1
00:00:00,000 --> 00:00:03,000
Sergey Litvinenko: Guess how
much our payout is so far?

2
00:00:03,000 --> 00:00:05,000
Audrow Nash: 20 cents.

3
00:00:05,000 --> 00:00:06,000
Sergey Litvinenko: Five.

4
00:00:06,000 --> 00:00:08,000
Audrow Nash: Holy cow.

5
00:00:08,000 --> 00:00:11,088
Sergey Litvinenko: that's, how
much better it is. Even like

6
00:00:11,088 --> 00:00:14,333
we're an early stage company, we
don't have a ton of data yet,

7
00:00:14,333 --> 00:00:17,578
but from the data that we have,
it's already that much better.

8
00:00:17,578 --> 00:00:20,718
And it's, literally printing,
underwriting profits. And what

9
00:00:20,718 --> 00:00:23,858
we're going to do is that we're
going to take that and we're

10
00:00:23,858 --> 00:00:26,841
going to, do the iteration of
the models that we have and

11
00:00:26,841 --> 00:00:30,086
we're going to start bringing
the rates down. So it's going to

12
00:00:30,086 --> 00:00:33,121
be even cheaper to get our
insurance. that's the magnitude

13
00:00:33,121 --> 00:00:36,000
that I'm talking about, and that
gets me really excited

14
00:00:36,000 --> 00:00:39,625
Audrow Nash: Hey everyone,
Audrow here. Today we're diving

15
00:00:39,625 --> 00:00:43,313
into a fascinating intersection
of robotics and finance you

16
00:00:43,313 --> 00:00:46,938
might not have thought about
before. I'm talking to Sergey

17
00:00:46,938 --> 00:00:51,000
Litvinenko, a co founder and the
CEO of Koop AI, a company that's

18
00:00:51,000 --> 00:00:54,625
revolutionizing insurance for
robotics companies. Now, you

19
00:00:54,625 --> 00:00:58,313
might wonder, why should I care
about insurance for robots?

20
00:00:58,313 --> 00:01:01,688
Well, it turns out that as
robots become more and more

21
00:01:01,688 --> 00:01:05,563
common in our world, figuring
out how to insure them is a huge

22
00:01:05,563 --> 00:01:09,500
challenge and opportunity. Koop
AI, which is backed by Hyundai,

23
00:01:09,500 --> 00:01:13,063
is using cutting edge data
analysis to solve this problem

24
00:01:13,063 --> 00:01:16,938
in a way that could accelerate
the adoption of robotics across

25
00:01:16,938 --> 00:01:20,313
industries. If you're into
robotics, AI or you're just

26
00:01:20,313 --> 00:01:23,813
curious about innovative
business models in tech, you're

27
00:01:23,813 --> 00:01:27,313
going to love this conversation.
We get into some really

28
00:01:27,313 --> 00:01:31,063
interesting stuff about how they
assess risk for robots from

29
00:01:31,063 --> 00:01:34,125
autonomous vehicles to
industrial robots and even

30
00:01:34,125 --> 00:01:38,063
submarines. We also discuss the
future of the industry and some

31
00:01:38,063 --> 00:01:42,000
surprising insights about robot
behavior versus human behavior.

32
00:01:42,000 --> 00:01:45,813
So, let's jump right into it
with Sergey Litvinenko from Koop

33
00:01:45,813 --> 00:01:46,000
ai.

34
00:01:46,000 --> 00:01:49,000
Audrow Nash: Hi, Sergey. Would
you introduce yourself?

35
00:01:49,000 --> 00:01:53,761
Sergey Litvinenko: Hi Audrow.
thanks for having me. yep, sure.

36
00:01:53,761 --> 00:01:58,139
my name is Sergey Litvinenko.
I'm a co founder and CEO at

37
00:01:58,139 --> 00:02:02,055
Koop. We are an insurance
technology company that's

38
00:02:02,055 --> 00:02:05,741
focused on everything
automation, robotics, self

39
00:02:05,741 --> 00:02:10,426
driving cars, drones, and we
underwrite all kinds of risks in

40
00:02:10,426 --> 00:02:15,110
that space and, have a lot of
things going on that we'll talk

41
00:02:15,110 --> 00:02:19,641
about today. We're an early
stage company, Series A, raised

42
00:02:19,641 --> 00:02:24,172
7 million to date. one of our
backers is Hyundai and Kia, a

43
00:02:24,172 --> 00:02:28,933
global automotive conglomerate.
15 employees in two locations,

44
00:02:28,933 --> 00:02:33,618
New York area, and also Dubai,
UAE, and, making progress step

45
00:02:33,618 --> 00:02:35,000
by step. So, yeah.

46
00:02:35,000 --> 00:02:37,000
Audrow Nash: Okay. Hell yeah.

47
00:02:37,000 --> 00:02:42,613
Audrow Nash: so can you tell me
why you, like, why is this

48
00:02:42,613 --> 00:02:46,000
important and what is
underwriting?

49
00:02:46,000 --> 00:02:49,082
Sergey Litvinenko: Yeah. Let's
start first with, what is

50
00:02:49,082 --> 00:02:52,108
underwriting or what is
insurance in general? insurance

51
00:02:52,108 --> 00:02:55,135
is a financial product that
people and companies buy in

52
00:02:55,135 --> 00:02:58,381
order to protect themselves when
something bad happens. For

53
00:02:58,381 --> 00:03:01,848
example, you have a car accident
and, you have to pay a hundred

54
00:03:01,848 --> 00:03:05,204
thousand dollars for liability
because you damaged somebody's

55
00:03:05,204 --> 00:03:08,341
car or there was a medical
injury or something like that.

56
00:03:08,341 --> 00:03:11,753
You don't have that money, but
because you pay a little bit of

57
00:03:11,753 --> 00:03:14,944
money to the insurance company
every single month, they're

58
00:03:14,944 --> 00:03:18,026
going to take care of that.
Okay, and what the insurance

59
00:03:18,026 --> 00:03:21,493
company does, they pull a lot of
money from a lot of people and

60
00:03:21,493 --> 00:03:24,464
then they redistribute this
capital to when the claims

61
00:03:24,464 --> 00:03:27,380
happen. Same applies to
businesses. As a business you

62
00:03:27,380 --> 00:03:30,847
pay money every month or every
single year and when a big claim

63
00:03:30,847 --> 00:03:33,819
happens, big accident happens
where you are liable for

64
00:03:33,819 --> 00:03:37,230
millions of dollars, insurance
companies will pay for that. So

65
00:03:37,230 --> 00:03:40,312
what does it do for you? In
exchange for a little bit of

66
00:03:40,312 --> 00:03:43,724
money that you pay on a month of
a year basis, you potentially

67
00:03:43,724 --> 00:03:46,860
get more money. You're self
protected against millions of

68
00:03:46,860 --> 00:03:50,052
dollars of liability, and that
will make sure you don't go

69
00:03:50,052 --> 00:03:53,573
bankrupt. That also satisfies a
lot of contractual requirements.

70
00:03:53,573 --> 00:03:56,930
If you try to work with other
people, because it's the safety

71
00:03:56,930 --> 00:04:00,287
net that people prefer to have
in place. And in general, as a

72
00:04:00,287 --> 00:04:03,753
business owner, it's a peace of
mind. if something bad happens,

73
00:04:03,753 --> 00:04:07,000
I know somebody got me. So
that's what it is fundamentally.

74
00:04:07,000 --> 00:04:10,594
Sergey Litvinenko: So what we
do, this, but for robotics.

75
00:04:10,594 --> 00:04:14,251
Companies and for the robotics
industry in general. And we

76
00:04:14,251 --> 00:04:17,970
ensure all kinds of companies
that, sell robots or at least

77
00:04:17,970 --> 00:04:21,627
robots or provide robotics as a
service. So think of them,

78
00:04:21,627 --> 00:04:25,473
anywhere between like day one
operations that are about to do

79
00:04:25,473 --> 00:04:29,256
their first pilot all the way to
companies that already sell

80
00:04:29,256 --> 00:04:32,913
industrial robots to many
customers all over the world. So

81
00:04:32,913 --> 00:04:36,885
that's who we're dealing with.
And, Okay, that's step one, just

82
00:04:36,885 --> 00:04:40,668
to set the stage. Step two, why
we're doing this? we believe

83
00:04:40,668 --> 00:04:44,577
that the robotics industry is
the next great frontier in terms

84
00:04:44,577 --> 00:04:48,486
of what's going to happen with
our world. And it's going to be

85
00:04:48,486 --> 00:04:51,764
a physical makeover over the
next couple of decades,

86
00:04:51,764 --> 00:04:55,547
hopefully sooner rather than
later. And when you think about

87
00:04:55,547 --> 00:04:59,456
this, there are a lot of great
companies and great people that

88
00:04:59,456 --> 00:05:02,987
build those robots and the
systems and the services. And

89
00:05:02,987 --> 00:05:06,706
there is this layer that goes on
top, which has to do with.

90
00:05:06,706 --> 00:05:10,805
Financial products and services,
specifically equipment financing

91
00:05:10,805 --> 00:05:14,398
or robot financing, warranties,
and also insurance. Okay.

92
00:05:14,398 --> 00:05:17,803
Insurance is one of the
financial products. And if you

93
00:05:17,803 --> 00:05:21,334
think about it, it's hard to
adopt robots if they cost a

94
00:05:21,334 --> 00:05:25,116
fortune or if they don't come
with a warranty and they break

95
00:05:25,116 --> 00:05:28,773
and you lose your money or if
they make a mistake and then

96
00:05:28,773 --> 00:05:32,367
you're liable for millions of
dollars. So those financial

97
00:05:32,367 --> 00:05:35,898
products, they are really
important for the adoption and

98
00:05:35,898 --> 00:05:39,302
that's why we do it. So
insurance is one of the things

99
00:05:39,302 --> 00:05:42,896
that you need to have in place
in order to make sure that

100
00:05:42,896 --> 00:05:46,427
People buy your robots and the
businesses installed more

101
00:05:46,427 --> 00:05:50,525
systems and that when people ask
you questions what if your robot

102
00:05:50,525 --> 00:05:54,119
makes a mistake and then I
cannot operate my business you

103
00:05:54,119 --> 00:05:57,523
tell them that We have the
insurance in place. They're

104
00:05:57,523 --> 00:06:01,495
going to give you the payout and
That's why you shouldn't be it

105
00:06:01,495 --> 00:06:05,152
shouldn't be a problem for you
to go and start adopting My

106
00:06:05,152 --> 00:06:08,683
particular system or similar
systems. So that's the why,

107
00:06:08,683 --> 00:06:12,466
that's the peace of mind. That's
that level of friction that

108
00:06:12,466 --> 00:06:16,311
we're trying to, take care of.
And, in general, we think that

109
00:06:16,311 --> 00:06:20,094
insurance at the end of the day
will save a lot of companies

110
00:06:20,094 --> 00:06:23,877
from having a bad taste about
robotics. especially as robots

111
00:06:23,877 --> 00:06:27,975
become more and more ubiquitous.
So that's the why. The why is at

112
00:06:27,975 --> 00:06:31,821
the end of the day. Get more
robots into the world, help them

113
00:06:31,821 --> 00:06:35,541
commercialize and do this at the
best cost possible to both

114
00:06:35,541 --> 00:06:38,000
robotics companies and their
customers.

115
00:06:38,000 --> 00:06:41,000
Audrow Nash: Yeah, that makes
sense.

116
00:06:41,000 --> 00:06:45,474
Audrow Nash: Now, how did you
come to this problem? this seems

117
00:06:45,474 --> 00:06:49,371
like a super necessary thing,
but as so I am primarily

118
00:06:49,371 --> 00:06:54,062
involved in the technology. I, I
just recently bought a house and

119
00:06:54,062 --> 00:06:58,247
that involves like insurance and
things like this umbrella

120
00:06:58,247 --> 00:07:02,722
insurance, but it's not like a
thing I often think about. so I

121
00:07:02,722 --> 00:07:06,763
would like, how, what is your
path been? and how did you

122
00:07:06,763 --> 00:07:09,000
realize this is such a problem?

123
00:07:09,000 --> 00:07:12,954
Sergey Litvinenko: The path has
been very non trivial, to say

124
00:07:12,954 --> 00:07:16,130
the least. so yeah, we
originally started with an

125
00:07:16,130 --> 00:07:20,278
extreme focus on the autonomous
vehicle space. And our original,

126
00:07:20,278 --> 00:07:24,426
the original problem that we're
trying to solve had, didn't have

127
00:07:24,426 --> 00:07:28,250
to do anything with insurance.
So when we started, we said,

128
00:07:28,250 --> 00:07:32,398
Hey, We made an assumption that
autonomous vehicles are going to

129
00:07:32,398 --> 00:07:36,481
happen in the near future. And,
back in 2020, it was late 2019,

130
00:07:36,481 --> 00:07:40,306
early 2020. it seemed like it
was going to happen. We had a

131
00:07:40,306 --> 00:07:44,389
five year horizon. And we said,
look, autonomous vehicles, like

132
00:07:44,389 --> 00:07:47,630
the companies behind the
autonomous vehicles, they

133
00:07:47,630 --> 00:07:51,000
operate, in like in a closed
loop. Okay, so whatever

134
00:07:51,000 --> 00:07:53,000
Audrow Nash: do you mean? Oh,

135
00:07:53,000 --> 00:07:56,389
Sergey Litvinenko: So yeah, it's
like whatever they data they

136
00:07:56,389 --> 00:07:59,500
produce, they use it for
training and then they ship the

137
00:07:59,500 --> 00:08:02,944
updates. It doesn't really go
anywhere else. Okay. And when we

138
00:08:02,944 --> 00:08:05,722
thought about, they have to
interact with a lot of

139
00:08:05,722 --> 00:08:08,444
stakeholders, with the
government, with insurance

140
00:08:08,444 --> 00:08:11,722
companies, with public, policy
making organizations and the

141
00:08:11,722 --> 00:08:15,111
like. And we're like, at some
point there has to be a company

142
00:08:15,111 --> 00:08:18,500
that will make it very easy. To
take the data from autonomous

143
00:08:18,500 --> 00:08:21,556
vehicle companies and put it
into the hands of external

144
00:08:21,556 --> 00:08:24,500
stakeholders. So for example,
let's say if you are an

145
00:08:24,500 --> 00:08:27,944
autonomous vehicle company and
you try to report to NHTSA, the

146
00:08:27,944 --> 00:08:29,000
way it's done right

147
00:08:29,000 --> 00:08:31,000
Audrow Nash: What is NHTSA?

148
00:08:31,000 --> 00:08:35,109
Sergey Litvinenko: the national
highway safety organization. So

149
00:08:35,109 --> 00:08:38,826
it's the comp, it's the, it's
the governmental body that,

150
00:08:38,826 --> 00:08:42,609
pretty much overlooks anything
that has to do with traffic

151
00:08:42,609 --> 00:08:43,000
safety

152
00:08:43,000 --> 00:08:45,000
Audrow Nash: Yeah. Okay.

153
00:08:45,000 --> 00:08:48,331
Sergey Litvinenko: and, so for
example, Autonomous vehicle

154
00:08:48,331 --> 00:08:51,374
companies, be those pure play
level four robotaxis or

155
00:08:51,374 --> 00:08:54,762
partially automated, systems,
they have to report to NHTSA.

156
00:08:54,762 --> 00:08:57,805
And there are certain
requirements, there are certain

157
00:08:57,805 --> 00:09:01,366
fields of data that they have to
show them. And right now it's

158
00:09:01,366 --> 00:09:04,754
done in a very manual way.
you've literally just pulled the

159
00:09:04,754 --> 00:09:08,256
data, a bunch of, Excel word
files, you put it into the cloud

160
00:09:08,256 --> 00:09:11,644
and then you ship it to them.
And we thought, what if there

161
00:09:11,644 --> 00:09:15,147
were a streamlined way where you
can just, define the fields,

162
00:09:15,147 --> 00:09:18,822
click the button, and then we're
going to build the API and pull

163
00:09:18,822 --> 00:09:22,325
all the fields, of course, with
the permission of the company

164
00:09:22,325 --> 00:09:25,770
and do this like seamlessly.
That was the original idea. And

165
00:09:25,770 --> 00:09:29,216
one of the use cases that we had
was, insurance and we said,

166
00:09:29,216 --> 00:09:32,833
wouldn't it be awesome for the
insurance company who are trying

167
00:09:32,833 --> 00:09:36,394
to insure autonomous vehicles to
just go to, a tool that we're

168
00:09:36,394 --> 00:09:39,954
going to sell them, click one
button, and they're going to see

169
00:09:39,954 --> 00:09:43,399
all their data, the need for
underwriting. that was correct.

170
00:09:43,399 --> 00:09:47,074
But when we tried to sell that,
it didn't work out. Why? Because

171
00:09:47,074 --> 00:09:50,405
insurance companies had two
problems. Number one, they had

172
00:09:50,405 --> 00:09:53,678
no idea which data to ask, and
number two, there were not

173
00:09:53,678 --> 00:09:56,894
enough autonomous vehicles on
the road for them to make,

174
00:09:56,894 --> 00:10:00,282
commercial sense out of it.
that's why they dismissed this.

175
00:10:00,282 --> 00:10:03,727
And for us, it was a signal
like, okay, what are we going to

176
00:10:03,727 --> 00:10:07,459
do? And then we made a decision,
that, Hey, if we know which data

177
00:10:07,459 --> 00:10:11,077
we can pull and we know how to
pull this data, and we, like our

178
00:10:11,077 --> 00:10:14,752
specialty was working with data
in the first place, why don't we

179
00:10:14,752 --> 00:10:18,197
become an underwriter or an
insurance company ourselves? And

180
00:10:18,197 --> 00:10:21,873
that's, we'll make a decision to
go from Koop to Koop insurance.

181
00:10:21,873 --> 00:10:25,261
And we went through all the
hoops that you have to go over,

182
00:10:25,261 --> 00:10:28,419
like getting licenses, getting,
relationships with like

183
00:10:28,419 --> 00:10:31,692
reinsurance companies and stuff
like that, and we started

184
00:10:31,692 --> 00:10:35,195
underwriting autonomous vehicles
ourselves. And shortly after

185
00:10:35,195 --> 00:10:38,755
we've done that. We, we said,
Hey, we can do the similar thing

186
00:10:38,755 --> 00:10:42,373
to other robotics use cases. we
can do mobile robots. We can do

187
00:10:42,373 --> 00:10:45,301
drones. We can do automated
vessels. we even insure

188
00:10:45,301 --> 00:10:48,000
submarines today, believe it or
not, autonomous

189
00:10:48,000 --> 00:10:49,000
Audrow Nash: Super cool.

190
00:10:49,000 --> 00:10:52,482
Sergey Litvinenko: And, that's
how we got to where we are

191
00:10:52,482 --> 00:10:56,331
today. And, when I say the path
was non trivial is that when we

192
00:10:56,331 --> 00:10:59,936
originally started, we didn't
envision that we would become

193
00:10:59,936 --> 00:11:03,601
like a I would like to brag
about it, but I think we are the

194
00:11:03,601 --> 00:11:07,328
number one robotics underwriter
in the US today, maybe in the

195
00:11:07,328 --> 00:11:08,000
world. And,

196
00:11:08,000 --> 00:11:11,000
Audrow Nash: know other ones. So
hi. Are there other ones?

