
positive that that's how it worked.

997
00:46:21,280 --> 00:46:23,519
Speaker 3: I have this seeking suspicion that most of the audience

998
00:46:23,559 --> 00:46:25,519
will have no idea what you're talking about anyway, So

999
00:46:25,760 --> 00:46:26,480
don't worry about it.

1000
00:46:28,800 --> 00:46:32,039
Speaker 1: I mean, w am I old now? Is that the Yeah?

1001
00:46:33,039 --> 00:46:36,440
Speaker 3: I think we're all technically old, so yeah, yeah, it's fair.

1002
00:46:37,440 --> 00:46:39,360
Speaker 2: That is true. Though, Like I've been doing this for

1003
00:46:39,519 --> 00:46:43,400
three decades, and I had to go around and ask

1004
00:46:43,519 --> 00:46:46,119
if I was the oldest person at the company, because

1005
00:46:46,159 --> 00:46:50,599
I was like pretty confident that I was, and it

1006
00:46:50,679 --> 00:46:52,599
turns out I wasn't. There's one guy who's a couple

1007
00:46:52,599 --> 00:46:55,800
of years older than me, but it's close enough in

1008
00:46:55,920 --> 00:46:58,239
terms of the ages of the rest of the people

1009
00:46:58,280 --> 00:47:01,280
in the company where we're practically the same age.

1010
00:47:01,679 --> 00:47:04,480
Speaker 4: Yeah, yeah, we have a we have a pretty young

1011
00:47:04,519 --> 00:47:07,440
company too, but we've as we've grown, we've gotten some

1012
00:47:07,480 --> 00:47:09,880
more maturity, and honestly, it's it's much better. It's much

1013
00:47:09,880 --> 00:47:12,199
better to have just a more mature company.

1014
00:47:13,639 --> 00:47:17,119
Speaker 2: I think that's an interesting dynamic to balance, you know,

1015
00:47:17,199 --> 00:47:23,360
because you have like the a lot of the enthusiasm

1016
00:47:23,400 --> 00:47:27,480
and the excitement that comes from people who are earlier

1017
00:47:27,519 --> 00:47:32,559
in their career, but there's also like a really nice

1018
00:47:32,599 --> 00:47:35,559
balance when you have people who are who are senior

1019
00:47:35,639 --> 00:47:41,559
in their career to provide like perspective on those ideas,

1020
00:47:41,559 --> 00:47:43,599
you know, and sometimes you get that dynamic where it

1021
00:47:43,639 --> 00:47:46,039
feels like the young people are just trying to do

1022
00:47:46,079 --> 00:47:47,960
something and the old people are just trying to say no.

1023
00:47:48,039 --> 00:47:51,360
But if you get that really great combination there, you

1024
00:47:51,400 --> 00:47:54,320
can get this dynamic where the young people are coming

1025
00:47:54,400 --> 00:47:59,840
up with new ideas and the older people can say like, yeah,

1026
00:48:00,079 --> 00:48:04,320
we used to do that in the nineties and here's

1027
00:48:04,360 --> 00:48:06,559
why we changed, and then you kind of iterate on

1028
00:48:06,639 --> 00:48:09,320
that and evolve into something completely different.

1029
00:48:10,239 --> 00:48:12,519
Speaker 3: I think you should hire him because of all his

1030
00:48:12,599 --> 00:48:15,320
great ideas.

1031
00:48:16,760 --> 00:48:19,000
Speaker 4: I mean, I completely agree, though it's you know, they've

1032
00:48:19,199 --> 00:48:21,000
really it's sort of a balance that's important.

1033
00:48:21,000 --> 00:48:22,400
Speaker 1: But also bringing just years of.

1034
00:48:22,360 --> 00:48:25,400
Speaker 4: Expertise on you know, I'm thinking go to market in

1035
00:48:25,400 --> 00:48:27,199
this case, but you know you have something, you can

1036
00:48:27,239 --> 00:48:29,599
sell it at the beginning by years of experience, just

1037
00:48:29,679 --> 00:48:31,599
help you sell something, like, really bring it to market,

1038
00:48:31,639 --> 00:48:34,320
really scale it. So I think about like kind of

1039
00:48:34,360 --> 00:48:37,480
you know, earlier talent is helping you get something off

1040
00:48:37,480 --> 00:48:40,079
the ground, and then late later career talent is really

1041
00:48:40,119 --> 00:48:41,440
helping you push it and scale it.

