1
00:00:00,000 --> 00:00:03,490
Audrow Nash: Like, I think the
feeling is because of things

2
00:00:03,490 --> 00:00:07,099
like ChatGPT that there's no
reason to go into something like

3
00:00:07,099 --> 00:00:10,885
computer science because you're
just going to get in a thing and

4
00:00:10,885 --> 00:00:14,552
be automated right away for this
kind of thing. I think that's

5
00:00:14,552 --> 00:00:18,042
the feeling. I wanted to hear
your thoughts on it. Like, do

6
00:00:18,042 --> 00:00:21,000
you think it's a good time to be
a roboticist, or,

7
00:00:21,000 --> 00:00:25,986
Melonee Wise: I think it's a
great time to be a roboticist.

8
00:00:25,986 --> 00:00:31,056
This is the, probably the next
50 years are going to be like

9
00:00:31,056 --> 00:00:33,000
the heyday of robotics.

10
00:00:33,000 --> 00:00:36,943
Audrow Nash: Are humanoids the
next big thing? How long before

11
00:00:36,943 --> 00:00:41,014
they take our jobs? To get some
perspective on this, I talk with

12
00:00:41,014 --> 00:00:44,894
Agility Robotics CTO, Melonee
Wise. Her answers will probably

13
00:00:44,894 --> 00:00:48,710
surprise you. I think you'll
like this interview if you want

14
00:00:48,710 --> 00:00:52,145
to understand the current
technology around humanoids,

15
00:00:52,145 --> 00:00:56,088
what's possible, and where the
opportunities are. I also think

16
00:00:56,088 --> 00:01:00,223
you'll like it if you're curious
about how LLMs like ChatGPT will

17
00:01:00,223 --> 00:01:03,784
impact robotics and jobs, or if
you'd like to know about

18
00:01:03,784 --> 00:01:07,219
manufacturing in the US,
especially the challenges and

19
00:01:07,219 --> 00:01:10,972
opportunities. As always, I'm
Audro Nash. This is the Audro

20
00:01:10,972 --> 00:01:15,042
Nash Podcast. After you listen,
I'd love to know. on X or in the

21
00:01:15,042 --> 00:01:18,095
comments if you agree or
disagree with Melonee's

22
00:01:18,095 --> 00:01:21,975
perspective on the timeline for
humanoids and where we'll see

23
00:01:21,975 --> 00:01:25,855
them first. Also, if you want to
talk about this interview or

24
00:01:25,855 --> 00:01:29,671
robotics in general, I host a
weekly space on X on Thursdays

25
00:01:29,671 --> 00:01:33,615
at 9 p. m. Eastern time or 6 p.
m. Pacific time. It's free and

26
00:01:33,615 --> 00:01:37,240
has been a lot of fun. All
right, here's the interview. I

27
00:01:37,240 --> 00:01:41,311
hope you find this conversation
as enjoyable and enlightening as

28
00:01:41,311 --> 00:01:45,000
I did. Alright. Hi, Melonee.
Would you introduce yourself?

29
00:01:45,000 --> 00:01:50,194
Melonee Wise: Hi, Audrow. I'm
Melonee Wise, CTO of Agility

30
00:01:50,194 --> 00:01:51,000
Robotics.

31
00:01:51,000 --> 00:01:54,370
Audrow Nash: Now, last time we
talked, you were at Fetch. tell

32
00:01:54,370 --> 00:01:57,848
me a bit about what's happened,
like, your path, for the last, I

33
00:01:57,848 --> 00:02:01,000
don't know, year and a half
since we've done an interview?

34
00:02:01,000 --> 00:02:06,746
Melonee Wise: Yeah, sure. well,
last time we talked, Fetch had

35
00:02:06,746 --> 00:02:12,307
just been acquired. I spent
about a year and a half, working

36
00:02:12,307 --> 00:02:18,146
at Zebra Technologies and then I
decided to take some time off.

37
00:02:18,146 --> 00:02:23,244
So about six months off,
traveled the world, that was a

38
00:02:23,244 --> 00:02:28,990
lot of fun. Went all over Asia
Pacific, went to Antarctica and

39
00:02:28,990 --> 00:02:34,459
then South America. Then I
decided I should get back in the

40
00:02:34,459 --> 00:02:39,000
game. And, I decided to go join
Agility Robotics.