197
00:11:11,000 --> 00:11:14,500
Sergey Litvinenko: we typically
compete with like bigger

198
00:11:14,500 --> 00:11:16,000
insurance companies that

199
00:11:16,000 --> 00:11:18,000
Audrow Nash: Yeah, but they're
so slow

200
00:11:18,000 --> 00:11:21,478
Sergey Litvinenko: robotics
companies. But yeah, it's, not a

201
00:11:21,478 --> 00:11:24,957
systemic competition. Yeah. so
yeah, that's how we got here.

202
00:11:24,957 --> 00:11:28,377
And then, we are a function of
the robotics industry. So as

203
00:11:28,377 --> 00:11:31,971
long, as the robotics industry
thrives, we're going to thrive.

204
00:11:31,971 --> 00:11:35,449
And we're trying to do a really
good job to make the life of

205
00:11:35,449 --> 00:11:38,000
robotics companies as easy as
possible. Yes.

206
00:11:38,000 --> 00:11:42,098
Audrow Nash: thing to be
explicit about with the business

207
00:11:42,098 --> 00:11:46,555
model if I understand correctly,
so you are underwriting which

208
00:11:46,555 --> 00:11:50,580
means that you are estimating
the risk Overall for these

209
00:11:50,580 --> 00:11:54,678
companies so because basically
the way that it works if I

210
00:11:54,678 --> 00:11:59,135
understand is you say for this
many companies that we have, or

211
00:11:59,135 --> 00:12:03,232
this many robots deployed of
this type, there's a risk of

212
00:12:03,232 --> 00:12:07,761
this much, like we have a risk
of this much, and the payout for

213
00:12:07,761 --> 00:12:12,218
that is this much. And so then
what you have to do is you have

214
00:12:12,218 --> 00:12:16,747
to multiply all that together,
basically, so you can try to get

215
00:12:16,747 --> 00:12:21,132
the expected risk for the whole
fleet of robots. And then you

216
00:12:21,132 --> 00:12:25,445
have to charge just a little bit
more than that, so that you

217
00:12:25,445 --> 00:12:29,974
probably need a buffer too. and
then you have to have enough to

218
00:12:29,974 --> 00:12:34,000
pay yourselves. And that's, the
business model, correct?

219
00:12:34,000 --> 00:12:37,451
Sergey Litvinenko: You are
correct. And you pretty much

220
00:12:37,451 --> 00:12:41,530
with what you described, you can
possibly qualify to be, to be an

221
00:12:41,530 --> 00:12:45,546
insurance person. So think about
it this way. It's what you just

222
00:12:45,546 --> 00:12:49,249
described is called frequency
and severity. And that's how,

223
00:12:49,249 --> 00:12:53,139
that's what insurance comes down
to. It doesn't matter whether

224
00:12:53,139 --> 00:12:56,465
it's robotics or anything else.
Frequency, what's the

225
00:12:56,465 --> 00:13:00,356
probability a bad thing is going
to happen. Severity, if a bad

226
00:13:00,356 --> 00:13:04,121
thing happens, how bad is it
going to be? Okay. Of course, a

227
00:13:04,121 --> 00:13:07,886
lot goes into both in terms of
how much data and statistical

228
00:13:07,886 --> 00:13:11,526
analysis, but that's what it
comes down to. And then there

229
00:13:11,526 --> 00:13:15,040
are a bunch of rules like
underwriting rules. It's okay,

230
00:13:15,040 --> 00:13:18,428
we're not going to touch
companies that are too big or

231
00:13:18,428 --> 00:13:22,131
that are too small or that do
international operations. But

232
00:13:22,131 --> 00:13:25,582
all of that is more like Rule
based system that details

233
00:13:25,582 --> 00:13:29,284
exactly. yep, you're correct,
and that's exactly what we do

234
00:13:29,284 --> 00:13:33,050
for robots. For every single
company, when we evaluate them,

235
00:13:33,050 --> 00:13:36,375
we evaluate, potential
frequency, potential severity,

236
00:13:36,375 --> 00:13:40,140
multiply one by the another,
that gives you a base rate, and

237
00:13:40,140 --> 00:13:43,780
then on top, you make, the,
percentage for the company to,

238
00:13:43,780 --> 00:13:47,545
to operate. yep, you're 100
percent on the spot. And we, we,

239
00:13:47,545 --> 00:13:51,499
also buy reinsurance. what does
it mean? for example, let's say

240
00:13:51,499 --> 00:13:55,264
you take frequency multiplied by
severity. It ends up being,

241
00:13:55,264 --> 00:13:59,092
let's say, 10, 000 for a year.
Okay? And then we, it includes

242
00:13:59,092 --> 00:14:03,045
some commercial considerations
for Koop. And then what we would

243
00:14:03,045 --> 00:14:06,685
do, we would collect that
premium. And then we would go to

244
00:14:06,685 --> 00:14:10,450
reinsurance and say, Hey, we're
going to give you a certain.

245
00:14:10,450 --> 00:14:13,838
percentage of what we,
collected. So you can re insure

246
00:14:13,838 --> 00:14:17,854
us. And that means that, if, we,
let's say if Koop is wrong with

247
00:14:17,854 --> 00:14:21,431
its underwriting model, and
let's say we have a big loss,

248
00:14:21,431 --> 00:14:23,000
then they would cover the

249
00:14:23,000 --> 00:14:25,000
Audrow Nash: for insurance. They
back you up.

250
00:14:25,000 --> 00:14:28,735
Sergey Litvinenko: And we do
that because it's a, in general,

251
00:14:28,735 --> 00:14:32,469
just a good practice until you
have a lot of money on balance

252
00:14:32,469 --> 00:14:35,776
sheet, you typically do
reinsurance. So that's what we

253
00:14:35,776 --> 00:14:39,571
do. And then we're left with,
whatever percentage is remaining

254
00:14:39,571 --> 00:14:43,000
as our, revenue, and then, it
goes down the P and L from

255
00:14:43,000 --> 00:14:46,150
Audrow Nash: Yeah. I feel like
it would be ironic if you didn't

256
00:14:46,150 --> 00:14:49,300
do reinsurance because it's like
you don't leave in the process

257
00:14:49,300 --> 00:14:50,000
in some sense.

258
00:14:50,000 --> 00:14:53,000
Sergey Litvinenko: Yeah. Yeah.

259
00:14:53,000 --> 00:14:57,436
Audrow Nash: Okay, yeah, and so
reinsurance in the event that

260
00:14:57,436 --> 00:15:02,018
something huge happens, and that
payout is much higher than the

261
00:15:02,018 --> 00:15:06,309
severity you expected, so then
it's too much cash from your

262
00:15:06,309 --> 00:15:10,309
perspective, you guys are
insured by a bigger insurance

263
00:15:10,309 --> 00:15:13,000
company for this kind of thing.
Okay,

264
00:15:13,000 --> 00:15:16,955
Sergey Litvinenko: Yep. So
that's how, it works. And, it's

265
00:15:16,955 --> 00:15:21,182
just a, it's a, it's an industry
standard until you become too

266
00:15:21,182 --> 00:15:22,000
big to fail,

267
00:15:22,000 --> 00:15:26,462
Audrow Nash: yeah. Yeah, too big
to fail. And then but one thing

268
00:15:26,462 --> 00:15:30,923
that is clever, I think is, and
I remember from talking earlier,

269
00:15:30,923 --> 00:15:34,618
I think I remember from talking
earlier, you guys are

270
00:15:34,618 --> 00:15:38,731
effectively a brokerage for
insurance. Is it true? Or is it

271
00:15:38,731 --> 00:15:42,983
like underwriting? You're not,
you're estimating the risk for

272
00:15:42,983 --> 00:15:46,817
some other guys who are
insuring, them, or are you guys

273
00:15:46,817 --> 00:15:50,651
doing the full insurance
yourself? Or how does all that

274
00:15:50,651 --> 00:15:51,000
work?

275
00:15:51,000 --> 00:15:54,513
Sergey Litvinenko: Yep. So
here's how it works. We act as

276
00:15:54,513 --> 00:15:58,149
an underwriter. So that means
that when we have a customer,

277
00:15:58,149 --> 00:16:01,970
we'll look at their data. we do
the pricing, we do the, rules,

278
00:16:01,970 --> 00:16:05,421
we apply rules to them,
underwriting rules, and then, we

279
00:16:05,421 --> 00:16:09,057
also do the whole transaction
from start to finish. We also

280
00:16:09,057 --> 00:16:12,878
are responsible for acquiring
the customer and maintaining the

281
00:16:12,878 --> 00:16:15,836
relationship. We are the
underwriter. Now we not

282
00:16:15,836 --> 00:16:19,657
underwriting everything. We're
underwriting only a part of the

283
00:16:19,657 --> 00:16:23,355
company. So what does it mean?
We do a general liability, or

284
00:16:23,355 --> 00:16:26,991
it's also known as product
liability, just for the company.

285
00:16:26,991 --> 00:16:30,750
We'll also do robotics errors
and emissions for your fleet of

286
00:16:30,750 --> 00:16:34,509
robots that you operate as a
service. And we'll also do cyber

287
00:16:34,509 --> 00:16:37,837
liability. In case you, your
robots get hacked or your

288
00:16:37,837 --> 00:16:41,597
company gets hacked. That's what
we under, oh, and we also of

289
00:16:41,597 --> 00:16:45,171
course do commercial auto. So
commercial auto it's, if you

290
00:16:45,171 --> 00:16:48,622
have, fully, or partially
automated vehicles on the road

291
00:16:48,622 --> 00:16:51,888
in a commercial fleet, we
underwrite that as well. So

292
00:16:51,888 --> 00:16:55,586
that's what we underwrite. We
also have customers who, let's

293
00:16:55,586 --> 00:16:59,160
say, come to Koop with other
needs. We call them non core,

294
00:16:59,160 --> 00:17:02,920
non core needs. And that could
be like directors and officers

295
00:17:02,920 --> 00:17:06,679
to protect your management. it
could be employment, practices

296
00:17:06,679 --> 00:17:10,500
liability. It could be workers
compensation liability. This is

297
00:17:10,500 --> 00:17:14,013
something that we do not
underwrite because it's a pretty

298
00:17:14,013 --> 00:17:17,710
straightforward how it's done.
That's what we're going to go

299
00:17:17,710 --> 00:17:21,470
and partner with somebody else
to take care of that. But it's

300
00:17:21,470 --> 00:17:25,229
going to be done through Koop.
Okay, so in that sense, we are

301
00:17:25,229 --> 00:17:29,050
brokerage for whatever is not
our specialty, and we underwrite

302
00:17:29,050 --> 00:17:32,809
the specialty lines of, lines of
insurance, which are product

303
00:17:32,809 --> 00:17:36,384
liability, errors and omissions,
cyber liability, and auto

304
00:17:36,384 --> 00:17:37,000
liability.

305
00:17:37,000 --> 00:17:41,923
Audrow Nash: awesome. Okay. And
so with all this, so that, to

306
00:17:41,923 --> 00:17:46,604
me, it all is very clear now,
which is great. How the main

307
00:17:46,604 --> 00:17:51,365
challenge that I see is now you
need a very accurate way of

308
00:17:51,365 --> 00:17:56,368
assessing risk because you could
be super conservative and you

309
00:17:56,368 --> 00:18:00,000
could charge your customers an
arm and a leg.

310
00:18:00,000 --> 00:18:03,714
Audrow Nash: but that's not
helping the companies and you're

311
00:18:03,714 --> 00:18:07,676
not as competitive as you could
be or you could be super minimal

312
00:18:07,676 --> 00:18:11,700
and if you're super minimal then
you guys go out of business real

313
00:18:11,700 --> 00:18:15,352
soon because you don't have
enough to do the payouts so how

314
00:18:15,352 --> 00:18:19,005
do you figure out a good risk
and the payouts especially in

315
00:18:19,005 --> 00:18:22,038
these novel spaces like
submarines and autonomous

316
00:18:22,038 --> 00:18:25,814
vehicles and all these kinds of
things how do you figure this

317
00:18:25,814 --> 00:18:26,000
out

318
00:18:26,000 --> 00:18:29,891
Sergey Litvinenko: Yep. So we're
pretty much built a model and,

319
00:18:29,891 --> 00:18:33,659
that gives us an estimate of how
much we charge. And to check

320
00:18:33,659 --> 00:18:37,612
ourselves, we compare that model
to the market benchmarks and we

321
00:18:37,612 --> 00:18:41,257
see how much, how different we
are. And I can tell you that

322
00:18:41,257 --> 00:18:45,210
sometimes we are lower than the
market, significantly lower than

323
00:18:45,210 --> 00:18:48,792
the market. Because the data
shows that it's a really good

324
00:18:48,792 --> 00:18:52,684
risk. And there is a good, both
on a quantitative side in terms

325
00:18:52,684 --> 00:18:55,772
of what we collect from the
fleet, and also on the

326
00:18:55,772 --> 00:18:59,416
qualitative side. Sometimes we
have customers who we charge

327
00:18:59,416 --> 00:19:02,875
over the market because we've
figured something out that

328
00:19:02,875 --> 00:19:06,767
they're not as great of a risk
and we need to charge them more.

329
00:19:06,767 --> 00:19:10,473
And, it's, what's. It's this,
it's, how they are and it's up

330
00:19:10,473 --> 00:19:13,870
to them whether they want to
accept that or not, so our

331
00:19:13,870 --> 00:19:17,638
model, there are two components
to our model. number one, the

332
00:19:17,638 --> 00:19:21,406
quantitative component is that
the API that I told you about,

333
00:19:21,406 --> 00:19:25,359
we use that API to collect data
directly from different kinds of

334
00:19:25,359 --> 00:19:29,188
robots. And when we collect the
data, our methodology is based

335
00:19:29,188 --> 00:19:32,524
on understanding A behavior of a
robot in a particular

336
00:19:32,524 --> 00:19:36,168
environment, and then we'll use
the behavior as a proxy for

337
00:19:36,168 --> 00:19:39,936
risk, as a proxy for an accident
that can lead to a financial

338
00:19:39,936 --> 00:19:43,889
loss, an accident that can lead
to a claim. And the idea is very

339
00:19:43,889 --> 00:19:47,719
simple, actually. The better the
performance of the robot, the

340
00:19:47,719 --> 00:19:51,486
better the behavior, the lower
the probability that something

341
00:19:51,486 --> 00:19:54,884
bad is going to happen. And
probably the most startling

342
00:19:54,884 --> 00:19:58,713
example that we have would be,
Auto, so automobiles, we have a

343
00:19:58,713 --> 00:20:02,666
ocean of data when it comes to,
human driving performance on the

344
00:20:02,666 --> 00:20:06,311
road and how it correlates to
accidents, which I believe in

345
00:20:06,311 --> 00:20:10,202
2023, there was 40, 000 deaths.
On, public roads in the U S and

346
00:20:10,202 --> 00:20:13,476
there were a million more
accidents, that led to some

347
00:20:13,476 --> 00:20:17,182
injury or damage. And there was
a lot of bad behavior on the

348
00:20:17,182 --> 00:20:21,011
road in general. So we have an
ocean of data. So from the data

349
00:20:21,011 --> 00:20:24,161
that we collected from
autonomous vehicle companies

350
00:20:24,161 --> 00:20:27,806
that we work with, and that
includes different use cases on

351
00:20:27,806 --> 00:20:31,697
public roads, robo taxis, vans,
trucks, shuttles, we're already

352
00:20:31,697 --> 00:20:35,465
seeing that autonomous vehicles
are behaving up to 70 percent

353
00:20:35,465 --> 00:20:39,480
better than humans. So what does
it mean? What does it mean? That

354
00:20:39,480 --> 00:20:43,248
means that, there are, there is
like a five to six common bad

355
00:20:43,248 --> 00:20:47,077
behavior examples that tend to
lead to accidents. For example,

356
00:20:47,077 --> 00:20:50,536
excessive speeding, hard
stopping, risky lane maneuvers,

357
00:20:50,536 --> 00:20:53,872
running the red light and stuff
like that. we see that

358
00:20:53,872 --> 00:20:57,763
autonomous vehicles just don't
do that. Even from the data that

359
00:20:57,763 --> 00:21:01,345
Koop has, 70 percent reduction
in risky maneuvers, okay? I

360
00:21:01,345 --> 00:21:04,804
think it's going to be 99
percent eventually. And that's

361
00:21:04,804 --> 00:21:08,758
what's driving, the quantitative
assessment that we have. That's

362
00:21:08,758 --> 00:21:12,711
what's driving the frequency. we
have a way how we say, based on

363
00:21:12,711 --> 00:21:16,293
that. We can estimate that the
frequency of an accident is

364
00:21:16,293 --> 00:21:20,061
going to be lower. now that's on
the qualitative side, on the

365
00:21:20,061 --> 00:21:23,952
qualitative side, every single
company that we work with, we of

366
00:21:23,952 --> 00:21:27,597
course evaluate how good they
are from the risk standpoint,

367
00:21:27,597 --> 00:21:31,612
and we have a certain, Technical
questionnaire that we developed.

368
00:21:31,612 --> 00:21:35,194
We asked them about their
software practices, what kind of

369
00:21:35,194 --> 00:21:39,085
hardware they have, what kind of
safety methodologies that they

370
00:21:39,085 --> 00:21:42,544
use, what kind of compliance
they have in place, and the

371
00:21:42,544 --> 00:21:46,374
companies that did more stuff,
we assume that they're going to

372
00:21:46,374 --> 00:21:50,080
have, less of a chance that
something bad is going to happen

373
00:21:50,080 --> 00:21:53,477
to the fleet or to the
organization. And that also goes

374
00:21:53,477 --> 00:21:57,060
into our underwriting. So once
we do both quantitative and

375
00:21:57,060 --> 00:22:00,828
qualitative, combine it together
and everything comes down to

376
00:22:00,828 --> 00:22:04,534
frequency and severity, and we
have the rate. We look, okay.

377
00:22:04,534 --> 00:22:08,178
This is how much we're going to
charge them and this is the

378
00:22:08,178 --> 00:22:12,193
market benchmark and we have the
access to market benchmarks. And

379
00:22:12,193 --> 00:22:16,146
most of the time we are charging
less than the market because we

380
00:22:16,146 --> 00:22:19,976
have the methodology that says
that this is a better risk. But

381
00:22:19,976 --> 00:22:23,805
most of the time, by most of the
time, 85 percent of the time.

382
00:22:23,805 --> 00:22:27,820
There are also 15 percent of the
time we see companies that could

383
00:22:27,820 --> 00:22:30,600
do a better job and we also
provide them with

384
00:22:30,600 --> 00:22:34,429
recommendations what they can
do. But unfortunately we have to

385
00:22:34,429 --> 00:22:38,135
overcharge them because they
are. A riskier company than the

386
00:22:38,135 --> 00:22:39,000
market thinks.

387
00:22:39,000 --> 00:22:41,897
Audrow Nash: Yep. And you have
to do the smart risk estimation.

388
00:22:41,897 --> 00:22:43,000
And so that makes sense.

389
00:22:43,000 --> 00:22:45,891
Sergey Litvinenko: yep. And then
if there isn't any other

390
00:22:45,891 --> 00:22:49,087
insurance company that wants to
take them, sure. But we figured

391
00:22:49,087 --> 00:22:51,877
that they are riskier than
people believe. And maybe we

392
00:22:51,877 --> 00:22:55,022
shouldn't insure them, because
the probability of having a bad

393
00:22:55,022 --> 00:22:57,000
accident is higher than
everybody else.