1042
00:48:41,760 --> 00:48:44,199
Speaker 3: I feel like we're losing I mean, even with years

1043
00:48:44,239 --> 00:48:47,239
of experience, I feel like we're losing actually the original

1044
00:48:47,280 --> 00:48:50,480
concept of later year talent because I think it came

1045
00:48:50,559 --> 00:48:54,679
from having managed systems that were alive for a very

1046
00:48:54,760 --> 00:48:57,960
long time and understanding the nuances. They're like, oh, yeah,

1047
00:48:58,159 --> 00:49:01,159
we've had a you know, in our data center, we've

1048
00:49:01,199 --> 00:49:03,679
had a mainframe there for fifty years that's been running

1049
00:49:03,719 --> 00:49:07,079
you know, whatever it was running from. I assume IBM

1050
00:49:07,440 --> 00:49:09,880
and you know, going and going, and these are the

1051
00:49:09,880 --> 00:49:13,199
weird things that we saw. And now it's fifty years. Like, no,

1052
00:49:13,320 --> 00:49:17,599
that's like twenty five different people were integrating with that, right, Like,

1053
00:49:17,639 --> 00:49:20,519
it's not one person who had seen everything there was

1054
00:49:20,559 --> 00:49:22,400
to see with that one piece of technology, that one

1055
00:49:22,440 --> 00:49:24,960
service or one product, And I don't know if that's

1056
00:49:25,000 --> 00:49:26,760
ever coming back, which means I feel like we're losing

1057
00:49:26,840 --> 00:49:27,639
as a society.

1058
00:49:27,719 --> 00:49:27,840
Speaker 1: This.

1059
00:49:29,239 --> 00:49:32,320
Speaker 3: It's a critically useful piece of information as far as

1060
00:49:32,360 --> 00:49:35,440
how we actually deal with these systems or those experiences

1061
00:49:35,480 --> 00:49:37,039
to be able to guide us in the right direction

1062
00:49:37,079 --> 00:49:37,639
going forward.

1063
00:49:38,440 --> 00:49:40,880
Speaker 4: Yeah, but I guess the question is is that important too,

1064
00:49:40,920 --> 00:49:42,920
because you now have this sort of abstraction of knowledge

1065
00:49:42,920 --> 00:49:45,639
where you have an AWS or a GCP managed service

1066
00:49:45,760 --> 00:49:46,320
that you know.

1067
00:49:46,360 --> 00:49:47,880
Speaker 1: I mean, I think it depends on the example.

1068
00:49:47,960 --> 00:49:50,920
Speaker 4: But you keep getting, you keep abstracting above and then

1069
00:49:50,960 --> 00:49:53,119
people can focus on like a different level of skill.

1070
00:49:53,880 --> 00:49:56,880
Speaker 3: I think it's something different like if we see that

1071
00:49:57,280 --> 00:49:59,639
the majority of people in the world are spending their

1072
00:49:59,679 --> 00:50:03,119
effort and how they're focusing on problems that become more

1073
00:50:03,159 --> 00:50:07,559
and more short term than we are losing those situations

1074
00:50:07,559 --> 00:50:10,599
where people have experience working with long term functionality. And

1075
00:50:10,599 --> 00:50:13,239
I think this is coming to the cloud providers and

1076
00:50:13,320 --> 00:50:17,119
hyperscalers out there unfortunately, and we'll see that if everyone

1077
00:50:17,159 --> 00:50:20,199
outside of them doesn't have long term experience and the

1078
00:50:20,199 --> 00:50:22,239
only people that for those companies to hire are people

1079
00:50:22,239 --> 00:50:25,400
without long term experience, and what the market cares about.

1080
00:50:25,880 --> 00:50:28,639
If I see lots of little data centers and cloud

1081
00:50:28,639 --> 00:50:30,920
providers pop up to say, oh, we're better than AWS

1082
00:50:30,960 --> 00:50:33,920
and GCP and whomever because we do this thing, and

1083
00:50:33,920 --> 00:50:37,239
it's like, well, you're better because you wrote a hack

1084
00:50:37,320 --> 00:50:39,920
that got it done in six months compared to a

1085
00:50:39,960 --> 00:50:42,599
company that's been around and has that particular service for

1086
00:50:42,599 --> 00:50:45,800
twenty years. And it's not as good, but it still

1087
00:50:45,840 --> 00:50:48,039
maybe solves your particular use case. And I think this

1088
00:50:48,079 --> 00:50:52,119
is similar to hard hardened API versus MTP server written

1089
00:50:52,119 --> 00:50:54,199
on top of it. You know, we wrote something quick

1090
00:50:54,480 --> 00:50:56,480
and dirty to get it done. It's a hack and

1091
00:50:56,679 --> 00:51:00,320
it now it becomes ingrained in what we're utilizing and

1092
00:51:00,400 --> 00:51:02,519
so there's a lot of risk associated with it. And

1093
00:51:02,559 --> 00:51:06,719
I think this is a this is a failure of

1094
00:51:06,760 --> 00:51:10,119
the human race to you know, be so short and

1095
00:51:10,239 --> 00:51:12,960
narrow like narrow sighted there, like you really focus on

1096
00:51:13,360 --> 00:51:15,920
the short term. Having a long term focus is very

1097
00:51:15,960 --> 00:51:16,760
difficult for people.