41
00:02:39,000 --> 00:02:42,851
Audrow Nash: would you give a
little bit of background on

42
00:02:42,851 --> 00:02:44,000
Agility Robotics?

43
00:02:44,000 --> 00:02:47,949
Melonee Wise: So, Agility
Robotics is a mobile

44
00:02:47,949 --> 00:02:53,529
manipulation company. So, it has
a humanoid ish Form Factor. That

45
00:02:53,529 --> 00:02:59,023
is targeting, tote manipulation
for machine assisted operations.

46
00:02:59,023 --> 00:03:03,144
So, every time you order
something or online, it

47
00:03:03,144 --> 00:03:07,951
potentially passed through a
tote. and Digit, our mobile

48
00:03:07,951 --> 00:03:12,587
manipulation robot, is one of
the robots that might be

49
00:03:12,587 --> 00:03:17,394
handling a tote that has
something you ordered in it. Or

50
00:03:17,394 --> 00:03:21,000
even parts for a thing that you
might buy.

51
00:03:21,000 --> 00:03:23,268
Audrow Nash: Oh yeah, and so a
tote, it's just a small bin? Is

52
00:03:23,268 --> 00:03:24,000
that what a tote is?

53
00:03:24,000 --> 00:03:28,870
Melonee Wise: Yeah, a tote is a
plastic container that it can

54
00:03:28,870 --> 00:03:33,660
be, it can range in sizes quite
a bit, so it can be anything

55
00:03:33,660 --> 00:03:38,529
from like 2 feet by 2 feet to
like 10 inches by 10 inches or,

56
00:03:38,529 --> 00:03:43,000
so the range on the size of the
container is very large.

57
00:03:43,000 --> 00:03:47,102
Audrow Nash: Mm hmm. Okay. And
so you have Digit. Digit's a

58
00:03:47,102 --> 00:03:51,342
humanoid. Digit is helping with
these TOTE related tasks. And

59
00:03:51,342 --> 00:03:55,652
you mentioned it's humanoid ish.
Can you tell me what it looks

60
00:03:55,652 --> 00:03:56,000
like?

61
00:03:56,000 --> 00:04:01,996
Melonee Wise: Yeah, so Digit, is
probably about five Feet two or

62
00:04:01,996 --> 00:04:07,804
three inches tall. it has two,
six degree of freedom arms as a

63
00:04:07,804 --> 00:04:13,613
head with LED lights for a face.
and it has two legs. however,

64
00:04:13,613 --> 00:04:19,047
the leg architecture is a little
bit reversed from the leg

65
00:04:19,047 --> 00:04:24,387
architecture that we have as
people. then a lot of people

66
00:04:24,387 --> 00:04:29,446
would say the knees are
backwards. However, Because it

67
00:04:29,446 --> 00:04:34,411
has more of an avian leg
structure, it's actually the

68
00:04:34,411 --> 00:04:39,939
ankles are backwards. so it's,
it's a little bit different,

69
00:04:39,939 --> 00:04:42,000
from that perspective.

70
00:04:42,000 --> 00:04:45,279
Audrow Nash: Yeah, I think of it
like having ostrich legs or

71
00:04:45,279 --> 00:04:48,503
something like that. Because
when it kind of, and it's cool

72
00:04:48,503 --> 00:04:51,727
because I was imagining while
watching it that it's kind of

73
00:04:51,727 --> 00:04:55,060
pragmatic to have the legs go
backwards, because then you can

74
00:04:55,060 --> 00:04:58,230
have it drop its and keep the
arms in front of it, but you

75
00:04:58,230 --> 00:05:01,508
don't have the knees occluding
whatever the arms may want to

76
00:05:01,508 --> 00:05:02,000
get into?

77
00:05:02,000 --> 00:05:06,870
Melonee Wise: Yeah. The knees
don't get in the way. And it

78
00:05:06,870 --> 00:05:11,824
also biases is the center of
gravity to drop straight down,

79
00:05:11,824 --> 00:05:13,000
which is nice.

80
00:05:13,000 --> 00:05:18,031
Audrow Nash: It's got a nice
squat. Hell yeah. Okay. And so

81
00:05:18,031 --> 00:05:22,806
how did you pick to come to
agility? Tell me a bit about

82
00:05:22,806 --> 00:05:24,000
that decision.