394
00:22:57,000 --> 00:23:01,218
Audrow Nash: Ah, okay. That's
pretty cool. So that makes good

395
00:23:01,218 --> 00:23:04,883
sense to me for autonomous
vehicles because it's cool

396
00:23:04,883 --> 00:23:09,170
because you, can become domain
experts to some degree, and you

397
00:23:09,170 --> 00:23:12,904
can look at behaviors and you
can use that to estimate

398
00:23:12,904 --> 00:23:17,399
probability of accidents. So you
can look at these indicators you

399
00:23:17,399 --> 00:23:20,926
said, stopping too quickly,
reckless lane changing,

400
00:23:20,926 --> 00:23:25,213
whatever. And you can see that
from the data. If they're doing

401
00:23:25,213 --> 00:23:29,154
that, and then you can say,
okay, we're never doing that,

402
00:23:29,154 --> 00:23:33,511
thus we're not creating a bunch
of preconditions for accidents,

403
00:23:33,511 --> 00:23:37,660
and then we're not having as
many accidents. Okay, so that's

404
00:23:37,660 --> 00:23:41,947
great. how do you do this for
Unknown areas like submarines or

405
00:23:41,947 --> 00:23:46,372
robots in hotels or things like
this like how do you because the

406
00:23:46,372 --> 00:23:50,314
thing is what strikes me is with
this we have the Highway

407
00:23:50,314 --> 00:23:54,670
Association you were mentioning
and they have and the insurance

408
00:23:54,670 --> 00:23:59,096
industry is really interested in
figuring out how to What in the

409
00:23:59,096 --> 00:24:02,899
vehicle space, not just
autonomous, but people are risk

410
00:24:02,899 --> 00:24:07,048
indicators so they can insure
more appropriately. but how do

411
00:24:07,048 --> 00:24:11,473
you do this in a new space that
doesn't have such a large amount

412
00:24:11,473 --> 00:24:15,000
of established information about
what creates risk?

413
00:24:15,000 --> 00:24:18,336
Sergey Litvinenko: Yeah, there
is a, trick that we do and

414
00:24:18,336 --> 00:24:21,906
you're completely right that if
you have a mobile robot, that

415
00:24:21,906 --> 00:24:25,417
transports, let's say, goods or
materials in a warehouse and

416
00:24:25,417 --> 00:24:28,870
it's not exposed to a lot of
people and there was no public

417
00:24:28,870 --> 00:24:31,855
data in terms of what's
happening with within those

418
00:24:31,855 --> 00:24:35,542
warehouses, what we're going to
do, or submarine or a drone, So

419
00:24:35,542 --> 00:24:38,410
we, we apply the same
methodology. So we'll still

420
00:24:38,410 --> 00:24:41,512
collect the data on, it's not
going to be called. Bad

421
00:24:41,512 --> 00:24:45,023
behavior, we'll call it bad
behavior, but we call it like a,

422
00:24:45,023 --> 00:24:48,652
what's the troubleshooting rate?
How many times a vehicle goes

423
00:24:48,652 --> 00:24:52,280
offline or has to do a minimal
risk maneuver or has to reach a

424
00:24:52,280 --> 00:24:55,441
minimal risk condition? And
every single company has a

425
00:24:55,441 --> 00:24:59,011
different definition of that,
and it's our job to standardize

426
00:24:59,011 --> 00:25:02,464
that. But it's pretty much.
Like, how often your robot goes

427
00:25:02,464 --> 00:25:06,092
offline because it can, cannot
do something. That's the, proxy

428
00:25:06,092 --> 00:25:09,604
that we take. And then we also
evaluate the environment. And

429
00:25:09,604 --> 00:25:13,232
then this says, okay, if there
is an environment where there's

430
00:25:13,232 --> 00:25:16,861
literally nobody around, what
does it mean? That means that we

431
00:25:16,861 --> 00:25:20,255
cannot injure anybody and we
cannot damage anything. Okay,

432
00:25:20,255 --> 00:25:23,591
now what's the worst case
scenario that you can think of?

433
00:25:23,591 --> 00:25:27,044
the worst case scenario is if
you have a company relying on

434
00:25:27,044 --> 00:25:30,614
your drone for a service, if
that's, if your drone goes, goes

435
00:25:30,614 --> 00:25:33,950
offline, then, You cause
business interruption. Okay. And

436
00:25:33,950 --> 00:25:37,637
that means that the company is
going to lose money and there is

437
00:25:37,637 --> 00:25:41,090
a way how to evaluate that. So
when, we do that, we realize

438
00:25:41,090 --> 00:25:44,485
that, Hey, in those, simpler
environments where the robots

439
00:25:44,485 --> 00:25:47,996
are not, let's say, carrying any
people as passengers or not

440
00:25:47,996 --> 00:25:51,274
driving with other people on the
road, we would take the

441
00:25:51,274 --> 00:25:54,785
troubleshooting rate and we
would just assume what the worst

442
00:25:54,785 --> 00:25:58,179
case scenario could be. And most
of the time it's business

443
00:25:58,179 --> 00:26:01,691
interruption. And it's business
interruption up to a certain

444
00:26:01,691 --> 00:26:04,851
dollar limit, let's say a
million dollars, two million

445
00:26:04,851 --> 00:26:08,070
dollars, or three million
dollars, depending on what we

446
00:26:08,070 --> 00:26:11,699
put in the policy. And then we
say, there is a 0. 0001 percent

447
00:26:11,699 --> 00:26:15,269
chance that this robot is going
to go like completely offline

448
00:26:15,269 --> 00:26:19,014
and not recover. And it's going
to be like the redundant systems

449
00:26:19,014 --> 00:26:22,467
are also going to fail. And it's
going to be, let's say a 2

450
00:26:22,467 --> 00:26:26,213
million loss. We'll multiply one
by another. And of course, it's

451
00:26:26,213 --> 00:26:29,607
not as good of an assumption
that we make because there is

452
00:26:29,607 --> 00:26:32,826
less data. But that's how we do
it. and then the second

453
00:26:32,826 --> 00:26:36,630
challenge is there are no market
benchmarks, meaning I don't know

454
00:26:36,630 --> 00:26:40,083
how people charge for robots in
warehouses, it just doesn't

455
00:26:40,083 --> 00:26:43,361
exist. So we cannot compare that
against the market, but

456
00:26:43,361 --> 00:26:47,048
typically we can say that, hey,
if you have a robot that, let's

457
00:26:47,048 --> 00:26:50,267
say, if you have a fleet of
robots that costs a million

458
00:26:50,267 --> 00:26:53,544
dollars and you are paying,
let's say, 10, 000 to insure

459
00:26:53,544 --> 00:26:57,056
that on a yearly basis, The
economics make sense to both the

460
00:26:57,056 --> 00:27:00,684
producer or the developer of the
robot and the customer. So we

461
00:27:00,684 --> 00:27:04,254
usually use customers as like
the sanity check, whether that,

462
00:27:04,254 --> 00:27:07,649
that, that price makes sense.
And of course we reserve the

463
00:27:07,649 --> 00:27:11,336
right to do adjustments to the
rate that we charge based on the

464
00:27:11,336 --> 00:27:14,789
underwriting assessment on the
market conditions, but we do

465
00:27:14,789 --> 00:27:18,417
that rarely. most of the time
it's just programmatic. so yeah,

466
00:27:18,417 --> 00:27:21,753
that's how we do it. And I got
to tell you, it's the same

467
00:27:21,753 --> 00:27:24,972
methodology. It's more
assumptions that go into it, but

468
00:27:24,972 --> 00:27:28,308
at the end of the day, it does
make economic sense to the

469
00:27:28,308 --> 00:27:31,000
customers. So that's the most
important thing.

470
00:27:31,000 --> 00:27:35,228
Audrow Nash: Yeah. And also, if
you are making more assumptions,

471
00:27:35,228 --> 00:27:39,126
you are effectively not. You're
having more risk because of

472
00:27:39,126 --> 00:27:42,825
those assumptions, but you can
charge a higher margin, I

473
00:27:42,825 --> 00:27:46,392
imagine, and that, that's
probably how you make up for

474
00:27:46,392 --> 00:27:50,290
these riskier environments. So
if it's autonomous vehicles,

475
00:27:50,290 --> 00:27:54,320
maybe, I'm just picking numbers
and I have no idea, maybe you

476
00:27:54,320 --> 00:27:57,953
charge 1. 1. Times are 10
percent more than you need to

477
00:27:57,953 --> 00:28:01,719
because that's your margin
because lots of them are going

478
00:28:01,719 --> 00:28:05,881
and you it's a volume thing in a
sense. Whereas if you're doing

479
00:28:05,881 --> 00:28:09,646
warehouse robotics for a new
robotics company, maybe it's

480
00:28:09,646 --> 00:28:13,544
like 1. 5 or something like
that. And that's because of the

481
00:28:13,544 --> 00:28:17,706
risk that you can't estimate and
you give yourself a buffer for

482
00:28:17,706 --> 00:28:21,604
that kind of thing. Is that,
kind of correct or what do you

483
00:28:21,604 --> 00:28:22,000
think?

484
00:28:22,000 --> 00:28:26,253
Sergey Litvinenko: it is, it's
actually more, more manual than

485
00:28:26,253 --> 00:28:30,644
you think. So the equation that
we have, it spits out the rates,

486
00:28:30,644 --> 00:28:34,142
the tech, we'll call it the
technical rate. So it's

487
00:28:34,142 --> 00:28:38,464
literally frequency multiplied
by severity. And then there is a

488
00:28:38,464 --> 00:28:42,649
certain percentage that we put
on top as our margin, like you

489
00:28:42,649 --> 00:28:46,834
said, but it's not dependent on
a customer. It's the same for

490
00:28:46,834 --> 00:28:51,087
everybody. And, and then. If and
then there is an underwriting

491
00:28:51,087 --> 00:28:55,203
judgment involved and we have
underwriters on staff and they

492
00:28:55,203 --> 00:28:59,456
say let's say if you insure an
autonomous submarine that does,

493
00:28:59,456 --> 00:29:03,504
let's say, R&amp; D or something
else, and the loss of that

494
00:29:03,504 --> 00:29:07,551
submarine could be really bad
for the customer that did use

495
00:29:07,551 --> 00:29:12,011
that submarine, you can, you are
able to add a certain percentage

496
00:29:12,011 --> 00:29:14,000
on top as a margin of safety,

497
00:29:14,000 --> 00:29:17,000
Audrow Nash: It comes in at a
different stage, basically.

498
00:29:17,000 --> 00:29:20,652
Sergey Litvinenko: it's not, we,
do not have a model that detects

499
00:29:20,652 --> 00:29:23,798
It dictates that. It's more
rules based. Okay? So if you

500
00:29:23,798 --> 00:29:27,112
have a rule that you have a very
tough robotics use case to

501
00:29:27,112 --> 00:29:29,978
ensure, that deserves
underwriting review, and then

502
00:29:29,978 --> 00:29:33,348
the underwriter can decide to
increase the margin of safety,

503
00:29:33,348 --> 00:29:37,000
but the, it's not baked into the
equation. It's more rules based.

504
00:29:37,000 --> 00:29:41,144
Audrow Nash: Okay. So tell me
more about the rules based

505
00:29:41,144 --> 00:29:45,584
stuff. what it is. So to me,
initially they seemed like just

506
00:29:45,584 --> 00:29:50,097
details that may not be that
interesting, but I'm starting to

507
00:29:50,097 --> 00:29:53,945
think of it. it's like an
algorithm that you guys go

508
00:29:53,945 --> 00:29:58,459
through in order to It's it's
too complex to make a model. So

509
00:29:58,459 --> 00:30:02,825
they're like simple heuristics
that you follow, in order to

510
00:30:02,825 --> 00:30:07,486
make decisions that are around
risk. Is that how to think about

511
00:30:07,486 --> 00:30:12,000
the rules that you may have, or
tell me more about the rules?

512
00:30:12,000 --> 00:30:15,855
Sergey Litvinenko: Yep, so the
rules are the heuristics that

513
00:30:15,855 --> 00:30:19,709
you have to follow and this is
done by a human being. So for

514
00:30:19,709 --> 00:30:23,243
example, let's say you have a
robotics company and that

515
00:30:23,243 --> 00:30:27,226
robotics company, so what are
the rules that we apply to them?

516
00:30:27,226 --> 00:30:31,081
geographical locations, you have
to operate within a certain

517
00:30:31,081 --> 00:30:35,000
geography. Right now it's the U.
S. We are unable to do that.

518
00:30:35,000 --> 00:30:37,000
Audrow Nash: Too outside of it,

519
00:30:37,000 --> 00:30:39,578
Sergey Litvinenko: exactly,
number one. Number two, for

520
00:30:39,578 --> 00:30:42,203
certain use cases, we have
heuristics, whether you go on

521
00:30:42,203 --> 00:30:44,922
public roads, you do not go on
public roads, that actually

522
00:30:44,922 --> 00:30:46,000
makes a huge difference

523
00:30:46,000 --> 00:30:49,000
Audrow Nash: I would imagine.
Makes

524
00:30:49,000 --> 00:30:52,635
Sergey Litvinenko: insurance,
we'll have the heuristics for,

525
00:30:52,635 --> 00:30:56,453
the management team. what does
it mean? That's really important

526
00:30:56,453 --> 00:31:00,331
that you have a person dedicated
to either, safety or compliance

527
00:31:00,331 --> 00:31:04,148
on staff, and that's, it makes a
big difference, not just, from

528
00:31:04,148 --> 00:31:07,299
the checklist standpoint, but
also from the. Company

529
00:31:07,299 --> 00:31:11,055
standpoint, it's like, if you
have a person dedicated to this,

530
00:31:11,055 --> 00:31:14,812
we have more trust that you will
do the right thing when a bad

531
00:31:14,812 --> 00:31:18,447
thing happens. And, sometimes,
we saw examples with some big

532
00:31:18,447 --> 00:31:22,022
autonomous vehicle companies
when they didn't go, right. we

533
00:31:22,022 --> 00:31:25,658
have also rules based around,
just some financial metrics of

534
00:31:25,658 --> 00:31:29,414
the company, because that's what
we look at. and, It could be,

535
00:31:29,414 --> 00:31:33,232
when in our, we have an internal
system underwriting system and

536
00:31:33,232 --> 00:31:37,049
that internal system spits out
the rate, and then it also shows

537
00:31:37,049 --> 00:31:40,745
you certain pattern changes with
the account and it says, The

538
00:31:40,745 --> 00:31:44,259
company, let's say, tripled in
size, or it tripled in size

539
00:31:44,259 --> 00:31:47,713
while reducing the headcount, or
it did not grow, but the

540
00:31:47,713 --> 00:31:50,924
headcount grew by 50%. We have
those certain, they're

541
00:31:50,924 --> 00:31:54,620
relatively simple, but those
certain metrics, when they, when

542
00:31:54,620 --> 00:31:58,377
the underwriter looks at them,
it gives an underwriter an idea

543
00:31:58,377 --> 00:32:01,831
whether something is off. And
typically, if you are, that

544
00:32:01,831 --> 00:32:05,224
gives an underwriter an ability
to go. And either call a

545
00:32:05,224 --> 00:32:08,981
customer or have a conversation,
maybe over email and ask them

546
00:32:08,981 --> 00:32:12,798
about whether everything is all
right, because don't forget the

547
00:32:12,798 --> 00:32:16,312
tendency when you purchase
insurance is that, I don't want

548
00:32:16,312 --> 00:32:19,826
to show you any data, just give
me the quote, and there is

549
00:32:19,826 --> 00:32:23,280
always the fight between how
much data we can get and how

550
00:32:23,280 --> 00:32:26,734
good of a quote we can give.
That's why we built a system

551
00:32:26,734 --> 00:32:30,430
that spots the pattern changes
that don't have to do anything

552
00:32:30,430 --> 00:32:33,762
with robotics, more with the
company. And if there is a

553
00:32:33,762 --> 00:32:37,580
pattern change, it activates a
certain rule that you have to go

554
00:32:37,580 --> 00:32:41,033
to check. to see whether the
data provided was incorrect,

555
00:32:41,033 --> 00:32:44,790
inconsistent, or something else.
And those rules are typically

556
00:32:44,790 --> 00:32:48,183
done not for every single
company, but they're typically

557
00:32:48,183 --> 00:32:51,758
done for the tougher use cases
where we lack some data. And

558
00:32:51,758 --> 00:32:54,000
that's what underwriters do
manually.

559
00:32:54,000 --> 00:32:56,000
Audrow Nash: very interesting.

560
00:32:56,000 --> 00:33:00,339
Audrow Nash: Do you have to, so
I imagine that there are not

561
00:33:00,339 --> 00:33:04,171
many robotics underwriters
existing to, like the vast

562
00:33:04,171 --> 00:33:08,004
percentage of underwriters are,
I don't know, regular

563
00:33:08,004 --> 00:33:12,198
underwriters that work at a
regular insurance company that

564
00:33:12,198 --> 00:33:16,247
have a standard established
processes. How is it hard to

565
00:33:16,247 --> 00:33:20,296
find underwriters or do they
require a lot of additional

566
00:33:20,296 --> 00:33:24,780
training or is it just you have
your very clear process put in

567
00:33:24,780 --> 00:33:29,118
place so you can just drop any
underwriter in to your system

568
00:33:29,118 --> 00:33:33,529
and then they start functioning
in your system or how does it

569
00:33:33,529 --> 00:33:37,000
work to actually train
underwriters? Definitely.

570
00:33:37,000 --> 00:33:40,643
Sergey Litvinenko: great
question. I don't think anybody

571
00:33:40,643 --> 00:33:44,807
actually asked me that question.
So great. You are right. If you

572
00:33:44,807 --> 00:33:48,841
look at the world of insurance,
you can find like all kinds of

573
00:33:48,841 --> 00:33:52,810
underwriters, but there is not a
single underwriter that does

574
00:33:52,810 --> 00:33:56,779
robots. And that's, that's, what
we're trying to solve. We're

575
00:33:56,779 --> 00:34:00,227
trying to become that
underwriter. So what we've done

576
00:34:00,227 --> 00:34:03,480
is that, when we hire
underwriters, we try to hire

577
00:34:03,480 --> 00:34:07,319
with some background. And that
background usually has to do

578
00:34:07,319 --> 00:34:10,898
either with traditional. Product
Liability Insurance or

579
00:34:10,898 --> 00:34:14,931
Traditional Errors and Omissions
Insurance or Cyber Liability.

580
00:34:14,931 --> 00:34:18,705
Understanding of the insurance
side of things. And then we

581
00:34:18,705 --> 00:34:22,088
train them on robotics
specifically, on all kinds of

582
00:34:22,088 --> 00:34:25,927
use cases, on the methodology,
on what to look out for, and

583
00:34:25,927 --> 00:34:29,635
also on the system that we
developed internally. So there

584
00:34:29,635 --> 00:34:33,344
is, we are required to have
Underwriters with established

585
00:34:33,344 --> 00:34:37,378
industry background, and then we
train them on top. So that's,

586
00:34:37,378 --> 00:34:41,347
how it works. there is no other
way around it. And I do think

587
00:34:41,347 --> 00:34:45,250
that we are, in terms of the
customers and data that we have

588
00:34:45,250 --> 00:34:49,154
with the insights, we are the
only true robotics underwriter

589
00:34:49,154 --> 00:34:50,000
in the world.

590
00:34:50,000 --> 00:34:54,355
Audrow Nash: Yeah, I wouldn't be
surprised if that's true. Very

591
00:34:54,355 --> 00:34:57,880
interesting. how, difficult is
the training for the

592
00:34:57,880 --> 00:35:02,028
underwriters? Like how, quickly
can you get them up to speed

593
00:35:02,028 --> 00:35:05,000
assuming they have a reasonable
background?