1098
00:51:16,800 --> 00:51:22,440
Speaker 2: I think true story.

1099
00:51:21,400 --> 00:51:23,599
Speaker 3: I like this will be my I think this will

1100
00:51:23,599 --> 00:51:25,559
be one of my future picks. But there's a great

1101
00:51:25,599 --> 00:51:29,400
science fiction television show called Well, actually written after a

1102
00:51:29,400 --> 00:51:34,760
book called The Expanse, and the one of the characters Abursala,

1103
00:51:34,840 --> 00:51:37,079
who's I don't need to go into it, she has

1104
00:51:37,119 --> 00:51:39,840
this really great quote that the failure of humanity is

1105
00:51:40,000 --> 00:51:42,239
too little, too late, And I think it really does

1106
00:51:42,280 --> 00:51:45,480
go to the fact that humans do really wait too

1107
00:51:45,519 --> 00:51:47,960
long to see a problem and attempt to fix it,

1108
00:51:48,000 --> 00:51:50,559
even even if without even putting in enough effort to

1109
00:51:50,599 --> 00:51:52,920
actually solve it. So you know, I'm with her. I'm

1110
00:51:52,960 --> 00:51:58,440
very pessimistic on the on the topic. But yeah, I just.

1111
00:51:58,360 --> 00:52:00,000
Speaker 4: Think of the word AI and I'm like, the world

1112
00:52:00,079 --> 00:52:02,039
as we know it is over and I just can't

1113
00:52:02,039 --> 00:52:03,000
think about the problems.

1114
00:52:03,079 --> 00:52:04,360
Speaker 1: But I know I'm with you.

1115
00:52:04,440 --> 00:52:07,159
Speaker 3: It's well, I think you know there's something there and

1116
00:52:07,360 --> 00:52:09,920
I am always the first to bash anything related to

1117
00:52:09,960 --> 00:52:12,360
AI because it's not AI as I understand it. It's

1118
00:52:12,400 --> 00:52:17,079
these probabilistic engines that are just returning a statistical result

1119
00:52:17,119 --> 00:52:19,519
that has no intelligence behind it whatsoever. I mean, there

1120
00:52:19,559 --> 00:52:22,599
was intelligence to build the system, but actually contained in

1121
00:52:22,639 --> 00:52:28,719
the computational matrix of the anyway. Yeah, so lack of

1122
00:52:28,760 --> 00:52:29,639
lack of intelligence.

1123
00:52:31,079 --> 00:52:33,039
Speaker 1: That's fine. I think most human brains are similar.

1124
00:52:33,079 --> 00:52:39,719
Speaker 3: So yeah, well there's a there's a huge debate there.

1125
00:52:39,719 --> 00:52:41,960
I mean, if LM is what we created in humanity

1126
00:52:42,000 --> 00:52:46,559
after downloading all of the information and compiling it, then

1127
00:52:46,840 --> 00:52:50,000
it should be an average. So I'd say that it

1128
00:52:50,079 --> 00:52:52,480
must be that lms are better than fifty percent of

1129
00:52:52,559 --> 00:52:55,239
humanity of the people who contributed to them, and worse

1130
00:52:55,280 --> 00:52:57,119
than the other fifty percent. And if you're okay with

1131
00:52:57,199 --> 00:52:59,679
the average of an output, the average code that's being

1132
00:52:59,679 --> 00:53:03,320
generated or workflow is being executed, then an LM may

1133
00:53:03,360 --> 00:53:05,920
be an improvement there an average amount of knowledge.

1134
00:53:05,760 --> 00:53:08,639
Speaker 4: Yeah, but it is also it is also all the

1135
00:53:08,719 --> 00:53:11,639
knowledge that everyone knew throughout their entire lifetimes versus not

1136
00:53:11,679 --> 00:53:13,119
all useful moment in time.

1137
00:53:13,320 --> 00:53:18,119
Speaker 3: Yeah, there's things there, right, I'm pretty sure it was like,

1138
00:53:18,199 --> 00:53:20,800
and I'm going to get this wrong, like Aristotle believed.

1139
00:53:20,920 --> 00:53:24,639
I mean, people recognize the name Aristotle believe that the

1140
00:53:24,679 --> 00:53:27,920
Sun went around the Earth, So you know, that's sort

1141
00:53:27,960 --> 00:53:30,000
of an interesting thing. So, yes, there's a lot of

1142
00:53:30,119 --> 00:53:33,440
history of information, and uh, there's a lot that's wrong.

1143
00:53:34,199 --> 00:53:38,639
Speaker 2: Yes, sure, yes, we need to feed LLM with the

1144
00:53:38,639 --> 00:53:42,920
the redlined version of our human collective consciousness.

1145
00:53:43,119 --> 00:53:45,119
Speaker 3: Well that's that, you know. And the problem is that

1146
00:53:45,159 --> 00:53:48,079
the information that's being fed in to train lms today

1147
00:53:48,280 --> 00:53:51,480
isn't being sanitized as much as data engineering was doing

1148
00:53:51,519 --> 00:53:53,519
in the past. They're, you know, they're realizing that there's

1149
00:53:53,559 --> 00:53:56,400
too much information to do this, so they're coming up

1150
00:53:56,400 --> 00:53:58,599
with tricks and strategies to try to filter stuff out.

1151
00:53:59,000 --> 00:54:02,480
But often you're picking you're filtering the sources and the

1152
00:54:02,599 --> 00:54:06,639
attributes and the functionality rather than the accuracy, which is very,

1153
00:54:06,719 --> 00:54:08,840
very difficult. And now, for sure, if you find something wrong,

1154
00:54:08,880 --> 00:54:10,840
you can go back to the training data and start

1155
00:54:10,840 --> 00:54:14,000
eliminating things, get it to solve certain problems. But I mean,

1156
00:54:14,239 --> 00:54:16,960
mathematics or equation solving today is something because of how

1157
00:54:17,000 --> 00:54:20,199
the lms are designed, just is almost never getting better.

1158
00:54:20,280 --> 00:54:23,639
Speaker 4: For instance, yeah, the cool thing there is is you know,

1159
00:54:23,679 --> 00:54:26,679
if it can successfully call a calculator tool via MCP,

1160
00:54:26,880 --> 00:54:29,239
then there you go, or even a local tool, then

1161
00:54:29,239 --> 00:54:30,880
at least they can formulate that and let let a

1162
00:54:30,880 --> 00:54:32,199
calculator statically.

1163
00:54:31,760 --> 00:54:34,480
Speaker 3: Do it MCP for Wolfram Alpha, that's what I'm hearing.

1164
00:54:34,480 --> 00:54:38,440
Speaker 4: I'm sure someone oh man, that was the most advanced

1165
00:54:38,440 --> 00:54:39,880
a they had when I was in college.

1166
00:54:39,920 --> 00:54:44,599
Speaker 2: So yeah, right on, this feels like a good point

1167
00:54:44,599 --> 00:54:45,920
to move into picks. What do you think?

1168
00:54:46,960 --> 00:54:50,159
Speaker 3: I think we should definitely do it, all right, Yeah,

1169
00:54:50,360 --> 00:54:53,079
so I have something non technical this week. I'm just

1170
00:54:53,119 --> 00:54:54,559
going to jump in and save mine because I know,

1171
00:54:54,639 --> 00:54:55,800
you know, if I don't will it's just gonna be

1172
00:54:55,880 --> 00:54:59,159
like Warren, you know what time it is time for

1173
00:54:59,280 --> 00:55:03,079
my pick, because I always go first. So my pick

1174
00:55:03,239 --> 00:55:06,559
is the book A Slash television show The Magicians by

1175
00:55:06,760 --> 00:55:11,440
Lev Grossman. I'm I don't know something something science fiction, fantasy.

1176
00:55:12,280 --> 00:55:14,480
There is a question is fantasy different from science fiction?

1177
00:55:14,960 --> 00:55:18,480
And I think that The Magicians gets into that quite

1178
00:55:18,519 --> 00:55:20,559
a little bit that it's hard to distinguish, which is

1179
00:55:20,559 --> 00:55:22,760
which I really like the book and the show. They're

1180
00:55:22,960 --> 00:55:25,280
they're good in different ways. Unfortunately, Like it's not like

1181
00:55:25,599 --> 00:55:28,199
they well recreated. It's sort of like a different storyline

1182
00:55:28,199 --> 00:55:30,760
and different stuff happen. Characters that are good in the

1183
00:55:30,840 --> 00:55:33,280
in the book are bad in the show, and vice versa.

1184
00:55:33,360 --> 00:55:35,880
So if you've only read or watched one, I highly

1185
00:55:35,880 --> 00:55:36,559
recommend the other.