594
00:35:05,000 --> 00:35:08,276
Sergey Litvinenko: under a
month, I would even say under

595
00:35:08,276 --> 00:35:11,728
three weeks, it's like a fire
drill, you, our met, when you

596
00:35:11,728 --> 00:35:15,296
implement the methodology, the
way we have it implemented, it

597
00:35:15,296 --> 00:35:18,865
looks relatively simple. Okay.
So it's okay, here are all the

598
00:35:18,865 --> 00:35:22,492
buttons you need to click here,
all the numbers you need to go

599
00:35:22,492 --> 00:35:26,060
through here, all the rules that
you need to complete. but in

600
00:35:26,060 --> 00:35:29,044
order to do that, our
methodology is like literally

601
00:35:29,044 --> 00:35:32,437
200 pages worth of just writing.
Stuff, right? And then we

602
00:35:32,437 --> 00:35:35,947
implemented that methodology.
we, pretty much onboard people

603
00:35:35,947 --> 00:35:39,516
onto those 200 pages in a very
user friendly way in less than

604
00:35:39,516 --> 00:35:43,201
three weeks. And the way we do
this is just, by just, just it's

605
00:35:43,201 --> 00:35:46,828
a part of the employee training
that we do. And we just have a

606
00:35:46,828 --> 00:35:50,338
process where in a very short
period of time, we're going to

607
00:35:50,338 --> 00:35:53,790
onboard you to everything. And
then as you start working on

608
00:35:53,790 --> 00:35:57,183
customers and you're going to
have any issues or that will

609
00:35:57,183 --> 00:36:00,634
have troubleshoot, we're just
going to do it on the fly. we

610
00:36:00,634 --> 00:36:03,676
don't have the luxury of
spending three months to do

611
00:36:03,676 --> 00:36:07,303
training, unfortunately. So we
have to do it really, fast. And

612
00:36:07,303 --> 00:36:09,000
we built and we built for it.

613
00:36:09,000 --> 00:36:12,359
Audrow Nash: Yeah, and you
mentioned, I might not remember

614
00:36:12,359 --> 00:36:15,486
at the beginning, did you say
you're like 10 people or

615
00:36:15,486 --> 00:36:18,846
something like this? And I
assume then you have a staff of

616
00:36:18,846 --> 00:36:22,205
underwriters for this kind of
thing, or is this 15 people?

617
00:36:22,205 --> 00:36:24,000
Does that include underwriters?

618
00:36:24,000 --> 00:36:26,000
Sergey Litvinenko: Yes.

619
00:36:26,000 --> 00:36:28,000
Audrow Nash: How many of those
are underwriters?

620
00:36:28,000 --> 00:36:29,000
Sergey Litvinenko: We have two
of them right now.

621
00:36:29,000 --> 00:36:33,159
Audrow Nash: Oh, okay. Very
interesting. are most of your

622
00:36:33,159 --> 00:36:37,538
customer, where, are, so you
have a lot of different domains

623
00:36:37,538 --> 00:36:41,843
that customers are in. Can you
just tell me a bit about the

624
00:36:41,843 --> 00:36:46,076
different domains that you're
serving, servicing, and What

625
00:36:46,076 --> 00:36:50,673
would be the rough percentages
for which domain? are you mostly

626
00:36:50,673 --> 00:36:55,416
autonomous vehicles or, how does
it change depending on different

627
00:36:55,416 --> 00:36:56,000
domains?

628
00:36:56,000 --> 00:37:00,840
Sergey Litvinenko: Yep. So I can
give you by the robot type I can

629
00:37:00,840 --> 00:37:03,000
also give you by the industry

630
00:37:03,000 --> 00:37:04,000
Audrow Nash: Both are very
interesting.

631
00:37:04,000 --> 00:37:08,286
Sergey Litvinenko: yep. Let's
start with the industry first.

632
00:37:08,286 --> 00:37:12,643
So we have Transportation or
like ride sharing which would be

633
00:37:12,643 --> 00:37:16,000
the robotaxis we have Holing
goods or tracking.

634
00:37:16,000 --> 00:37:19,124
Audrow Nash: Can you say the
companies or, like I'd be

635
00:37:19,124 --> 00:37:22,364
curious to hear who the
companies are if it's allowed to

636
00:37:22,364 --> 00:37:23,000
be exposed.

637
00:37:23,000 --> 00:37:26,375
Sergey Litvinenko: yep. I can
tell you the ones that we have

638
00:37:26,375 --> 00:37:29,863
on the website. Those are the,
we have the permissions for. we

639
00:37:29,863 --> 00:37:32,000
work with a Honda and Kia on the
robot

640
00:37:32,000 --> 00:37:33,000
Audrow Nash: Makes sense.

641
00:37:33,000 --> 00:37:37,608
Sergey Litvinenko: and, on the
tracking, I unfortunately cannot

642
00:37:37,608 --> 00:37:42,143
tell you, but it's one of the,
it's only a handful of players,

643
00:37:42,143 --> 00:37:46,385
so it's one of them. we. Also,
there is also, construction

644
00:37:46,385 --> 00:37:50,628
robotics. there is, mobile
robots, not it's, not just the,

645
00:37:50,628 --> 00:37:54,870
mobile robots that go into
warehousing specifically. Okay.

646
00:37:54,870 --> 00:37:59,332
So let's call it warehousing. we
also have the medical space,

647
00:37:59,332 --> 00:38:03,355
which is by the way, really
tough for the robotics, but

648
00:38:03,355 --> 00:38:07,597
we'll still have that. if there
is a, aerial use cases, so

649
00:38:07,597 --> 00:38:11,328
aerial transportation it's
drones, done for, either

650
00:38:11,328 --> 00:38:15,716
transporting goods or doing
aerial surveys or doing any kind

651
00:38:15,716 --> 00:38:20,178
of data collection. There are
marine use cases primarily used

652
00:38:20,178 --> 00:38:24,347
for research and development
purposes, and that has to do

653
00:38:24,347 --> 00:38:28,590
with kind of on surface vessels
and also subsurface or the

654
00:38:28,590 --> 00:38:32,905
submarines. We also have boring
machines, the machines that

655
00:38:32,905 --> 00:38:37,586
actually Dig Tunnels, there is a
huge human component to that as

656
00:38:37,586 --> 00:38:42,267
well, but it's still considered
to be a robot and we ensure that

657
00:38:42,267 --> 00:38:46,656
as well. that's just on a high
level, I have a very detailed

658
00:38:46,656 --> 00:38:50,825
breakdown in, in, in our
financials, but just to give you

659
00:38:50,825 --> 00:38:55,068
on a high level, and, in terms
of the robot types, that's,

660
00:38:55,068 --> 00:38:59,456
Everything when it comes to the
new generation of robots. So

661
00:38:59,456 --> 00:39:03,187
that would be, robotized
vehicles, public vehicles,

662
00:39:03,187 --> 00:39:07,648
mobile robots, robotic arms that
do all kinds of manipulation

663
00:39:07,648 --> 00:39:11,964
with different number of degrees
of freedom. Some are super

664
00:39:11,964 --> 00:39:16,353
simple, some are super complex.
we have robots that do, like

665
00:39:16,353 --> 00:39:20,449
moving pallets in open field for
agricultural use cases,

666
00:39:20,449 --> 00:39:25,130
construction robots that do all
kinds of, paintings on the roads

667
00:39:25,130 --> 00:39:29,592
or all kinds of fencing or any
kind of welding. That would be

668
00:39:29,592 --> 00:39:33,834
security robots that just go
around the perimeter and they

669
00:39:33,834 --> 00:39:38,515
are not armed, okay, but they go
around the perimeter collecting

670
00:39:38,515 --> 00:39:42,758
data in the real time about any
potential intrusion or any

671
00:39:42,758 --> 00:39:47,000
potential suspicious behavior.
The same applies to drones.

672
00:39:47,000 --> 00:39:49,000
Audrow Nash: Cobalt or something
like this

673
00:39:49,000 --> 00:39:52,730
Sergey Litvinenko: Yeah, so one
thing that we don't have in our

674
00:39:52,730 --> 00:39:55,808
portfolio, is like the old
generation of robots. For

675
00:39:55,808 --> 00:39:59,065
example, if you are, an
automotive manufacturer and you

676
00:39:59,065 --> 00:40:02,854
have the automated assembly line
and you have those KUKA robots,

677
00:40:02,854 --> 00:40:05,991
we do not do that primarily
because it's not just the

678
00:40:05,991 --> 00:40:09,188
technical consideration. It's
also more of a financial

679
00:40:09,188 --> 00:40:12,741
consideration. Companies that
buy those robots, they have so

680
00:40:12,741 --> 00:40:16,174
much money that if something
goes wrong, they can just pay

681
00:40:16,174 --> 00:40:19,016
for that. They do not
necessarily buy insurance.

682
00:40:19,016 --> 00:40:22,805
That's why we deal with a newer
generation of robots where there

683
00:40:22,805 --> 00:40:25,943
is not a lot of precedent, and
they're also in a less

684
00:40:25,943 --> 00:40:29,258
constrained environment with
more degrees of freedom out

685
00:40:29,258 --> 00:40:32,692
there. that's what we are
focused on. And probably the two

686
00:40:32,692 --> 00:40:36,126
startling examples would be
Robotech system, public roads,

687
00:40:36,126 --> 00:40:39,560
and also humanoids when we get
those like for Koop, that's

688
00:40:39,560 --> 00:40:43,290
going to be, that's our thing.
We're going to ensure that a day

689
00:40:43,290 --> 00:40:44,000
in, day out.

690
00:40:44,000 --> 00:40:46,000
Audrow Nash: Why?

691
00:40:46,000 --> 00:40:49,148
Sergey Litvinenko: I think. For
first for economic

692
00:40:49,148 --> 00:40:52,799
considerations, we do believe
that the next gen robots are

693
00:40:52,799 --> 00:40:56,703
going to be the future and it's
a chicken in the neck problem.

694
00:40:56,703 --> 00:41:00,606
do you first ship the robot and
then you do insurance? But you

695
00:41:00,606 --> 00:41:04,572
need the insurance to ship the
robot in the first place because

696
00:41:04,572 --> 00:41:08,413
your partners, customers, and
people would want to have that.

697
00:41:08,413 --> 00:41:12,379
So that's what we're trying to
solve, the economic transaction.

698
00:41:12,379 --> 00:41:16,031
And two, our own. selfish
motivation is that we want to be

699
00:41:16,031 --> 00:41:20,060
the first mover. We want to grab
the market. We want to have the

700
00:41:20,060 --> 00:41:24,089
most data on that. And it's, the
best way to do this is, just to

701
00:41:24,089 --> 00:41:27,489
do this at the time when
everything happens. So that's

702
00:41:27,489 --> 00:41:29,000
what we're, planning on.

703
00:41:29,000 --> 00:41:31,990
Audrow Nash: Yeah. So you're
going to be there, right? you'll

704
00:41:31,990 --> 00:41:34,000
be the early bird for this kind
of thing.

705
00:41:34,000 --> 00:41:35,000
Sergey Litvinenko: exactly.

706
00:41:35,000 --> 00:41:38,867
Audrow Nash: Awesome. And, but
why humanoids? Why are they

707
00:41:38,867 --> 00:41:41,000
especially interesting for this?

708
00:41:41,000 --> 00:41:44,665
Sergey Litvinenko: I'm honestly,
so we are a function of the

709
00:41:44,665 --> 00:41:48,330
market. So if there is a new use
case that appears, like the

710
00:41:48,330 --> 00:41:52,300
first thing that we think about,
we want to see how it plays out.

711
00:41:52,300 --> 00:41:56,026
Okay. In order to see that we
need to ensure that I think the

712
00:41:56,026 --> 00:41:59,691
reason why I mentioned that is
because there is just so much

713
00:41:59,691 --> 00:42:03,417
news about humanoids and I'm
like, okay, what's going on? and

714
00:42:03,417 --> 00:42:07,265
then we figured that there are
like 10 companies in the U S and

715
00:42:07,265 --> 00:42:10,930
I think 20 companies in China
that are working on humanoids,

716
00:42:10,930 --> 00:42:14,534
they're at a demo stage, but
the, progress has been, pretty

717
00:42:14,534 --> 00:42:18,077
tremendous over the last year
and a half, I would say. and

718
00:42:18,077 --> 00:42:21,558
we're thinking, when the first
humanoids start to ship to

719
00:42:21,558 --> 00:42:24,979
customers or to the first
companies that are going to be

720
00:42:24,979 --> 00:42:28,705
adopted at, Like what insurance
applications are going to be,

721
00:42:28,705 --> 00:42:32,003
how we can pull the data from
them to understand their

722
00:42:32,003 --> 00:42:35,974
performance. Okay. And apply our
methodology. And how can we save

723
00:42:35,974 --> 00:42:39,761
money to both the developers and
to the customers? That's what

724
00:42:39,761 --> 00:42:42,754
I'm thinking. And since
humanoids are all over my

725
00:42:42,754 --> 00:42:46,113
Twitter or X feed, I'm like,
that's, I just cannot stop

726
00:42:46,113 --> 00:42:49,778
thinking about it. that's why
I'm really curious to see what

727
00:42:49,778 --> 00:42:51,000
this is going to go.

728
00:42:51,000 --> 00:42:55,864
Audrow Nash: Okay. That's very
funny. So yeah. Yeah. It'll be

729
00:42:55,864 --> 00:43:00,728
really interesting to see where
it goes and what the timeline

730
00:43:00,728 --> 00:43:05,592
is. Another question that I have
with this, how do you handle

731
00:43:05,592 --> 00:43:09,818
software updates? if I am a
robotics company and I am

732
00:43:09,818 --> 00:43:14,603
shipping a new version, but
there's definitely precedent for

733
00:43:14,603 --> 00:43:18,988
a new version just completely
bricking the robot in its

734
00:43:18,988 --> 00:43:23,533
behavior, or fundamentally
changing the behavior and thus

735
00:43:23,533 --> 00:43:28,238
your past estimates are now
broken. different than what the

736
00:43:28,238 --> 00:43:32,942
or your past estimates are
assuming different behavior than

737
00:43:32,942 --> 00:43:37,886
the current behavior. How do you
deal with robotics being more

738
00:43:37,886 --> 00:43:42,591
iterative? Or do you only deal
with robots that have locked

739
00:43:42,591 --> 00:43:47,136
down behavior? Or is part of
your risk estimation how the

740
00:43:47,136 --> 00:43:52,000
companies are doing testing and
their rollout of updates? Ah,

741
00:43:52,000 --> 00:43:55,494
Sergey Litvinenko: very tough
question, but I have a simple

742
00:43:55,494 --> 00:43:59,225
answer for that. So here's what
we'll do. The, the quantitative

743
00:43:59,225 --> 00:44:02,541
data that we track, which will
be the performance of the

744
00:44:02,541 --> 00:44:06,035
behavior of the robot, the
troubleshooting rate that we use

745
00:44:06,035 --> 00:44:09,529
for estimates. we're going to
put it on a timeline. So it's

746
00:44:09,529 --> 00:44:13,319
going to be a time series, okay,
of the rate, of the performance

747
00:44:13,319 --> 00:44:16,991
rate. And we're going to layer
the performance versions on top

748
00:44:16,991 --> 00:44:20,722
of that. And then what we want
to see is that the more software

749
00:44:20,722 --> 00:44:24,512
updates you ship, the lower the
troubleshooting rate becomes, or

750
00:44:24,512 --> 00:44:27,651
the higher the performance
becomes. That's what we're

751
00:44:27,651 --> 00:44:31,085
looking for. We are not too
surgical about it because it's

752
00:44:31,085 --> 00:44:34,639
very hard. Like we're going to
cross the border into safety.

753
00:44:34,639 --> 00:44:37,659
which we do not do, we do not
want to, or, into the

754
00:44:37,659 --> 00:44:41,390
capabilities of the robot, which
we do not want to do. if there

755
00:44:41,390 --> 00:44:44,706
is a fundamental change, that
you, that the robot or the

756
00:44:44,706 --> 00:44:48,022
fundamental change in the
behavioral model that you ship

757
00:44:48,022 --> 00:44:51,161
there, we want to see the
improvement no matter what.

758
00:44:51,161 --> 00:44:54,833
Okay. And the only way to see
the improvement is that when the

759
00:44:54,833 --> 00:44:57,912
robot goes into the real world
environment, that the

760
00:44:57,912 --> 00:45:01,584
troubleshooting rate goes down.
that's what we're looking for.

761
00:45:01,584 --> 00:45:05,256
If we don't see that, Then that
means that the software update

762
00:45:05,256 --> 00:45:08,987
was maybe not good. Maybe it was
something else. We don't know.

763
00:45:08,987 --> 00:45:11,000
But that's what we're looking
for.

764
00:45:11,000 --> 00:45:15,734
Audrow Nash: Yeah, but so what
I'm thinking is like someone

765
00:45:15,734 --> 00:45:20,950
ships an update of how the robot
handles. It could even be just a

766
00:45:20,950 --> 00:45:25,523
bug. say, the robot can't
reliably like the camera images

767
00:45:25,523 --> 00:45:30,177
the robot has are off by one
index now. And so what is the

768
00:45:30,177 --> 00:45:34,831
whole image like every row is
really confusing and you get

769
00:45:34,831 --> 00:45:39,806
like these weird stripes across
the image. Now the robot can't

770
00:45:39,806 --> 00:45:44,781
do perception. and it. Bricks
the robots effectively, but they

771
00:45:44,781 --> 00:45:49,515
don't realize this until they
put them into production, and

772
00:45:49,515 --> 00:45:54,329
this camera only turns on in a
specific part of the process.

773
00:45:54,329 --> 00:45:59,144
And so then it fails here. I'm
sure new updates break things

774
00:45:59,144 --> 00:46:04,199
occasionally. how, do you handle
this kind of thing? Or, maybe,

775
00:46:04,199 --> 00:46:09,334
you just expect strict improving
performance. but I would expect

776
00:46:09,334 --> 00:46:14,149
that bugs occasionally happen.
That our performance breaking

777
00:46:14,149 --> 00:46:18,000
even if the company has good
processes in place.

778
00:46:18,000 --> 00:46:22,073
Sergey Litvinenko: the way to
catch that, fortunately or not,

779
00:46:22,073 --> 00:46:26,012
is through the qualitative
evaluation. when it comes to the

780
00:46:26,012 --> 00:46:30,084
real world data, we will, if
there is a bug that gets shipped

781
00:46:30,084 --> 00:46:34,224
to production, we're going to
see that, and unfortunately it's

782
00:46:34,224 --> 00:46:38,363
going to affect the performance
and the evaluation on our end.

783
00:46:38,363 --> 00:46:41,968
The way to prevent that is,
okay, how do you stage the

784
00:46:41,968 --> 00:46:45,974
release? What is the company's
procedure to pushing to prod?

785
00:46:45,974 --> 00:46:49,913
And, we, by the way, do ask
questions related to that. it's

786
00:46:49,913 --> 00:46:53,518
okay, what kind of testing
procedures do you have? how

787
00:46:53,518 --> 00:46:57,324
much, data do you have to
collect? Or performance data on

788
00:46:57,324 --> 00:47:01,663
the robot to make sure that it's
good to go? And, there's a bunch

789
00:47:01,663 --> 00:47:04,000
of other additional questions.
And,

790
00:47:04,000 --> 00:47:07,000
Audrow Nash: So you fall back on
qualitative things,

791
00:47:07,000 --> 00:47:08,000
Sergey Litvinenko: yes.

792
00:47:08,000 --> 00:47:12,000
Audrow Nash: because it's a very
hard judgment to make.

793
00:47:12,000 --> 00:47:17,088
Audrow Nash: what, so what I
was, what I'm getting at with

794
00:47:17,088 --> 00:47:22,175
this is if there's a large
performance break in the robot,

795
00:47:22,175 --> 00:47:26,912
does it make it so that it
invalidates your insurance?

796
00:47:26,912 --> 00:47:32,263
because they're not living up to
the estimates that you based

797
00:47:32,263 --> 00:47:37,175
your policy on or something like
that. if the robot does

798
00:47:37,175 --> 00:47:42,351
something completely crazy
because of a recent bug that was

799
00:47:42,351 --> 00:47:47,000
not there when you insured them,
how is that handled?

800
00:47:47,000 --> 00:47:50,583
Sergey Litvinenko: so we, what I
think you are implying,

801
00:47:50,583 --> 00:47:54,485
something that's called as a
performance guarantee. let's say

802
00:47:54,485 --> 00:47:58,644
you have a cloud service and you
guarantee that 99. 99 percent of

803
00:47:58,644 --> 00:48:02,675
the time it's going to be up and
then for 0. 01%, there's going

804
00:48:02,675 --> 00:48:06,577
to be a performance guarantee
that we're going to pay you out

805
00:48:06,577 --> 00:48:10,352
certain amount of money, blah,
blah, blah. It's called It's

806
00:48:10,352 --> 00:48:14,382
parametric insurance. there is a
very clear parameter, and once

807
00:48:14,382 --> 00:48:18,541
you go out of the bound, that's,
you're gonna get a payout. We do

808
00:48:18,541 --> 00:48:22,316
not do that. there is no way
that a certain bug or a sudden

809
00:48:22,316 --> 00:48:26,410
decrease in performance is going
to invalidate our insurance. We

810
00:48:26,410 --> 00:48:30,505
do not do parametric insurance.
Yeah, at least, we haven't found

811
00:48:30,505 --> 00:48:33,000
a use case for robotics for it
for now.

812
00:48:33,000 --> 00:48:34,000
Audrow Nash: gotcha.

813
00:48:34,000 --> 00:48:37,998
Sergey Litvinenko: And, yep, so
if, let's say, the only thing

814
00:48:37,998 --> 00:48:41,930
that can invalidate insurance if
the, the erroneous, not the

815
00:48:41,930 --> 00:48:45,862
erroneous, but the purposefully,
skewed data or purposefully

816
00:48:45,862 --> 00:48:49,138
incorrect data that you
submitted as a part of the

817
00:48:49,138 --> 00:48:53,070
process. But when it comes to
the troubleshooting, it's very

818
00:48:53,070 --> 00:48:57,002
hard to, provide bad data or API
because that comes directly

819
00:48:57,002 --> 00:49:00,803
from, the, robots or from the
fleet. qualitative, yes, you

820
00:49:00,803 --> 00:49:04,867
can, misrepresent things. And
that's why. when, you mentioned,

821
00:49:04,867 --> 00:49:08,340
how we catch that, yes, we have
to catch that through

822
00:49:08,340 --> 00:49:12,206
understanding what steps the
company has in order to get to

823
00:49:12,206 --> 00:49:16,400
prod. If they said that they had
the steps, but they didn't have

824
00:49:16,400 --> 00:49:20,201
those steps, and during the
investigation after the claim,

825
00:49:20,201 --> 00:49:24,068
we figured that out, yeah, then
we can invalidate that. The

826
00:49:24,068 --> 00:49:28,000
insurance and say, you're not
getting paid because you lied.

827
00:49:28,000 --> 00:49:29,000
Audrow Nash: You misrepresent.
Yeah.

828
00:49:29,000 --> 00:49:31,706
Sergey Litvinenko: yeah. And
what's, what's funny. It has

829
00:49:31,706 --> 00:49:34,601
nothing to do with robots. It
has to do with humans. The fact

830
00:49:34,601 --> 00:49:37,639
that you lie on the application.
but that's, that's the business

831
00:49:37,639 --> 00:49:40,155
that we're in. we have to deal
with this, things. And

832
00:49:40,155 --> 00:49:43,051
fortunately we haven't had a
situation like that. I think the

833
00:49:43,051 --> 00:49:46,041
way we run the process and the
way we market ourselves is that.

834
00:49:46,041 --> 00:49:49,032
People are trustworthy and they
understand that the worst thing

835
00:49:49,032 --> 00:49:51,642
that can happen is that when
they need to pay a million

836
00:49:51,642 --> 00:49:54,538
dollars in liability and their
insurance becomes invalidated,

837
00:49:54,538 --> 00:49:57,528
that's literally the worst thing
that you can, that can happen.

838
00:49:57,528 --> 00:49:59,000
And they don't want to do that.

839
00:49:59,000 --> 00:50:02,703
Audrow Nash: Yeah. Cause that's
what I think of. I haven't had

840
00:50:02,703 --> 00:50:06,585
any negative experiences myself,
but I've heard things where like

841
00:50:06,585 --> 00:50:10,168
insurance, it's right when you
need the claim, they're like,

842
00:50:10,168 --> 00:50:13,692
Nope, it wasn't validated for
some reason, whatever. And, I

843
00:50:13,692 --> 00:50:16,678
could imagine that A robot
shipping a bug and just

844
00:50:16,678 --> 00:50:20,321
destroying its own performance
because of that bug, hopefully

845
00:50:20,321 --> 00:50:24,083
rare, but I could imagine that
would be one of those cases that

846
00:50:24,083 --> 00:50:27,547
you would want to terminate
because it's I don't know, say

847
00:50:27,547 --> 00:50:31,369
it does something just wonky and
it's different than what you've

848
00:50:31,369 --> 00:50:35,013
expected with all your behavior
analysis. so that it's almost

849
00:50:35,013 --> 00:50:38,297
just, it's justifiable to me in
some way, but hopefully

850
00:50:38,297 --> 00:50:42,000
companies are more reliable with
what they ship to production.

851
00:50:42,000 --> 00:50:44,000
Sergey Litvinenko: I can add one
more thing here.

852
00:50:44,000 --> 00:50:47,706
Sergey Litvinenko: A, what the
beauty about robotics is exactly

853
00:50:47,706 --> 00:50:51,412
what you just said. If there is
a bug, then we expect that this

854
00:50:51,412 --> 00:50:54,706
bug is going to be fixed, and
this bug is never going to

855
00:50:54,706 --> 00:50:58,235
happen again, okay? Very hard
with humans, because it's very

856
00:50:58,235 --> 00:51:00,000
hard to change human behavior,

857
00:51:00,000 --> 00:51:04,000
Audrow Nash: Oh, that's super
interesting. Yes.

858
00:51:04,000 --> 00:51:07,058
Sergey Litvinenko: have a fleet
of robots that's going to be

859
00:51:07,058 --> 00:51:09,862
continuously and probably
exponentially improving, with

860
00:51:09,862 --> 00:51:12,767
humans, impossible. That's why
our side of insurance, the

861
00:51:12,767 --> 00:51:16,029
robotics insurance, is going to
be so rapidly evolving up to the

862
00:51:16,029 --> 00:51:18,934
point that it's going to be
Super easy, super economical,

863
00:51:18,934 --> 00:51:21,993
and super user friendly, because
we can control the robotics

864
00:51:21,993 --> 00:51:25,000
behavior, one way or another.
With humans, very hard to do.

865
00:51:25,000 --> 00:51:30,378
Audrow Nash: is hilarious. That
is such a thing. because humans,

866
00:51:30,378 --> 00:51:35,000
so humans, we know how to drive,
but we keep getting in

867
00:51:35,000 --> 00:51:36,000
Sergey Litvinenko: we know.

868
00:51:36,000 --> 00:51:40,281
Audrow Nash: but, we literally
are able to get in our car and

869
00:51:40,281 --> 00:51:44,491
this kind of thing, but then
people are reckless, they're on

870
00:51:44,491 --> 00:51:48,772
their phone, they're drinking a
big slushy, whatever it might

871
00:51:48,772 --> 00:51:52,702
be, and then they get in
accidents. And you can't change

872
00:51:52,702 --> 00:51:56,982
that, In the whole population,
but with robots, you can hit a

873
00:51:56,982 --> 00:52:01,263
bug once. And so if you hit the
bug once, then everyone knows

874
00:52:01,263 --> 00:52:05,684
about this. Then you make a good
fix for it. The fix is robust,

875
00:52:05,684 --> 00:52:09,754
hopefully. And then from there,
that failure case is not a

876
00:52:09,754 --> 00:52:13,895
thing. And that actually makes
it seem like this is like an

877
00:52:13,895 --> 00:52:16,000
insurance person's dream, in a

878
00:52:16,000 --> 00:52:18,611
Sergey Litvinenko: Yes, it is.
It absolutely is. That's why we

879
00:52:18,611 --> 00:52:20,000
are, that's why we're doing
this.

880
00:52:20,000 --> 00:52:22,791
Sergey Litvinenko: And I got to
tell you our, there are two

881
00:52:22,791 --> 00:52:25,628
performance metrics that we
have. One performance metrics is

882
00:52:25,628 --> 00:52:28,703
of course, how many customers we
have and how much business we've

883
00:52:28,703 --> 00:52:31,399
produced. The other performance
metrics, which is equally

884
00:52:31,399 --> 00:52:34,000
important, is how many losses we
had. And I got to tell

885
00:52:34,000 --> 00:52:35,000
Audrow Nash: they keep going
down

886
00:52:35,000 --> 00:52:39,355
Sergey Litvinenko: yeah, I got
to tell you, like the, If you

887
00:52:39,355 --> 00:52:43,855
look at property and casualty
industry, which is like casualty

888
00:52:43,855 --> 00:52:48,427
means, third party liability and
property is property. Industry

889
00:52:48,427 --> 00:52:52,710
average is around 70 to 75
percent. of, losses paid out. So

890
00:52:52,710 --> 00:52:57,137
what does it mean? Out of every
dollar that insurance company

891
00:52:57,137 --> 00:53:01,710
collects, you pay out 70 cents.
Guess how much our payout is so

892
00:53:01,710 --> 00:53:02,000
far?

893
00:53:02,000 --> 00:53:05,000
Audrow Nash: 20 cents. Five.

894
00:53:05,000 --> 00:53:06,000
Sergey Litvinenko: Five.

895
00:53:06,000 --> 00:53:08,000
Audrow Nash: Holy cow.

896
00:53:08,000 --> 00:53:11,134
Sergey Litvinenko: that's, how
much better it is. Even like

897
00:53:11,134 --> 00:53:14,427
we're an early stage company, we
don't have a ton of data yet,

898
00:53:14,427 --> 00:53:17,720
but from the data that we have,
it's already that much better.

899
00:53:17,720 --> 00:53:20,907
And it's, literally printing,
underwriting profits. And what

900
00:53:20,907 --> 00:53:24,093
we're going to do is that we're
going to take that and we're

901
00:53:24,093 --> 00:53:27,121
going to, do the iteration of
the models that we have and

902
00:53:27,121 --> 00:53:30,414
we're going to start bringing
the rates down. So it's going to

903
00:53:30,414 --> 00:53:33,495
be even cheaper to get our
insurance. that's the magnitude

904
00:53:33,495 --> 00:53:36,788
that I'm talking about, and that
gets me really excited, about

905
00:53:36,788 --> 00:53:37,000
this

906
00:53:37,000 --> 00:53:41,283
Audrow Nash: That is exciting.
Yeah, super, super exciting.

907
00:53:41,283 --> 00:53:45,711
You're in a, you're in a hot,
space for this kind of thing, I

908
00:53:45,711 --> 00:53:50,430
think. and it's very interesting
that conventional insurance risk

909
00:53:50,430 --> 00:53:54,350
stays about the same over time.
and yours decreases as

910
00:53:54,350 --> 00:53:58,923
performance improves. That's a
fundamentally different dynamic,

911
00:53:58,923 --> 00:54:02,916
but you still get this old
industry, and people do feel

912
00:54:02,916 --> 00:54:06,836
better with insurance. They
sleep better, and you know

913
00:54:06,836 --> 00:54:11,554
you're not going out of business
tomorrow because of some mishap.

914
00:54:11,554 --> 00:54:15,910
but it's just like you're on the
right side of this dynamic,

915
00:54:15,910 --> 00:54:20,628
which is super, super cool. What
a thing. Yeah, that point about,

916
00:54:20,628 --> 00:54:24,911
the robot only does it once, or
you run into this bug once,

917
00:54:24,911 --> 00:54:26,000
that's amazing.

918
00:54:26,000 --> 00:54:31,511
Audrow Nash: related to this,
how do you, so machine learning

919
00:54:31,511 --> 00:54:37,022
Deep Learning, and all of these,
all the AI related stuff. it

920
00:54:37,022 --> 00:54:42,804
often has very good performance
for most cases, but then you get

921
00:54:42,804 --> 00:54:47,863
these weird edge cases.
occasionally when the conditions

922
00:54:47,863 --> 00:54:53,193
are Outside of the training
data, for example, how, so it's

923
00:54:53,193 --> 00:54:58,704
similar to the bug question, but
maybe a little bit different

924
00:54:58,704 --> 00:55:04,486
where the behavior might be hard
to evaluate. You can look at it

925
00:55:04,486 --> 00:55:09,726
from a completely external way
where you just see that out

926
00:55:09,726 --> 00:55:15,237
behavior, how, any thoughts on
evaluating risk in AI systems,

927
00:55:15,237 --> 00:55:20,025
deep learning, using, I don't
know, deep learning for

928
00:55:20,025 --> 00:55:24,000
perception or control or
anything like this.

929
00:55:24,000 --> 00:55:27,474
Sergey Litvinenko: we do not
interfere to that level. so we

930
00:55:27,474 --> 00:55:31,006
do not collect any metrics on
how, let's say, the perception

931
00:55:31,006 --> 00:55:34,775
algorithm performs, or let's say
how many Subjects that classify

932
00:55:34,775 --> 00:55:38,307
correctly or incorrectly, or
how, or like we do not evaluate

933
00:55:38,307 --> 00:55:41,604
the quality of the path
planning, we do not do that. And

934
00:55:41,604 --> 00:55:45,137
primarily because we believe
it's the job of the engineering

935
00:55:45,137 --> 00:55:48,729
team of the safety team. And we
like our thing is that we. We

936
00:55:48,729 --> 00:55:52,026
want to see how the robot
interacts with the real world.

937
00:55:52,026 --> 00:55:55,558
And there are certain metrics
that we can use. They are high

938
00:55:55,558 --> 00:55:59,268
level metrics, but they are we
believe they are good predictors

939
00:55:59,268 --> 00:56:02,388
of the accidents that can
happen. So for example, the

940
00:56:02,388 --> 00:56:05,862
troubleshooting rate, how many
times the robot cannot, does

941
00:56:05,862 --> 00:56:09,571
know what it has to do. It can
come from any part of the stack.

942
00:56:09,571 --> 00:56:13,222
Okay, but we do not, we do not
have an instant idea where it's

943
00:56:13,222 --> 00:56:16,519
exactly coming from. usually
what we would do is that we

944
00:56:16,519 --> 00:56:19,816
would, follow up with the
company and collect additional

945
00:56:19,816 --> 00:56:22,937
data in terms of the
distribution that they're seeing

946
00:56:22,937 --> 00:56:26,410
on, the, what's failing the
most, but it's not happening in

947
00:56:26,410 --> 00:56:30,061
the real time. And it's just,
just economic, practically, It's

948
00:56:30,061 --> 00:56:33,652
very hard to go to that level.
And I know there are companies

949
00:56:33,652 --> 00:56:37,067
who try to do that. I don't
think it worked out. It's very

950
00:56:37,067 --> 00:56:40,364
hard to sell an insurance
product that's literally going

951
00:56:40,364 --> 00:56:43,249
to evaluate your whole
engineering stack. Yep. So

952
00:56:43,249 --> 00:56:46,664
that's why we are, have to deal
with a bunch of high level

953
00:56:46,664 --> 00:56:50,138
metrics. Okay. but I can tell
you the way we deal with edge

954
00:56:50,138 --> 00:56:53,906
cases is in the insurance, there
are two ways to be found. first

955
00:56:53,906 --> 00:56:57,380
of all, you cannot predict in
edge case. Okay. It's just no

956
00:56:57,380 --> 00:57:01,089
way it's possible. even if you
look at the performance, you can

957
00:57:01,089 --> 00:57:04,563
assume a drastic change in the
performance, but it's. It's,

958
00:57:04,563 --> 00:57:08,096
it's just an assumption. Okay,
we don't know how good it is.

959
00:57:08,096 --> 00:57:11,628
One thing you can do though, is
that if you say that an edge

960
00:57:11,628 --> 00:57:15,220
case, you, first of all, you
don't know what the edge case is

961
00:57:15,220 --> 00:57:18,399
going to happen, so the
frequency component of it, you

962
00:57:18,399 --> 00:57:21,696
throw it out, we don't know. But
what we can do, we can,

963
00:57:21,696 --> 00:57:25,170
Approximate, what edge case
would look like on the severity

964
00:57:25,170 --> 00:57:28,703
side of things. And we can say,
if an edge case is something

965
00:57:28,703 --> 00:57:32,177
that we never thought would
happen, and the worst case that

966
00:57:32,177 --> 00:57:35,945
can happen is that it costs us a
fortune. So let's say that one,

967
00:57:35,945 --> 00:57:39,419
once, like once in a while,
there's going to be an accident

968
00:57:39,419 --> 00:57:43,246
that's going to cost us a ton of
money and we're going to call it

969
00:57:43,246 --> 00:57:46,778
an edge case, whatever it's
going to be. And then it's going

970
00:57:46,778 --> 00:57:50,193
to be like max out all the
possible severity at 10 million

971
00:57:50,193 --> 00:57:53,549
loss. Okay. Or 10 million
lawsuit or something like that.

972
00:57:53,549 --> 00:57:57,141
And then, we're like, okay, if
we assume that. or incorporate

973
00:57:57,141 --> 00:58:00,850
that into our model, what is it
going to do to it? And then, we

974
00:58:00,850 --> 00:58:04,324
do see the rates that they do
change, but not significantly

975
00:58:04,324 --> 00:58:07,974
because it's more likely, most
likely going to be a rare event

976
00:58:07,974 --> 00:58:11,448
at the end of the day. that's
one way we can do this on the

977
00:58:11,448 --> 00:58:15,040
model side. On the qualitative
side, that's probably where we

978
00:58:15,040 --> 00:58:18,749
can do. The most have the most
impact on our customers, because

979
00:58:18,749 --> 00:58:22,164
think about it this way. it's
not just about the edge case

980
00:58:22,164 --> 00:58:25,755
happening. It's about what are
you going to do right after it

981
00:58:25,755 --> 00:58:29,465
happens. so for example, there
was a public example with one AV

982
00:58:29,465 --> 00:58:33,115
company in San Francisco. It was
like an insane edge case. you

983
00:58:33,115 --> 00:58:36,000
probably know what I'm talking
about, right? And,

984
00:58:36,000 --> 00:58:39,403
Audrow Nash: the specific
instance. I assume it's

985
00:58:39,403 --> 00:58:41,000
something with, Cruise.

986
00:58:41,000 --> 00:58:42,000
Sergey Litvinenko: yeah,
exactly.