1186
00:55:36,840 --> 00:55:40,360
Speaker 2: Right on, Gil, you gave us a sneak preview of

1187
00:55:40,400 --> 00:55:43,400
your pick before we started recording, so I'm excited what

1188
00:55:43,800 --> 00:55:44,159
you got.

1189
00:55:44,320 --> 00:55:44,599
Speaker 1: Yeah.

1190
00:55:44,639 --> 00:55:47,400
Speaker 4: So I'm a big watch fan, and a topic that

1191
00:55:47,440 --> 00:55:50,519
I've been rereading about recently that I'm really into is

1192
00:55:50,800 --> 00:55:53,440
the historical notion it's called of constant escapement. It's a

1193
00:55:53,480 --> 00:55:58,599
problem in watchmaking where and clockmaking and all formsquare and sorry,

1194
00:55:58,639 --> 00:56:00,199
And the reason I bring this up is because with

1195
00:56:00,239 --> 00:56:03,159
AI technology just changing so fast, what I love about

1196
00:56:03,199 --> 00:56:06,519
watches is it's technology. There's no electronics and nothing really

1197
00:56:06,559 --> 00:56:09,519
changes except maybe material science gets better. And so with

1198
00:56:09,599 --> 00:56:12,800
the constant escaements this idea of you know, in watches

1199
00:56:12,880 --> 00:56:15,400
or in clocks, you wind something up. When you have

1200
00:56:15,440 --> 00:56:18,559
a wound up spring, as it unwinds, the power that

1201
00:56:18,599 --> 00:56:21,639
it gives off decreases. Right, It's more powerful when it's

1202
00:56:21,679 --> 00:56:23,719
really tightly wound. And so if you picture what that

1203
00:56:23,760 --> 00:56:26,440
does on a clock, you're gonna have really fast movement

1204
00:56:26,440 --> 00:56:28,199
of the hands and then it'll start to slow down

1205
00:56:28,639 --> 00:56:31,000
and that doesn't work. And so the huge problem always

1206
00:56:31,000 --> 00:56:33,559
has been constant escapement. How do you get constant forced

1207
00:56:33,599 --> 00:56:36,960
to be emitted from that that spring? And that's why

1208
00:56:37,000 --> 00:56:39,119
you see this idea of something you know balance wee

1209
00:56:39,159 --> 00:56:42,039
all going back and forth. It helps slowly release tension

1210
00:56:42,039 --> 00:56:44,480
from that spring. Or on a clock you have a

1211
00:56:44,519 --> 00:56:46,639
pendulum that swings back and forth. You need to have

1212
00:56:46,719 --> 00:56:50,320
ways to very like let gravity help you release that.

1213
00:56:50,559 --> 00:56:53,039
So I think it's a super interesting topic if you're

1214
00:56:53,039 --> 00:56:54,679
into attack and you want to read about something that's

1215
00:56:54,760 --> 00:56:58,039
just not changing so fast. But it's just a cool

1216
00:56:58,119 --> 00:57:00,840
historical thing. Look up constant escapement and read about all

1217
00:57:00,840 --> 00:57:02,480
the different ways it's been solved over time.

1218
00:57:02,679 --> 00:57:05,840
Speaker 3: It sounds like a conservation of angular momentum problem.

1219
00:57:06,400 --> 00:57:08,159
Speaker 1: Yes, it's a similar idea.

1220
00:57:10,000 --> 00:57:12,239
Speaker 3: I don't if you tell I I took a math

1221
00:57:12,280 --> 00:57:14,920
in physics as my U that was who I was

1222
00:57:14,960 --> 00:57:17,719
before I went into the philosophy of software engineering.

1223
00:57:19,840 --> 00:57:22,119
Speaker 1: It's it's such an interesting problem. I really love that.

1224
00:57:22,239 --> 00:57:24,400
Speaker 4: Again, like it's it's it's not to say that AI

1225
00:57:24,440 --> 00:57:28,199
and technology isn't great but it's just a fun different

1226
00:57:28,239 --> 00:57:29,320
way of thinking about things.

1227
00:57:29,760 --> 00:57:32,880
Speaker 3: Is there like a particular watch or a clock type

1228
00:57:32,960 --> 00:57:35,239
or something that you much prefer over others? Like do

1229
00:57:35,239 --> 00:57:37,599
you have like a giant grandfather clock sitting in your home?