987
00:58:42,000 --> 00:58:46,517
Audrow Nash: Was it, they pulled
out of San Francisco. I just,

988
00:58:46,517 --> 00:58:51,106
there were so many little mess
ups. was it like blocking police

989
00:58:51,106 --> 00:58:53,000
officers or something? Or,

990
00:58:53,000 --> 00:58:57,062
Sergey Litvinenko: it was a
situation when. So a cruise car

991
00:58:57,062 --> 00:59:01,125
was just driving by and there
was a human driver. the human

992
00:59:01,125 --> 00:59:05,462
driver hit a pedestrian who was
running in the wrong place. And

993
00:59:05,462 --> 00:59:09,662
then after the human car hit the
pedestrian, it fell onto the

994
00:59:09,662 --> 00:59:13,518
cruise vehicle and then the
cruise vehicle went over the

995
00:59:13,518 --> 00:59:14,000
person,

996
00:59:14,000 --> 00:59:15,000
Audrow Nash: oh,

997
00:59:15,000 --> 00:59:18,268
Sergey Litvinenko: remember that
one? Yeah, it's it's an edge

998
00:59:18,268 --> 00:59:21,482
case of edge cases. Cruise had
nothing to do with it. Just a

999
00:59:21,482 --> 00:59:24,482
person got hit by a human driver
first and fell into the

1000
00:59:24,482 --> 00:59:27,804
cruiser's lap. And, then, the
vehicle apparently didn't notice

1001
00:59:27,804 --> 00:59:30,643
that there was a person
underneath the vehicle and it

1002
00:59:30,643 --> 00:59:33,000
dragged that person for 20 feet
to the curb.

1003
00:59:33,000 --> 00:59:34,710
Audrow Nash: Oh, yeah. I
remember hearing about that.

1004
00:59:34,710 --> 00:59:35,000
huh. It's

1005
00:59:35,000 --> 00:59:37,400
Sergey Litvinenko: That's an
edge case of edge cases, if you

1006
00:59:37,400 --> 00:59:38,000
think about it.

1007
00:59:38,000 --> 00:59:39,000
Audrow Nash: horrifying.

1008
00:59:39,000 --> 00:59:42,800
Sergey Litvinenko: and the thing
is, I think it's not the fact

1009
00:59:42,800 --> 00:59:46,538
that it happened, but it's the
fact that, how the cruises the

1010
00:59:46,538 --> 00:59:50,154
company handled it, which led to
the, to pull it out of San

1011
00:59:50,154 --> 00:59:53,280
Francisco, there was like the
governance issue, the

1012
00:59:53,280 --> 00:59:57,141
communication issue. that's what
led to it. And when we look at

1013
00:59:57,141 --> 01:00:01,002
robotics companies, it's really
important that, we evaluate for

1014
01:00:01,002 --> 01:00:04,741
a, fire department, or is there,
do you have a literally fire

1015
01:00:04,741 --> 01:00:08,173
department at a robotics company
is going to handle that

1016
01:00:08,173 --> 01:00:12,156
particular issue? And believe it
or not, most of the companies Do

1017
01:00:12,156 --> 01:00:15,527
not have that. because they
focus on building stuff and

1018
01:00:15,527 --> 01:00:19,204
shipping the stuff. They're not
focusing about the financial

1019
01:00:19,204 --> 01:00:22,820
downside, which God knows if
it's ever going to happen. And

1020
01:00:22,820 --> 01:00:26,620
understandably, that's our job,
by the way. And, what we found

1021
01:00:26,620 --> 01:00:30,359
is that, the best way to deal
with edge cases is to make sure

1022
01:00:30,359 --> 01:00:34,097
that whenever it happens, that
you do not do stupid stuff and

1023
01:00:34,097 --> 01:00:37,345
you do not destroy your
reputation and you do not get

1024
01:00:37,345 --> 01:00:39,000
dragged into the courtroom.

1025
01:00:39,000 --> 01:00:40,000
Audrow Nash: Yeah.

1026
01:00:40,000 --> 01:00:41,690
Sergey Litvinenko: you need to
have specific things that you

1027
01:00:41,690 --> 01:00:42,000
need to do.

1028
01:00:42,000 --> 01:00:46,000
Audrow Nash: okay. So tell me
more about that.

1029
01:00:46,000 --> 01:00:49,163
Sergey Litvinenko: So pretty
much specific things, we, we do

1030
01:00:49,163 --> 01:00:51,799
not do official training, but we
provide a certain

1031
01:00:51,799 --> 01:00:55,120
recommendations, but pretty much
when an edge case happens, you

1032
01:00:55,120 --> 01:00:58,177
have to have a first responder
team within your company is

1033
01:00:58,177 --> 01:01:01,024
going to get deployed and
collect all the data. Number

1034
01:01:01,024 --> 01:01:03,923
one, number two, you block
communications for a certain

1035
01:01:03,923 --> 01:01:06,980
period of time. And number
three, if you have to report to

1036
01:01:06,980 --> 01:01:09,774
the government or to any
stakeholders, you have to do

1037
01:01:09,774 --> 01:01:12,884
this right away. And you have to
be very explicit with your

1038
01:01:12,884 --> 01:01:16,205
actions to say, Hey, Here's what
happened. We collected all the

1039
01:01:16,205 --> 01:01:19,474
data. We tried to mitigate that
it doesn't spread or whatever.

1040
01:01:19,474 --> 01:01:22,742
we talked to the government. We
notified them. This is exactly

1041
01:01:22,742 --> 01:01:25,958
what happened. And trust me,
most of the time, it's going to,

1042
01:01:25,958 --> 01:01:29,015
in the end, it's going to be
okay. Most of the time. Funky

1043
01:01:29,015 --> 01:01:32,283
things happen when you want to
hide the edge case, or you want

1044
01:01:32,283 --> 01:01:35,604
to hide the fact that you made a
mistake. And it's, such a soft

1045
01:01:35,604 --> 01:01:38,345
thing. It's such a non
engineering thing, but it can

1046
01:01:38,345 --> 01:01:41,666
blow up a lot of things. And we
just provide recommendations to

1047
01:01:41,666 --> 01:01:44,935
robotics companies that, Hey,
Oh, by the way, if, if you don't

1048
01:01:44,935 --> 01:01:48,256
know how to do this, or if you
don't have a way, we at Koop can

1049
01:01:48,256 --> 01:01:51,313
help you. But it's like an extra
service that we of course

1050
01:01:51,313 --> 01:01:53,000
provide on top of our insurance.

1051
01:01:53,000 --> 01:01:57,780
Audrow Nash: Yeah. Makes sense.
And it makes it so they're a

1052
01:01:57,780 --> 01:02:02,719
better customer for you and you
get paid to help them set this

1053
01:02:02,719 --> 01:02:07,498
up and it potentially saves them
if something does occur. So

1054
01:02:07,498 --> 01:02:12,198
that, seems like win-win to me,
as an initiative and having

1055
01:02:12,198 --> 01:02:16,978
those, yeah. 'cause then that,
makes 'em much better in your

1056
01:02:16,978 --> 01:02:22,156
qualitative, analysis. And then.
In the event that something like

1057
01:02:22,156 --> 01:02:26,936
this does occur, they're doing
significant damage mitigation

1058
01:02:26,936 --> 01:02:31,476
because this is in place, and
that makes them not tank as

1059
01:02:31,476 --> 01:02:36,176
significantly as you would, as
they would if they handle it

1060
01:02:36,176 --> 01:02:40,000
badly and lie or cover up or
anything like this.

1061
01:02:40,000 --> 01:02:43,301
Sergey Litvinenko: Just one
example, very simple. We haven't

1062
01:02:43,301 --> 01:02:46,272
experienced that. We haven't
insured that. Very simple

1063
01:02:46,272 --> 01:02:49,793
example. Let's say you have two
mobile robot developers. Both of

1064
01:02:49,793 --> 01:02:52,874
them do, let's say, materials
handling or goods handling

1065
01:02:52,874 --> 01:02:56,065
within a warehouse. Think of
Amazon, right? Like they have

1066
01:02:56,065 --> 01:02:57,000
what Kiva robots?

1067
01:02:57,000 --> 01:02:58,000
Audrow Nash: They have

1068
01:02:58,000 --> 01:03:01,207
Sergey Litvinenko: two, two
similar examples. Let's say you

1069
01:03:01,207 --> 01:03:04,467
have one company. They had some
nasty edge case for whatever

1070
01:03:04,467 --> 01:03:07,457
reason, the whole warehouse
stopped and, it was, Poorly

1071
01:03:07,457 --> 01:03:10,772
handled, poorly communicated.
There was no dispatch that took

1072
01:03:10,772 --> 01:03:13,924
care of that. And the news got
out and there was like two,

1073
01:03:13,924 --> 01:03:17,293
three articles out there that
bashed this company for how poor

1074
01:03:17,293 --> 01:03:20,500
of a job they did. And then when
you have customers coming,

1075
01:03:20,500 --> 01:03:23,489
customer is going to be like,
Oh, you guys had this bad

1076
01:03:23,489 --> 01:03:26,913
accident. I don't want to work
with You what if that happens to

1077
01:03:26,913 --> 01:03:30,174
me? I'm going to go work with
these guys. So that's why it's

1078
01:03:30,174 --> 01:03:33,380
going to haunt you down. If you
don't take care of it, it's

1079
01:03:33,380 --> 01:03:36,261
going to haunt you down
eventually, and it's going to

1080
01:03:36,261 --> 01:03:39,576
cost you money. So we advise our
customers is that, Hey, take

1081
01:03:39,576 --> 01:03:43,000
care of this right now. You're
going to thank yourself 10 years

1082
01:03:43,000 --> 01:03:47,116
Audrow Nash: Oh, definitely.
100%. And I also, I think this

1083
01:03:47,116 --> 01:03:50,953
is this is, it's just a good
exercise that you guys are

1084
01:03:50,953 --> 01:03:55,349
bringing into their perception,
in a sense, because they should

1085
01:03:55,349 --> 01:03:59,535
be considering the worst case
scenario. Scenario, and making

1086
01:03:59,535 --> 01:04:03,581
appropriate plans. Like I, to
me, that just seems wise for

1087
01:04:03,581 --> 01:04:07,977
this kind of thing. It's like in
your personal life, discussing

1088
01:04:07,977 --> 01:04:12,512
the event of a death of yourself
or a loved one or something like

1089
01:04:12,512 --> 01:04:16,419
this can make everything so much
easier in the unlikely,

1090
01:04:16,419 --> 01:04:20,395
hopefully unlikely event that it
does occur. Gotcha. Very

1091
01:04:20,395 --> 01:04:22,000
interesting. let's see.

1092
01:04:22,000 --> 01:04:27,339
Audrow Nash: So now backing out
a little bit, tell me about how

1093
01:04:27,339 --> 01:04:32,169
you guys. are gathering this
behavioral information. What

1094
01:04:32,169 --> 01:04:37,169
does your API look like? What
does it require of the robot?

1095
01:04:37,169 --> 01:04:42,000
what kinds of information are
you typically transmitting?

1096
01:04:42,000 --> 01:04:46,045
Sergey Litvinenko: we collect,
typically we collect data from,

1097
01:04:46,045 --> 01:04:50,156
the customer's cloud, we do the
integration with the cloud, and

1098
01:04:50,156 --> 01:04:53,679
we collect data that they
already have. This is really

1099
01:04:53,679 --> 01:04:57,398
important. We're not trying to
collect, new types of data

1100
01:04:57,398 --> 01:05:01,313
because, from the engineering
standpoint it takes resources.

1101
01:05:01,313 --> 01:05:05,228
But there are two categories of
data that we collect. Number

1102
01:05:05,228 --> 01:05:09,012
one, utilization data. So we
want to know where the robots

1103
01:05:09,012 --> 01:05:13,123
are, how much time or how much
mileage they go for. And, what's

1104
01:05:13,123 --> 01:05:17,168
the general, time online that
they are out there. It will also

1105
01:05:17,168 --> 01:05:21,083
collect the telemetry data. So
the telemetry, data means how

1106
01:05:21,083 --> 01:05:24,672
fast they, how fast they go.
what's the, where are they

1107
01:05:24,672 --> 01:05:28,456
turning, where they're not
turning? what are the, if there

1108
01:05:28,456 --> 01:05:32,502
are any predefined routes, we
also want to know that, but it's

1109
01:05:32,502 --> 01:05:36,547
typically done over qualitative
evaluation and, any hard stops

1110
01:05:36,547 --> 01:05:40,266
or hard accelerations over a
certain G, G means, if they,

1111
01:05:40,266 --> 01:05:43,659
yep, exactly. And, we also
collect the, that's where

1112
01:05:43,659 --> 01:05:47,704
probably the most challenging
part, the field that I mentioned

1113
01:05:47,704 --> 01:05:51,358
about the minimal risk maneuver,
like going offline, the

1114
01:05:51,358 --> 01:05:55,273
troubleshooting rate. we also
collect that, but every single

1115
01:05:55,273 --> 01:05:58,992
customer has a different
definition of that. And for some

1116
01:05:58,992 --> 01:06:02,907
people, it might be that, if it
takes, let's say, More than,

1117
01:06:02,907 --> 01:06:06,952
let's say, I don't know, five
seconds to, pave your, path that

1118
01:06:06,952 --> 01:06:10,541
is considered to be a bad event.
Some people, or like a

1119
01:06:10,541 --> 01:06:14,456
behavioral event. Some people
say, if, my safety zone buffer

1120
01:06:14,456 --> 01:06:18,632
gets violated and I get another
person or a car too close to me,

1121
01:06:18,632 --> 01:06:22,612
it's going to be considered a
risky event. Or some people say

1122
01:06:22,612 --> 01:06:26,527
it could be that the vehicle.
Doesn't know what to do and it

1123
01:06:26,527 --> 01:06:30,311
has to like literally stop and
ask for help and we have to

1124
01:06:30,311 --> 01:06:34,226
Standardize those things within
our model and that's what we

1125
01:06:34,226 --> 01:06:37,880
deal with and you know at the
end of the day We take the

1126
01:06:37,880 --> 01:06:41,599
performance. We take the
telemetry. We kind of layer that

1127
01:06:41,599 --> 01:06:45,644
on top of the utilization so we
know how much vehicles are out

1128
01:06:45,644 --> 01:06:49,624
there and how they're performing
and Then we layer a bunch of

1129
01:06:49,624 --> 01:06:53,148
other things like software
updates and the qualitative

1130
01:06:53,148 --> 01:06:57,323
evaluation And that's what we do
Gets the, gets the underwriting

1131
01:06:57,323 --> 01:07:01,434
done, but yeah, fundamentally,
those are the two key categories

1132
01:07:01,434 --> 01:07:03,000
of data that we'll pull.

1133
01:07:03,000 --> 01:07:07,290
Audrow Nash: It's interesting to
me. It feels like for the

1134
01:07:07,290 --> 01:07:11,877
telemetry data, it feels like
you're treating other domains of

1135
01:07:11,877 --> 01:07:15,575
robotics like autonomous
vehicles, but maybe it's,

1136
01:07:15,575 --> 01:07:19,570
autonomous vehicles like
autonomous cars. is that, Did

1137
01:07:19,570 --> 01:07:23,934
you find that a lot of those
risk factors translate over to

1138
01:07:23,934 --> 01:07:28,595
other domains, or you have the
infrastructure built out? Or how

1139
01:07:28,595 --> 01:07:30,000
do you think of it?

1140
01:07:30,000 --> 01:07:33,989
Sergey Litvinenko: Yeah, so we,
the model that were built

1141
01:07:33,989 --> 01:07:37,768
originally for autonomous
vehicles, we have reasons to

1142
01:07:37,768 --> 01:07:42,107
believe that it is translatable
to other use cases, because at

1143
01:07:42,107 --> 01:07:46,516
the end of the day, our job is
to understand the performance of

1144
01:07:46,516 --> 01:07:50,785
the robot. And that performance,
how it will correlate to the

1145
01:07:50,785 --> 01:07:54,983
eventual potential losses. And,
for example, with autonomous

1146
01:07:54,983 --> 01:07:59,462
vehicles, if let's say you have
a, a behavioral event or a risky

1147
01:07:59,462 --> 01:08:03,241
event, according to our
definition that happens every,

1148
01:08:03,241 --> 01:08:07,440
let's say a hundred thousand
miles, right? we will take that

1149
01:08:07,440 --> 01:08:11,639
and then we will assume. That
there's going to be a certain,

1150
01:08:11,639 --> 01:08:15,698
accident rate coming out of
that, which could be, 10 times

1151
01:08:15,698 --> 01:08:19,617
or a hundred times. It's going
to, it's, depends on the,

1152
01:08:19,617 --> 01:08:23,816
customer and the data. Similar
thing with, let's say, mobile

1153
01:08:23,816 --> 01:08:28,155
robots in closed environments,
or let's say with drones. okay,

1154
01:08:28,155 --> 01:08:32,564
you have one, behavioral event
or one troubleshooting event per

1155
01:08:32,564 --> 01:08:36,483
100, 000 hours worth of
operation, or, we can do this on

1156
01:08:36,483 --> 01:08:40,891
the mileage or we can do flight
hours. So there is usually some

1157
01:08:40,891 --> 01:08:45,440
utilization expressed in mileage
or in hours or a number of tasks

1158
01:08:45,440 --> 01:08:49,639
performed. And there is some
performance metric, whether you

1159
01:08:49,639 --> 01:08:53,908
go offline or you need help, or
you need to be tele operated,

1160
01:08:53,908 --> 01:08:58,037
something like this. And, we
trying to do a really good job

1161
01:08:58,037 --> 01:09:02,096
with categorizing those, there
is a category of. of mobile

1162
01:09:02,096 --> 01:09:06,225
robots, category of drones,
categories of vessels, and then

1163
01:09:06,225 --> 01:09:10,354
within each, we're collecting
data and see how it performs.

1164
01:09:10,354 --> 01:09:14,273
But it's the same model that
applies to all. And that's,

1165
01:09:14,273 --> 01:09:18,541
that's our novelty. That's our
IP. We believe we have a model

1166
01:09:18,541 --> 01:09:22,810
that applies to all kinds of
robots, and we can underwrite on

1167
01:09:22,810 --> 01:09:24,000
the back of that.

1168
01:09:24,000 --> 01:09:28,133
Audrow Nash: hell yeah. So your
model, the way I'm thinking of

1169
01:09:28,133 --> 01:09:31,600
it, is you have your model. Your
model has different

1170
01:09:31,600 --> 01:09:35,667
parameterizations depending on
what domain you're in. Is that

1171
01:09:35,667 --> 01:09:36,000
fair?

1172
01:09:36,000 --> 01:09:37,000
Sergey Litvinenko: That's fair.

1173
01:09:37,000 --> 01:09:41,817
Audrow Nash: Okay, that's pretty
cool. And what is it, is it too

1174
01:09:41,817 --> 01:09:46,183
simplistic to think of your
model as the, you're basically

1175
01:09:46,183 --> 01:09:50,398
doing mean, so over, super
oversimplified, you're doing,

1176
01:09:50,398 --> 01:09:55,140
you're trying to figure out the
expected Value or the mean time

1177
01:09:55,140 --> 01:09:59,806
to failure, say, and then you
have an idea in the domain about

1178
01:09:59,806 --> 01:10:04,473
how costly a failure might be.
And obviously you probably have

1179
01:10:04,473 --> 01:10:08,763
a distribution and you're
handling this in distributions,

1180
01:10:08,763 --> 01:10:12,000
but is it, is it a fair way to
think of it?

1181
01:10:12,000 --> 01:10:15,760
Sergey Litvinenko: Yeah, that's,
fair. Yeah. And you're right. We

1182
01:10:15,760 --> 01:10:19,000
are looking at the, distribution
for every single domain

1183
01:10:19,000 --> 01:10:21,000
Audrow Nash: Yeah, like any good
statistics people.

1184
01:10:21,000 --> 01:10:24,913
Sergey Litvinenko: yep. Yep.
And, there is a distribution of

1185
01:10:24,913 --> 01:10:28,696
the, failure rates and the
accident rates, and the results

1186
01:10:28,696 --> 01:10:32,674
of the distribution for how bad
those are. And, distributions

1187
01:10:32,674 --> 01:10:36,848
look different for different use
cases. maybe. Maybe they all in

1188
01:10:36,848 --> 01:10:40,761
a limit are going to be all
normally distributed and you can

1189
01:10:40,761 --> 01:10:44,217
do like perfect prediction, but
now it's not normally

1190
01:10:44,217 --> 01:10:45,000
distributed.

1191
01:10:45,000 --> 01:10:49,711
Audrow Nash: Yeah, so one, one
thought, and this would be

1192
01:10:49,711 --> 01:10:55,000
funny, or like it'd be wonderful
for robotics. I'm sure that it,

1193
01:10:55,000 --> 01:11:00,041
I'm sure that there's not, I
don't know, in, in the limit, if

1194
01:11:00,041 --> 01:11:05,248
robotics keeps getting better,
will your margin keep decreasing

1195
01:11:05,248 --> 01:11:10,207
for this kind of thing? Or will
there be so many robots that

1196
01:11:10,207 --> 01:11:15,496
it'll be like Home insurance or
something where you just have so

1197
01:11:15,496 --> 01:11:20,702
many entities insured like how
do you view the limit as robotic

1198
01:11:20,702 --> 01:11:25,000
performance improves and how
does that look for Koop

1199
01:11:25,000 --> 01:11:29,386
Sergey Litvinenko: yeah, that's
a really good question. So to

1200
01:11:29,386 --> 01:11:33,771
repeat this back to you, pretty
much if, robots eradicate all

1201
01:11:33,771 --> 01:11:36,000
possible failures and they just

1202
01:11:36,000 --> 01:11:37,000
Audrow Nash: crush it all the
time

1203
01:11:37,000 --> 01:11:41,448
Sergey Litvinenko: or crush it
all the time. Yeah. Why do you

1204
01:11:41,448 --> 01:11:46,188
need insurance? I think, I think
what we're going to come down to

1205
01:11:46,188 --> 01:11:50,052
is, something like the
following. we're going to have

1206
01:11:50,052 --> 01:11:54,573
robots that are going to be so
good that we will likely reduce

1207
01:11:54,573 --> 01:11:59,240
the frequency by 95%, maybe like
98%, okay. And there's going to

1208
01:11:59,240 --> 01:12:03,323
be that remaining. 10 to 1
percent of accidents that are

1209
01:12:03,323 --> 01:12:07,990
still going to be present. Maybe
those will be some form of edge

1210
01:12:07,990 --> 01:12:12,000
cases, which will just be hard
to take care of because,

1211
01:12:12,000 --> 01:12:13,000
Audrow Nash: sure

1212
01:12:12,000 --> 01:12:15,695
Sergey Litvinenko: and when
those happens, the all eyes are

1213
01:12:15,695 --> 01:12:19,139
going to be on the company
behind the failure. and it's

1214
01:12:19,139 --> 01:12:22,959
going to be most likely very
high severity. So we're going to

1215
01:12:22,959 --> 01:12:26,778
go from, for humans, it's okay,
there is some frequency, some

1216
01:12:26,778 --> 01:12:30,285
severity. Okay. For robots, it's
going to be a lot lower

1217
01:12:30,285 --> 01:12:33,729
frequency, a lot higher
severity, because it could be a

1218
01:12:33,729 --> 01:12:37,549
fleet wide outage, okay? If I
have a million mobile robots in

1219
01:12:37,549 --> 01:12:41,557
operation, and, for some reason,
half of them go offline, and I,

1220
01:12:41,557 --> 01:12:45,126
my customers lose a billion
dollars, that's going to be a

1221
01:12:45,126 --> 01:12:49,009
nightmare. The severity is going
to go through the roof. And I

1222
01:12:49,009 --> 01:12:52,891
think we're going to get to the
point where our model is going

1223
01:12:52,891 --> 01:12:56,648
to actually not just we'll turn
into a catastrophe insurance

1224
01:12:56,648 --> 01:13:00,468
model where we're going to be
insuring for like very outsized

1225
01:13:00,468 --> 01:13:04,413
events, but we'll have the data
from the performance to know at

1226
01:13:04,413 --> 01:13:08,358
what frequency and severity of
those might happen. And here we,

1227
01:13:08,358 --> 01:13:11,803
look, we will probably make a
lot less per robot in the

1228
01:13:11,803 --> 01:13:15,372
future, but it's going to be so
many more robots that the

1229
01:13:15,372 --> 01:13:17,000
economics will make sense.

1230
01:13:17,000 --> 01:13:21,233
Audrow Nash: yeah that's what I
would expect too very

1231
01:13:21,233 --> 01:13:26,185
interesting that you may have to
like far down the road in the

1232
01:13:26,185 --> 01:13:31,296
limit it's the catastrophe model
Which, okay, makes sense. And I

1233
01:13:31,296 --> 01:13:36,248
love the idea of, oh, there's
like tons and tons of robots and

1234
01:13:36,248 --> 01:13:40,561
you just charge the smallest
insurance premium on them

1235
01:13:40,561 --> 01:13:45,034
because most of them are
reliable all the time. And it's

1236
01:13:45,034 --> 01:13:49,746
very cool that you get to,
leverage improvements over time.

1237
01:13:49,746 --> 01:13:54,538
Actually, so one question that
leads to is how do you update

1238
01:13:54,538 --> 01:13:58,931
your rate? for companies over
time as their performance

1239
01:13:58,931 --> 01:14:03,643
improves. 'cause I'm imagining
if, say I'm Tesla and I ship

1240
01:14:03,643 --> 01:14:08,435
the, I think, 11 to 12 for the
self-driving, there was a big

1241
01:14:08,435 --> 01:14:13,228
improvement there. And say I'm
insured at this one level and

1242
01:14:13,228 --> 01:14:18,100
now it's a big improvement. or
say we go to 13, it's probably

1243
01:14:18,100 --> 01:14:23,291
that they don't wanna pay the 11
rates when they're on. Full self

1244
01:14:23,291 --> 01:14:27,844
driving 13 How do you handle
adjustments over time as the

1245
01:14:27,844 --> 01:14:30,000
robots become more capable?

1246
01:14:30,000 --> 01:14:34,015
Sergey Litvinenko: the way we
approach this at Koop, it's a

1247
01:14:34,015 --> 01:14:38,098
function of volume of data that
we have. If we have a lot of

1248
01:14:38,098 --> 01:14:42,318
data coming, let's say, over the
course of the month, we would

1249
01:14:42,318 --> 01:14:46,401
look at updating the rates on a
monthly basis. The way we do

1250
01:14:46,401 --> 01:14:50,757
this right now is that we update
rates on a yearly basis because

1251
01:14:50,757 --> 01:14:54,976
over the course of the year, we
collect enough data to see how

1252
01:14:54,976 --> 01:14:58,651
the customers perform and
specifically to see if there

1253
01:14:58,651 --> 01:15:03,074
were any losses that happened in
that period, if there were major

1254
01:15:03,074 --> 01:15:07,294
improvements in the performance,
if there are major changes in

1255
01:15:07,294 --> 01:15:11,513
the qualitative questionnaires.
Now, if we, if, we, let's say,

1256
01:15:11,513 --> 01:15:15,801
let's say go from, getting 200
customers per year, every single

1257
01:15:15,801 --> 01:15:19,884
customer sharing some form of
data with us to getting a, 200

1258
01:15:19,884 --> 01:15:23,967
customers per month, then is
going to warrant a, higher rate

1259
01:15:23,967 --> 01:15:28,255
of updates on our end. So it's a
function of data. So right now

1260
01:15:28,255 --> 01:15:32,202
we're all set, which is also the
industry standard and the

1261
01:15:32,202 --> 01:15:36,013
insurance that you review the
rates on a yearly basis. I

1262
01:15:36,013 --> 01:15:40,300
think. Give it probably another
year, and a half. I think we're

1263
01:15:40,300 --> 01:15:44,179
going to do this on a, a
biannual basis. So twice a year,

1264
01:15:44,179 --> 01:15:48,058
the goal, I think the goal, the
minimum, like the highest

1265
01:15:48,058 --> 01:15:52,346
frequency of updates that we can
do will be on a monthly basis,

1266
01:15:52,346 --> 01:15:55,000
most likely. But we were not
there yet.

1267
01:15:55,000 --> 01:15:58,684
Audrow Nash: Yeah I would love
to see like in the absolute

1268
01:15:58,684 --> 01:16:02,686
limit that every maneuver that
the robot does is a small charge

1269
01:16:02,686 --> 01:16:06,180
For their insurance and it's
like highly scaled to what

1270
01:16:06,180 --> 01:16:10,119
they're doing right now Like I
think that would be the coolest

1271
01:16:10,119 --> 01:16:13,803
thing. The robot is navigating
between a crowd. that's six

1272
01:16:13,803 --> 01:16:17,551
cents or something like that.
And, it's highly usage based,

1273
01:16:17,551 --> 01:16:21,363
like API calls basically, but
for insurance. But that, yeah,

1274
01:16:21,363 --> 01:16:25,174
monthly is already way faster
for this kind of thing. And I,

1275
01:16:25,174 --> 01:16:26,000
oh, go ahead.

1276
01:16:26,000 --> 01:16:29,545
Sergey Litvinenko: do it. Yeah,
something like that, like usage

1277
01:16:29,545 --> 01:16:32,810
or maneuver based or risk
scenario based, insurance model.

1278
01:16:32,810 --> 01:16:36,242
like technically we can do it
right now. The reason why we're

1279
01:16:36,242 --> 01:16:39,000
not doing this is because
there's just not enough

1280
01:16:39,000 --> 01:16:40,000
Audrow Nash: fit the model.

1281
01:16:40,000 --> 01:16:41,000
Sergey Litvinenko: there. Yeah.
Yep.

1282
01:16:41,000 --> 01:16:45,197
Audrow Nash: Yeah. It makes
sense. That's super cool. let's

1283
01:16:45,197 --> 01:16:49,252
see. I wanted to get your
thoughts on some, so this might

1284
01:16:49,252 --> 01:16:53,805
be a complete. ridiculous idea,
but it seems to me that you guys

1285
01:16:53,805 --> 01:16:58,215
are getting a lot of really good
data about robotics companies

1286
01:16:58,215 --> 01:17:02,413
while they're operating. It
seems like a very good place to

1287
01:17:02,413 --> 01:17:06,681
have, a venture arm to your
company, where you invest in the

1288
01:17:06,681 --> 01:17:10,665
companies that are doing really
well. if I was a venture

1289
01:17:10,665 --> 01:17:14,862
capitalist, and I had access to
the data that you guys have

1290
01:17:14,862 --> 01:17:16,000
access to insure

1291
01:17:16,000 --> 01:17:17,000
Sergey Litvinenko: huh.

1292
01:17:16,000 --> 01:17:20,349
Audrow Nash: it would, you would
be making such intelligent bets.

1293
01:17:20,349 --> 01:17:24,229
I'm just, I wonder what your
thoughts are on this, because

1294
01:17:24,229 --> 01:17:28,310
you, can see their performance
curve over time, where they're

1295
01:17:28,310 --> 01:17:31,454
improving the, rates of
accidents, the rates of

1296
01:17:31,454 --> 01:17:35,602
accidents or bad performance are
decreasing. I, would think it

1297
01:17:35,602 --> 01:17:39,616
lends itself really well to
making investments yourself as a

1298
01:17:39,616 --> 01:17:43,764
company in the companies that
you are insuring. and know a lot

1299
01:17:43,764 --> 01:17:47,444
about but what are your thoughts
on that is that a like

1300
01:17:47,444 --> 01:17:51,324
incredibly stupid idea or
doesn't work because of conflict

1301
01:17:51,324 --> 01:17:54,000
of interest or what are you any
thoughts

1302
01:17:54,000 --> 01:17:58,333
Sergey Litvinenko: you're really
thinking this through, but I got

1303
01:17:58,333 --> 01:18:02,333
to tell you that we, we haven't
thought about becoming a, VC

1304
01:18:02,333 --> 01:18:06,000
shop ourselves, but we partnered
with a lot of other VC

1305
01:18:06,000 --> 01:18:07,000
Audrow Nash: oh

1306
01:18:07,000 --> 01:18:10,627
Sergey Litvinenko: And, it's not
a formal partnership. It's

1307
01:18:10,627 --> 01:18:14,130
actually, if you look online,
you're not going to find it

1308
01:18:14,130 --> 01:18:17,880
anywhere. That's something that
we do, I would say privately,

1309
01:18:17,880 --> 01:18:21,629
but we'll usually have. A, a
venture capitalist, that invests

1310
01:18:21,629 --> 01:18:25,317
in robotics companies and then
that venture capitalist would

1311
01:18:25,317 --> 01:18:29,067
want their portfolio companies
to work with Koop for reasons.

1312
01:18:29,067 --> 01:18:33,062
Number one, we can keep track of
them. Number two, we can protect

1313
01:18:33,062 --> 01:18:36,750
them. And, that's how they also
do this because they want to

1314
01:18:36,750 --> 01:18:40,316
protect their investment, and
it's, it's, they made a bet.

1315
01:18:40,316 --> 01:18:43,942
That this company is gonna
become a billion dollar company.

1316
01:18:43,942 --> 01:18:47,507
There's gonna be so many hiccups
on the way that paying, a

1317
01:18:47,507 --> 01:18:51,257
certain amount of money per year
just to protect yourself and

1318
01:18:51,257 --> 01:18:54,945
protect your attack is a no
brainer. And I can tell you. we,

1319
01:18:54,945 --> 01:18:58,449
of course, work with the
investors that invested in Koop,

1320
01:18:58,449 --> 01:19:01,952
that's AlleyCorp Ubiquiti,
Hyundai is an investor, Fusion

1321
01:19:01,952 --> 01:19:05,825
Funds, we work with all of them,
but we also work with non Koop

1322
01:19:05,825 --> 01:19:09,513
investors, and we have 10 funds
signed up so far, and all of

1323
01:19:09,513 --> 01:19:13,017
them are like, mid stage, not
mid stage, but mid size, so

1324
01:19:13,017 --> 01:19:14,000
anywhere between

1325
01:19:14,000 --> 01:19:18,909
Audrow Nash: they're doing like
series a kind of thing series a

1326
01:19:18,909 --> 01:19:20,000
or below maybe

1327
01:19:20,000 --> 01:19:23,283
Sergey Litvinenko: So you're
really thinking this through.

1328
01:19:23,283 --> 01:19:26,396
You're right. There is
definitely the investor angle. I

1329
01:19:26,396 --> 01:19:29,679
don't think we're big enough for
it to become a, to open a

1330
01:19:29,679 --> 01:19:32,000
venture arm ourselves, but look,
why not?

1331
01:19:32,000 --> 01:19:36,048
Audrow Nash: you have such good
data. It's a huge competitive

1332
01:19:36,048 --> 01:19:39,830
advantage to other venture arms,
or other venture capital

1333
01:19:39,830 --> 01:19:43,945
companies, I would think. to me
it just makes, it makes really

1334
01:19:43,945 --> 01:19:48,059
good sense to try to be involved
in the venture capital scene,

1335
01:19:48,059 --> 01:19:51,709
for this kind of thing. But
because also you just get a

1336
01:19:51,709 --> 01:19:55,624
little additional, you get some
revenue information for the

1337
01:19:55,624 --> 01:19:59,672
company and then, you know about
their customers and then you

1338
01:19:59,672 --> 01:20:03,123
have all the usage data. it
seems like an incredible

1339
01:20:03,123 --> 01:20:06,905
competitive advantage as a
venture company. I hope to see

1340
01:20:06,905 --> 01:20:10,621
you guys, doing this pivot
because I think it could be a

1341
01:20:10,621 --> 01:20:14,536
really good fit. It would be
like the most informed venture

1342
01:20:14,536 --> 01:20:18,584
capital and I know, one model
that could be done with this is

1343
01:20:18,584 --> 01:20:22,101
that you sell the data that
you're, collecting on the

1344
01:20:22,101 --> 01:20:26,282
company and of course let them
know and maybe give them some of

1345
01:20:26,282 --> 01:20:30,197
the profit too, but you could
sell this to venture firms to

1346
01:20:30,197 --> 01:20:34,444
make their decisions. That would
be one, but you would get a lot

1347
01:20:34,444 --> 01:20:38,359
less of the upside if you know
that a company is absolutely

1348
01:20:38,359 --> 01:20:42,606
killing it. and so I would think
that would why, be why you want

1349
01:20:42,606 --> 01:20:44,000
your own venture arm.

1350
01:20:44,000 --> 01:20:46,966
Sergey Litvinenko: Yeah, I can
tell you, data products are

1351
01:20:46,966 --> 01:20:50,136
something that we have on our
longer term roadmap. And I think

1352
01:20:50,136 --> 01:20:53,000
data products could be very
profitable. Very profitable.

1353
01:20:53,000 --> 01:20:57,865
Audrow Nash: think so too. Yeah.
And you could even like it,

1354
01:20:57,865 --> 01:21:02,973
probably you have Koop, but you
set up a second company that is

1355
01:21:02,973 --> 01:21:08,000
just a venture capital company
and then you pass the data over

1356
01:21:08,000 --> 01:21:12,297
and you integrate it really
well. and that would be a

1357
01:21:12,297 --> 01:21:17,162
wonderful opportunity. I would
imagine. Seems awesome. super

1358
01:21:17,162 --> 01:21:21,784
cool. it, seems like a really
good thing for the robotics

1359
01:21:21,784 --> 01:21:26,081
because also, It seems like a
lot of finance is about

1360
01:21:26,081 --> 01:21:31,027
intelligent capital allocation
and, you guys have really good

1361
01:21:31,027 --> 01:21:35,568
information. And if you're
assessing risk and you have a

1362
01:21:35,568 --> 01:21:40,432
model for risk, that seems like
a wonderful setup for making

1363
01:21:40,432 --> 01:21:44,000
intelligent allocations from my
perspective.

1364
01:21:44,000 --> 01:21:48,479
Sergey Litvinenko: Yeah. It's
also, you're correct. I think

1365
01:21:48,479 --> 01:21:53,337
the challenge with that is going
to be if we, let's say, collect

1366
01:21:53,337 --> 01:21:58,044
data from early stage companies
and we see that ones are doing

1367
01:21:58,044 --> 01:22:02,371
better, the other are doing
worse, and then the ones that

1368
01:22:02,371 --> 01:22:06,774
are doing worse change something
and then they start doing

1369
01:22:06,774 --> 01:22:11,556
better, it's like the, you have
to be able to track the rate of

1370
01:22:11,556 --> 01:22:16,415
change really, good because the,
you can make a bet on something

1371
01:22:16,415 --> 01:22:20,893
that you thought was bad, but
then. Things changed over the

1372
01:22:20,893 --> 01:22:25,524
course of the month. but I think
in general, having access to

1373
01:22:25,524 --> 01:22:30,079
data products, that you can use
however you want anonymized,

1374
01:22:30,079 --> 01:22:34,634
maybe by sector or by the robot
type or something like that.