1230
00:57:38,039 --> 00:57:41,679
Speaker 4: I so, I actually really like this clock called it's

1231
00:57:41,800 --> 00:57:46,039
JLC or La Colts. It's a it's called the atmost clock,

1232
00:57:46,519 --> 00:57:48,880
and you've probably seen them around, but it's sits in

1233
00:57:48,880 --> 00:57:51,360
your room, and it also can run forever no electricity,

1234
00:57:51,400 --> 00:57:52,599
and it's not on your wrist, so it's not.

1235
00:57:52,559 --> 00:57:53,440
Speaker 1: Wound or anything.

1236
00:57:53,960 --> 00:57:55,960
Speaker 4: It has a big bulb in it that's filled with

1237
00:57:56,000 --> 00:57:59,239
a gas, and even one degree of temperature variation in

1238
00:57:59,280 --> 00:58:01,760
the room makes it expand and then that is is

1239
00:58:01,880 --> 00:58:04,440
somehow winding up a mechanism when it expands and contracts.

1240
00:58:04,480 --> 00:58:05,880
Speaker 1: So I think that was really cool.

1241
00:58:06,320 --> 00:58:09,679
Speaker 2: Oh wow, that's pretty wild. That's super cool.

1242
00:58:10,480 --> 00:58:13,800
Speaker 3: I always get concerned when there's no like battery electricity,

1243
00:58:13,840 --> 00:58:16,400
that if it's not if it's not taking energy, it's

1244
00:58:16,440 --> 00:58:21,199
you know, giving off radiation, And.

1245
00:58:21,199 --> 00:58:23,360
Speaker 4: A lot of watches do give off radiation from the

1246
00:58:23,360 --> 00:58:26,000
the tritium tubes in the you know, in the the loom.

1247
00:58:26,159 --> 00:58:30,000
Speaker 3: So interesting besides being like low in the dark or

1248
00:58:30,159 --> 00:58:32,360
is it different purpose?

1249
00:58:33,039 --> 00:58:34,719
Speaker 1: Yeah, no, just for going the dark.

1250
00:58:35,239 --> 00:58:36,480
Speaker 4: You don't you know a lot of these days you

1251
00:58:36,480 --> 00:58:38,119
don't see a lot of tridium tubes, but you still

1252
00:58:38,159 --> 00:58:39,800
see them. They're not I think it's not to a

1253
00:58:39,880 --> 00:58:42,960
level that's considered dangerous, like a lot of this lead

1254
00:58:42,960 --> 00:58:44,760
paint that you see on like old plates and stuff,

1255
00:58:44,800 --> 00:58:48,159
but it's still you know, I I personally wouldn't wear one.

1256
00:58:48,800 --> 00:58:53,199
Speaker 2: Yeah, it's the It's an alpha wave emitter, and alpha

1257
00:58:53,760 --> 00:58:58,400
wave radiation can be stopped by the outer dead layers

1258
00:58:58,440 --> 00:59:04,039
of skin, so its ability to impact you or do

1259
00:59:04,119 --> 00:59:07,159
anything is super super low risk.

1260
00:59:07,440 --> 00:59:11,280
Speaker 3: Yeah, let's talk about radiation. Yeah, I mean alpha beta

1261
00:59:11,320 --> 00:59:13,199
is fine, but you know usually when we talk about

1262
00:59:13,239 --> 00:59:16,360
bad radiation, it's it's gamma or something stronger, which is

1263
00:59:16,920 --> 00:59:20,599
you know, multiple particle size, not just a single hydrogen

1264
00:59:20,639 --> 00:59:22,079
atom or electron.

1265
00:59:23,239 --> 00:59:25,199
Speaker 1: Yeah, so no nuclear reactor is on your risk.

1266
00:59:26,400 --> 00:59:28,760
Speaker 3: Well, I'm worried about wearing a piece of technology that

1267
00:59:28,800 --> 00:59:31,840
has five g antenna in it. So you know, that's

1268
00:59:31,880 --> 00:59:34,280
that's my own I would have to say conspiracy theory,

1269
00:59:34,320 --> 00:59:36,239
but that that is my own personal fear.

1270
00:59:37,519 --> 00:59:39,599
Speaker 4: I mean, I I get that, like we don't know,

1271
00:59:39,679 --> 00:59:41,239
so why take the risk. I mean, I know a

1272
00:59:41,239 --> 00:59:43,320
lot of people like my sister and her husband, they

1273
00:59:43,320 --> 00:59:45,760
put their phones on or on on airplane mode next

1274
00:59:45,760 --> 00:59:47,800
to their beds every night, or they keep their phone

1275
00:59:47,840 --> 00:59:48,480
across the room.

1276
00:59:48,760 --> 00:59:51,440
Speaker 3: I think outside like I think it's actually the heat

1277
00:59:51,519 --> 00:59:56,760
is worse than the radio. Yeah, or in your pocket.

1278
00:59:56,800 --> 00:59:58,840
I think there's a bunch of research done so like

1279
00:59:59,039 --> 01:00:03,800
keeping it outside your b it is for sure better there. Well,

1280
01:00:03,960 --> 01:00:05,840
you seem like you know about radiation and watch as.

1281
01:00:07,960 --> 01:00:08,440
Speaker 1: A little bit.

1282
01:00:08,480 --> 01:00:11,760
Speaker 2: You know, I was a former nuclear engineer in the Navy,

1283
01:00:12,159 --> 01:00:14,400
so I studied a bit about radiation.

1284
01:00:15,079 --> 01:00:18,039
Speaker 3: What's his what's his favorite kind of reactor? He's like

1285
01:00:18,079 --> 01:00:24,199
the authorium you yellow cake, uranium two fifty eight like water, right,

1286
01:00:24,760 --> 01:00:26,800
it's it's got to be like that was one of

1287
01:00:26,840 --> 01:00:30,639
the interesting things because when I went through nuclear power school,

1288
01:00:31,280 --> 01:00:38,639
it was just after the Chernobyl incident and and so

1289
01:00:38,760 --> 01:00:42,760
like the US was parading around the fact that we

1290
01:00:42,920 --> 01:00:48,360
use all water based coolant or water cooled reactors. And

1291
01:00:48,400 --> 01:00:52,559
the interesting thing about that is as as the water

1292
01:00:53,159 --> 01:00:56,280
gets heated up, the density and this is we're going

1293
01:00:56,320 --> 01:00:59,039
back like over, We're going back to the late eighties.

1294
01:00:59,039 --> 01:01:01,000
For me, to remember this sum likely going to botch

1295
01:01:01,039 --> 01:01:04,800
most of these facts. But as the water heated up,

1296
01:01:04,840 --> 01:01:09,039
the atoms got closer together, so even though the radioactivity

1297
01:01:09,119 --> 01:01:13,519
of the nuclear reactor was increasing, the increased density of

1298
01:01:13,679 --> 01:01:20,159
the molecules of water effectively shunted that increase in radiation.

1299
01:01:20,360 --> 01:01:24,679
So it was almost impossible for a water cooled reactor

1300
01:01:24,760 --> 01:01:29,440
to overheat the way that Chernobyl did because Chernobyl used

1301
01:01:29,800 --> 01:01:33,679
liquid sodium as they're coolant. And then Three Mile Island

1302
01:01:33,880 --> 01:01:35,920
and Fukushima.

1303
01:01:35,599 --> 01:01:39,920
Speaker 2: Yeah, and Three Mile Island was great because they got

1304
01:01:39,960 --> 01:01:43,239
the first alert telling that there was a problem, assumed

1305
01:01:43,239 --> 01:01:46,960
that it was a faulty sensor, and then they got

1306
01:01:47,000 --> 01:01:50,480
the second alert, which was downstream from that one, and

1307
01:01:50,519 --> 01:01:54,039
they were like, damn, we got two sensors that failed today.

1308
01:01:54,719 --> 01:01:58,760
And then it finally started spewing out into the river

1309
01:01:58,840 --> 01:01:59,400
and they're.

1310
01:01:59,199 --> 01:02:01,880
Speaker 3: Like, oh, oh, oh.

1311
01:02:01,559 --> 01:02:03,239
Speaker 2: Shit, we're going to fill out some sleeper work on

1312
01:02:03,280 --> 01:02:05,960
this one.

1313
01:02:06,639 --> 01:02:09,960
Speaker 4: Oh man, much worse than software, you know, ignoring an

1314
01:02:10,000 --> 01:02:11,599
alert that pops up, right.

1315
01:02:11,920 --> 01:02:14,280
Speaker 3: You just got to put the MCP server in front

1316
01:02:14,280 --> 01:02:16,960
of the nuclear reactor and problem right, you.

1317
01:02:16,880 --> 01:02:18,719
Speaker 1: Know, solves everything.

1318
01:02:21,079 --> 01:02:23,360
Speaker 3: I'm surprised. I also I also did Uh, I didn't

1319
01:02:23,360 --> 01:02:25,400
go as far as you did, well, but I definitely

1320
01:02:25,440 --> 01:02:27,639
got the book, and I'm pretty sure that in my

1321
01:02:27,880 --> 01:02:31,440
nuclear physics book there was no mention of MCP servers

1322
01:02:31,480 --> 01:02:36,800
anywhere in there, and it's clearly an oversight, clearly an oversite.