1375
01:22:34,634 --> 01:22:36,000
it's a great idea.

1376
01:22:36,000 --> 01:22:40,713
Audrow Nash: Yeah, for sure. I
think that's super, super cool.

1377
01:22:40,713 --> 01:22:45,579
so now one other thing I wanted
to get your opinion on. it seems

1378
01:22:45,579 --> 01:22:50,140
So I, most of my interviews are
with companies like robotics

1379
01:22:50,140 --> 01:22:54,550
companies. I'm talking to
technologists. but it strikes me

1380
01:22:54,550 --> 01:22:59,263
by talking with you. and you're
definitely a technologist too,

1381
01:22:59,263 --> 01:23:03,673
but like you're not making a
physical robot. It strikes me

1382
01:23:03,673 --> 01:23:07,626
that there's a lot of
opportunity in robotics around

1383
01:23:07,626 --> 01:23:12,339
the infrastructure required to
make robotics companies go. And

1384
01:23:12,339 --> 01:23:16,749
I think one large domain for
this is probably finance. And

1385
01:23:16,749 --> 01:23:21,310
you mentioned at the beginning,
like the ability to get like

1386
01:23:21,310 --> 01:23:25,491
good models for getting
equipment and things like this.

1387
01:23:25,491 --> 01:23:29,977
but I wanted to see if you have
other financial, like other

1388
01:23:29,977 --> 01:23:34,386
opportunities in the financial
space or. Other spaces that

1389
01:23:34,386 --> 01:23:39,099
aren't directly building a robot
and selling it for a use case

1390
01:23:39,099 --> 01:23:43,509
that you think are good, but
they're not quite things that

1391
01:23:43,509 --> 01:23:48,070
you have the time to pursue,
given that you're already doing

1392
01:23:48,070 --> 01:23:52,480
Koop and there's a lot of
exciting direction that you guys

1393
01:23:52,480 --> 01:23:54,000
can keep going down.

1394
01:23:54,000 --> 01:23:58,387
Sergey Litvinenko: if I great
question, I think that actually

1395
01:23:58,387 --> 01:24:02,917
your question implies Something
even bigger is like hey what is

1396
01:24:02,917 --> 01:24:07,448
the vision for Koop? Okay, and
that kind of those two questions

1397
01:24:07,448 --> 01:24:11,834
overlap. You're right There is
this financial layer and there

1398
01:24:11,834 --> 01:24:15,933
is the robot financing. So
whenever you purchase a robot,

1399
01:24:15,933 --> 01:24:20,176
you can finance and split the
payments for it. Okay. so you

1400
01:24:20,176 --> 01:24:24,491
can literally like open a bank
and you can provide leases on

1401
01:24:24,491 --> 01:24:28,949
the robots, make it very easy to
buy one. Two, warranty. don't

1402
01:24:28,949 --> 01:24:33,336
forget that the warranties that
you provide with your robots,

1403
01:24:33,336 --> 01:24:37,866
they're backed by somebody. if
you're a big company, you can do

1404
01:24:37,866 --> 01:24:42,397
it yourself. If you're another
big company, you have to partner

1405
01:24:42,397 --> 01:24:46,855
with a warranty underwriter, or
Also could be very profitable.

1406
01:24:46,855 --> 01:24:51,314
Three, insurance. In addition to
insurance, I would also add a

1407
01:24:51,314 --> 01:24:55,629
compliance automation software.
Robotics companies, they are

1408
01:24:55,629 --> 01:24:59,943
required to be compliant with
different standards, like ISO,

1409
01:24:59,943 --> 01:25:04,474
SOC2, or SOC in general, NIST.
and see, there are lots of them,

1410
01:25:04,474 --> 01:25:08,860
like just automating that and
making that as easy as possible

1411
01:25:08,860 --> 01:25:13,031
and as integrated as possible
with your systems is another

1412
01:25:13,031 --> 01:25:17,274
tremendous opportunity. And that
can be done with insurance

1413
01:25:17,274 --> 01:25:19,000
together, actually, they

1414
01:25:19,000 --> 01:25:20,000
Audrow Nash: goes hand in hand.

1415
01:25:20,000 --> 01:25:23,408
Sergey Litvinenko: Yeah. And
another thing, which you

1416
01:25:23,408 --> 01:25:26,881
mentioned is data products.
Okay, now data products is

1417
01:25:26,881 --> 01:25:30,547
probably going to be more
technical, out of all of those.

1418
01:25:30,547 --> 01:25:34,534
but the data products could be
sold to all kinds of interested

1419
01:25:34,534 --> 01:25:38,392
parties, governments, insurance
companies, like maybe legacy

1420
01:25:38,392 --> 01:25:40,000
insurance companies, any,

1421
01:25:40,000 --> 01:25:43,084
Audrow Nash: insurance companies
is a funny way to put it, and I

1422
01:25:43,084 --> 01:25:44,000
think you're right.

1423
01:25:44,000 --> 01:25:47,963
Sergey Litvinenko: Yeah. so when
you have. Equipment Financing,

1424
01:25:47,963 --> 01:25:51,611
Robot Financing, you have
Warranty, Insurance, Compliance,

1425
01:25:51,611 --> 01:25:55,574
and Data Products. Those are the
five things that I think is of

1426
01:25:55,574 --> 01:25:59,285
that layer. And one of the
visions, or one, one vision that

1427
01:25:59,285 --> 01:26:02,933
I have for Koop is that Koop
will, at scale, be able to do

1428
01:26:02,933 --> 01:26:06,896
all of those things. And be a,
true go to Financial partner and

1429
01:26:06,896 --> 01:26:10,481
data partner and insurance
partner and compliance partner

1430
01:26:10,481 --> 01:26:14,381
for the robotics industry. And
you have multiple offerings and

1431
01:26:14,381 --> 01:26:18,343
they all can talk to each other
because, if I ensure the robots

1432
01:26:18,343 --> 01:26:22,117
that you sell, I want to know
the financing terms, or I want

1433
01:26:22,117 --> 01:26:25,828
to know, I want to use that,
that certain data products for

1434
01:26:25,828 --> 01:26:28,973
that, or I want to use the,
whatever you do on the

1435
01:26:28,973 --> 01:26:32,496
compliance front. So you're
exactly right that, there is

1436
01:26:32,496 --> 01:26:36,270
more. than what we just do. And
I think at scale is going to

1437
01:26:36,270 --> 01:26:39,981
become very important that you
have that layer functioning.

1438
01:26:39,981 --> 01:26:43,000
Hopefully we can be that layer
for the industry.

1439
01:26:43,000 --> 01:26:47,076
Audrow Nash: That'll be awesome.
And if you do that, you'll get a

1440
01:26:47,076 --> 01:26:50,212
huge advantage from the
integration you're able to

1441
01:26:50,212 --> 01:26:54,038
provide between those services.
like that, will be wonderful.

1442
01:26:54,038 --> 01:26:57,863
And it, I wonder, I don't know
too much about the space and I

1443
01:26:57,863 --> 01:27:01,312
wonder how much overlap there
is, between I don't know,

1444
01:27:01,312 --> 01:27:04,887
helping with financing and
helping with, one of the other

1445
01:27:04,887 --> 01:27:08,211
parts that you've mentioned.
But, there might be some

1446
01:27:08,211 --> 01:27:12,225
overlap. I'm sure there is. That
will one big thing is you don't

1447
01:27:12,225 --> 01:27:16,113
have to do customer discovery,
and that's a big advantage. You

1448
01:27:16,113 --> 01:27:19,688
just say, Oh, by the way, we
also have this, you probably

1449
01:27:19,688 --> 01:27:23,764
need this, too. and that's I bet
you cost per customer is a thing

1450
01:27:23,764 --> 01:27:27,590
that is very difficult to break
into this market. Like it's a

1451
01:27:27,590 --> 01:27:31,540
You have to spend a lot to get a
customer, but then if you make

1452
01:27:31,540 --> 01:27:35,178
this zero because you already
have the customer because of

1453
01:27:35,178 --> 01:27:38,000
another service, then that's a
big advantage.

1454
01:27:38,000 --> 01:27:41,443
Sergey Litvinenko: And I can
tell you the biggest benefit

1455
01:27:41,443 --> 01:27:45,067
across doing those things under
one company is going to be a

1456
01:27:45,067 --> 01:27:47,000
unified system of record. So for

1457
01:27:47,000 --> 01:27:49,000
Audrow Nash: Yes. Oh

1458
01:27:49,000 --> 01:27:50,000
Sergey Litvinenko: I finance

1459
01:27:49,000 --> 01:27:50,000
Audrow Nash: you're

1460
01:27:50,000 --> 01:27:53,124
Sergey Litvinenko: when I
finance your, robots, I know

1461
01:27:53,124 --> 01:27:56,769
exactly what you have. So now I
don't have to bother you when I

1462
01:27:56,769 --> 01:28:00,066
need to provide you with
insurance because I know all the

1463
01:28:00,066 --> 01:28:03,190
robots. Same applies to
warranty. Okay. Or when, let's

1464
01:28:03,190 --> 01:28:06,719
say you do compliance, you don't
have to fill out a technical

1465
01:28:06,719 --> 01:28:10,074
questionnaire for insurance
because we're going to pull it

1466
01:28:10,074 --> 01:28:11,000
from compliance.

1467
01:28:11,000 --> 01:28:15,084
Audrow Nash: then strap on a
venture arm and you guys are set

1468
01:28:15,084 --> 01:28:19,102
for this kind of thing. That
way, it would be such a benefit

1469
01:28:19,102 --> 01:28:23,119
to the robotics space too. and
having a larger, like most of

1470
01:28:23,119 --> 01:28:27,136
the companies that I've talked
or investment firms that I've

1471
01:28:27,136 --> 01:28:30,819
talked to That invest in
robotics are small and they do

1472
01:28:30,819 --> 01:28:35,104
series A funding, and then when
you get bigger, it becomes a bit

1473
01:28:35,104 --> 01:28:39,322
easier to, go with regular large
investment firms. But Series B

1474
01:28:39,322 --> 01:28:42,670
is very hard typically for
robotics companies as I

1475
01:28:42,670 --> 01:28:46,687
understand it. and then like
when you get to like d and e or

1476
01:28:46,687 --> 01:28:50,570
whatever, if you go that far,
then it's easier because you

1477
01:28:50,570 --> 01:28:54,788
just look at. regular business
numbers, but it would be nice to

1478
01:28:54,788 --> 01:28:58,739
have a larger player in the
robotics space that helps with,

1479
01:28:58,739 --> 01:29:02,488
that is well suited to fund
later rounds. so if you guys

1480
01:29:02,488 --> 01:29:06,706
grow that direction, I think it
will be a very good thing, even

1481
01:29:06,706 --> 01:29:10,991
the insurance and everything. I
think it's a very good thing for

1482
01:29:10,991 --> 01:29:13,000
the robotics space in general.

1483
01:29:13,000 --> 01:29:16,205
Sergey Litvinenko: And we can
even, add to the financial

1484
01:29:16,205 --> 01:29:19,582
layer, a venture arm. We don't
even have to be like, we can

1485
01:29:19,582 --> 01:29:22,902
definitely be a lead investor,
but we can also bring other

1486
01:29:22,902 --> 01:29:25,879
people on board, like a
syndicate, and the syndicate

1487
01:29:25,879 --> 01:29:29,542
will make an investment decision
based on what we know about the

1488
01:29:29,542 --> 01:29:30,000
company.

1489
01:29:30,000 --> 01:29:32,760
Audrow Nash: yes, because you
can really make it so you get

1490
01:29:32,760 --> 01:29:35,520
great reports and you have all
the information and you have

1491
01:29:35,520 --> 01:29:38,000
your risk models. That's all
very useful information.

1492
01:29:38,000 --> 01:29:39,000
Sergey Litvinenko: Yep.

1493
01:29:39,000 --> 01:29:41,000
Audrow Nash: Okay, hell yeah.

1494
01:29:41,000 --> 01:29:45,850
Audrow Nash: so we've talked
about all those, like the

1495
01:29:45,850 --> 01:29:51,419
potential directions that Koop
could go. what do you, where do

1496
01:29:51,419 --> 01:29:56,000
you see yourself going in the
next five years, say?

1497
01:29:56,000 --> 01:29:59,823
Sergey Litvinenko: I believe
that, The, so first of all,

1498
01:29:59,823 --> 01:30:03,782
we're making a bet on the
robotics industry and we believe

1499
01:30:03,782 --> 01:30:08,014
that over the next decade Robots
are going to go from where we

1500
01:30:08,014 --> 01:30:12,109
are today to being 10x more
ubiquitous. Okay, hopefully 100x

1501
01:30:12,109 --> 01:30:16,273
So that's the bet that we're
making across all the use cases.

1502
01:30:16,273 --> 01:30:20,437
Okay, so I believe that the Koop
is going to become a company

1503
01:30:20,437 --> 01:30:24,328
that will consolidate certain
non core functions, such as

1504
01:30:24,328 --> 01:30:28,560
insurance, compliance, and data
products and financing in one.

1505
01:30:28,560 --> 01:30:32,792
It's going to be one company,
one platform, one partner that's

1506
01:30:32,792 --> 01:30:36,410
going to make it very
economically attractive for the

1507
01:30:36,410 --> 01:30:40,164
robotics companies and as a
result for the customers of

1508
01:30:40,164 --> 01:30:44,464
robotics companies. And we want,
we do not want to be a company

1509
01:30:44,464 --> 01:30:48,833
that does one thing. We want to
be a platform, like a layer that

1510
01:30:48,833 --> 01:30:53,133
takes care of all of that. And
like you said, we're going to be

1511
01:30:53,133 --> 01:30:57,092
producing value, through,
different means, but at the end,

1512
01:30:57,092 --> 01:31:01,324
it's going to come down to sell
your robots cheaper, sell your

1513
01:31:01,324 --> 01:31:05,556
robots faster, sell your robots
easier. That's what it's going

1514
01:31:05,556 --> 01:31:09,925
to come down to. So if you have
a robotics company and you don't

1515
01:31:09,925 --> 01:31:14,157
have Koop, it's going to be a
problem. That's the vision. that

1516
01:31:14,157 --> 01:31:18,184
we're driving towards. And,
yeah, I think, we're definitely

1517
01:31:18,184 --> 01:31:22,280
on our way. I think probably
happens sooner than five years.

1518
01:31:22,280 --> 01:31:26,307
but becoming that brand name
that everybody knows about and

1519
01:31:26,307 --> 01:31:30,744
that nobody's companies can live
without. That's definitely very,

1520
01:31:30,744 --> 01:31:34,771
if we do this, it's going to be
fantastic. I'll be proud of

1521
01:31:34,771 --> 01:31:36,000
this, of the team.

1522
01:31:36,000 --> 01:31:40,895
Audrow Nash: Hell yeah. Yeah,
that'll be great. any, do you

1523
01:31:40,895 --> 01:31:46,039
have any things, like for our
watchers and listeners, is there

1524
01:31:46,039 --> 01:31:51,017
any major takeaways that you
hope, if they get anything from

1525
01:31:51,017 --> 01:31:55,000
this episode, what do you hope
they get from it?

1526
01:31:55,000 --> 01:31:58,964
Sergey Litvinenko: I think, it's
something that I think about

1527
01:31:58,964 --> 01:32:02,408
every day and it maybe doesn't
just apply to robotics

1528
01:32:02,408 --> 01:32:05,852
companies, it applies to any,
startups or early stage

1529
01:32:05,852 --> 01:32:09,946
companies out there. I sometimes
find myself thinking about the

1530
01:32:09,946 --> 01:32:14,105
product more than thinking about
the customer. and you can build

1531
01:32:14,105 --> 01:32:18,004
a lot of cool stuff, but unless
it gets the job done for the

1532
01:32:18,004 --> 01:32:21,708
customer, it doesn't matter.
there is a funny meme on the

1533
01:32:21,708 --> 01:32:25,606
internet where you have like a
Mario and then you have some,

1534
01:32:25,606 --> 01:32:29,570
your product, and then you have
a Mario on like steroids. And

1535
01:32:29,570 --> 01:32:33,404
it's this is your Mario as a
customer. This, thing, is your

1536
01:32:33,404 --> 01:32:37,433
product, right? And then when
you combine those together, this

1537
01:32:37,433 --> 01:32:41,137
is the cool stuff that your
customers can do. You need to

1538
01:32:41,137 --> 01:32:44,711
focus on the cool stuff your
customers can do, not on a

1539
01:32:44,711 --> 01:32:48,610
product. And I would encourage
all the robotics companies to

1540
01:32:48,610 --> 01:32:52,379
think super, super hard about
their customers, how they're

1541
01:32:52,379 --> 01:32:55,888
going to commercialize, what
problems they're solving.

1542
01:32:55,888 --> 01:32:59,982
Sometimes I'm just seeing robots
for the sake of robots. And it

1543
01:32:59,982 --> 01:33:03,881
really makes me upset because,
you have such talented people

1544
01:33:03,881 --> 01:33:07,975
that could be solving some big
problems, but they're working on

1545
01:33:07,975 --> 01:33:11,939
the robot that just looks like a
cool robot without any clear

1546
01:33:11,939 --> 01:33:15,383
application. And, that's
probably one thing. And that

1547
01:33:15,383 --> 01:33:19,477
applies to everybody, not just
robotics companies, also applies

1548
01:33:19,477 --> 01:33:23,505
to us, probably to any, every
single one out there. that's one

1549
01:33:23,505 --> 01:33:25,000
thing that I can share.

1550
01:33:25,000 --> 01:33:28,944
Audrow Nash: All right. Hell
yeah. Let's see. This was a

1551
01:33:28,944 --> 01:33:33,169
blast. Wonderful hearing about
what you're doing. seems like

1552
01:33:33,169 --> 01:33:37,606
it's a great initiative for the
robotics community and, I think

1553
01:33:37,606 --> 01:33:40,000
it'll make a lot of things
easier.

1554
01:33:40,000 --> 01:33:44,000
Sergey Litvinenko: Thank you.
And we'll do our best.

1555
01:33:44,000 --> 01:33:47,673
Audrow Nash: Hell yeah. All
right. Bye everyone. And that

1556
01:33:47,673 --> 01:33:51,669
wraps up our conversation with
Sergey Litvinenko from Koop AI.

1557
01:33:51,669 --> 01:33:55,342
I don't know about you, but I'm
really excited to see how

1558
01:33:55,342 --> 01:33:59,015
companies like Koop AI will
shape the future of robotics.

1559
01:33:59,015 --> 01:34:02,753
It's fascinating to see how
something as seemingly mundane

1560
01:34:02,753 --> 01:34:06,812
as insurance could actually be a
key factor in accelerating the

1561
01:34:06,812 --> 01:34:10,743
adoption of robots in our daily
lives. So I want to leave you

1562
01:34:10,743 --> 01:34:14,481
with a question: How do you
think that increasing presence

1563
01:34:14,481 --> 01:34:18,154
of robots in our world will
change the way we think about

1564
01:34:18,154 --> 01:34:21,634
risk and insurance? And what
other industries might be

1565
01:34:21,634 --> 01:34:25,565
transformed by the growth of
robotics insurance? If you enjoy

1566
01:34:25,565 --> 01:34:29,560
this episode, please share it
with a friend who's into tech or

1567
01:34:29,560 --> 01:34:33,556
robots. And as always, I'd love
to hear your thoughts. You can

1568
01:34:33,556 --> 01:34:37,229
reach me on social media or
leave a comment on the video.

1569
01:34:37,229 --> 01:34:40,000
Thank you for listening. See you
next time.