1323
01:02:37,599 --> 01:02:38,880
So uh, what's your pick?

1324
01:02:39,880 --> 01:02:42,400
Speaker 2: So my pick this is a repeat pick for me,

1325
01:02:42,920 --> 01:02:46,119
because I'm still pissed at you over this warn. I'm

1326
01:02:46,119 --> 01:02:50,800
working my way through the Dungeon Crawler Carl series because

1327
01:02:50,920 --> 01:02:54,679
every book is so great, and I've pieced together at

1328
01:02:54,679 --> 01:02:56,960
this point that there's I think there's seven books in

1329
01:02:56,960 --> 01:02:58,039
this series right.

1330
01:02:59,400 --> 01:02:59,679
Speaker 1: Starting.

1331
01:03:00,159 --> 01:03:02,119
Speaker 3: I don't know why you're mad at me because it

1332
01:03:02,199 --> 01:03:04,639
was Matt Lee that brought up a dungeon collar pull

1333
01:03:04,719 --> 01:03:07,239
and I haven't read it yet, so you can be

1334
01:03:07,280 --> 01:03:09,039
mad at me all you want, but I don't know

1335
01:03:09,039 --> 01:03:09,719
what I don't know.

1336
01:03:10,239 --> 01:03:13,039
Speaker 2: Then I flipped that conversation in my head because I

1337
01:03:13,519 --> 01:03:16,440
remember it as you bringing it up. But okay, I

1338
01:03:16,800 --> 01:03:18,119
haven't read it.

1339
01:03:18,119 --> 01:03:20,599
Speaker 3: It's now on my list, and I you know, it's

1340
01:03:20,639 --> 01:03:23,199
always great when there's like multiple books in that series,

1341
01:03:23,239 --> 01:03:25,079
so you know, every single time you bring this up,

1342
01:03:25,199 --> 01:03:27,440
I'm like, okay, a good reminder that this is going

1343
01:03:27,519 --> 01:03:28,960
to have to be the next thing that I read.

1344
01:03:29,599 --> 01:03:32,880
Speaker 2: The disappointing thing is I'm five books into it now,

1345
01:03:33,239 --> 01:03:36,519
and there's seven books in a series. But I'm starting

1346
01:03:36,559 --> 01:03:41,000
to piece together the picture that the story's not going

1347
01:03:41,039 --> 01:03:43,719
to be complete by the time I finished book seven,

1348
01:03:43,960 --> 01:03:45,719
and then I'm going to be stuck waiting for him

1349
01:03:45,760 --> 01:03:47,679
to write the rest of the damn books so that

1350
01:03:47,679 --> 01:03:49,519
I get closure on this whole story.

1351
01:03:50,320 --> 01:03:52,199
Speaker 3: Aren't you taking salts in the fact the you'll propaly

1352
01:03:52,320 --> 01:03:55,119
You know, given given your years of experience, you may

1353
01:03:55,159 --> 01:03:57,599
forget what happened and then go back and reread the

1354
01:03:57,639 --> 01:04:00,280
book and you know, relived that greatness all the verga

1355
01:04:00,400 --> 01:04:01,360
when the next one comes out.

1356
01:04:02,159 --> 01:04:05,800
Speaker 2: Late career talent, yeah yeah, or just be dead by

1357
01:04:05,800 --> 01:04:08,079
the time it happens and it's not my problem anymore

1358
01:04:09,960 --> 01:04:15,159
either way. W's aw cool, all right, Gil, Thank you

1359
01:04:15,199 --> 01:04:17,639
so much, man. This has been super insightful. I appreciate

1360
01:04:17,719 --> 01:04:20,840
your insights, and you're taking the time to join us today.

1361
01:04:21,760 --> 01:04:23,639
Speaker 1: Awesome. Thank you so much for having me really enjoyed

1362
01:04:23,639 --> 01:04:25,360
the conversation. This is great.

1363
01:04:25,679 --> 01:04:28,400
Speaker 2: Yeah, Warren, thank you as always for joining me here

1364
01:04:28,400 --> 01:04:34,360
and carrying the conversation whenever I space Out and all

1365
01:04:34,360 --> 01:04:36,800
our listeners. Thank you guys for listening. Be sure and

1366
01:04:37,280 --> 01:04:39,599
hit us up if there's anything you want to see.

1367
01:04:39,639 --> 01:04:44,519
Elaborate on comments, thoughts, feedbacks, smart ass jabs.

1368
01:04:44,719 --> 01:04:45,280
Speaker 1: It's all good.

1369
01:04:45,519 --> 01:04:48,440
Speaker 2: Bring it on and we'll see everyone next week.