83
00:05:24,000 --> 00:05:28,760
Melonee Wise: so when I, when I,
you know, looked out in the

84
00:05:28,760 --> 00:05:33,757
robotic space, there were a lot
of things that I was interested

85
00:05:33,757 --> 00:05:38,913
in. I thought very briefly about
starting my own company, another

86
00:05:38,913 --> 00:05:43,752
one, and I decided I wanted a
little bit of a break. and so I

87
00:05:43,752 --> 00:05:48,036
thought I'd take a less
stressful role, as like a CTO,

88
00:05:48,036 --> 00:05:53,034
not a CEO, cause that's a very
stressful role. And I spoke with

89
00:05:53,034 --> 00:05:57,952
a couple of different companies,
um, and the thing that really

90
00:05:57,952 --> 00:06:02,870
excited me about agility were,
was a couple of things. One, it

91
00:06:02,870 --> 00:06:07,788
was very much in a market that
was similar to Fetch, so I knew

92
00:06:07,788 --> 00:06:12,310
a lot of the customers, I knew a
lot of the, the customer

93
00:06:12,310 --> 00:06:17,228
challenges, and why people buy.
Robots. so it helped me really

94
00:06:17,228 --> 00:06:21,909
understand whether I believe
that the product had a product

95
00:06:21,909 --> 00:06:26,351
market fit and whether people
would even want to buy the

96
00:06:26,351 --> 00:06:31,190
thing. Two, Agility's product is
relatively mature. They have

97
00:06:31,190 --> 00:06:36,267
physical robots. They have been
working with customers. I didn't

98
00:06:36,267 --> 00:06:40,947
want to start with a company
that was still in the does the

99
00:06:40,947 --> 00:06:45,945
technology even work, you know,
past the first prototype stage.

100
00:06:45,945 --> 00:06:50,863
and so that was another aspect
of why I was very interested in

101
00:06:50,863 --> 00:06:55,464
joining Agility. And then the,
the third thing was, is You

102
00:06:55,464 --> 00:06:59,589
know, one of the, one of the
things I've always been

103
00:06:59,589 --> 00:07:03,873
interested in is mobile
manipulation. I mean, at Fetch

104
00:07:03,873 --> 00:07:08,712
we had a mobile manipulator, but
also as we worked in the AMR

105
00:07:08,712 --> 00:07:13,550
market, it was very clear that
there was, there's a whole set

106
00:07:13,550 --> 00:07:18,548
of tasks that are better suited
towards mobile manipulation and

107
00:07:18,548 --> 00:07:22,673
not just mobile. And I thought
that that would be an

108
00:07:22,673 --> 00:07:27,353
interesting, Direction to go.
And if you're wondering why I

109
00:07:27,353 --> 00:07:32,192
didn't go and do more mobile,
well, I had a non compete. So I

110
00:07:32,192 --> 00:07:36,000
couldn't just go off and do
another AMR company.

111
00:07:36,000 --> 00:07:38,864
Audrow Nash: That's very funny.
How long does, I mean, I'm just

112
00:07:38,864 --> 00:07:41,682
curious about the non competes.
How long does that last? Is it

113
00:07:41,682 --> 00:07:44,000
like for five years you can't be
in an AMR company?

114
00:07:44,000 --> 00:07:48,168
Melonee Wise: Yeah, so my non
compete, because I was a key

115
00:07:48,168 --> 00:07:51,976
person in a material
transaction, as they call it, my

116
00:07:51,976 --> 00:07:56,000
non compete is three years from
the date of acquisition.

117
00:07:56,000 --> 00:07:58,942
Audrow Nash: Gotcha. That's
pretty interesting as a

118
00:07:58,942 --> 00:08:02,000
constraint for how to pick the
next robotics company.

119
00:08:02,000 --> 00:08:06,621
Melonee Wise: Well, it It's, it
is a material constraint, you

120
00:08:06,621 --> 00:08:07,000
know,

121
00:08:07,000 --> 00:08:11,509
Audrow Nash: yeah, for sure.
Okay. So then because of this, I

122
00:08:11,509 --> 00:08:15,278
guess, because of the
constraint, then digit having

123
00:08:15,278 --> 00:08:19,565
legs makes it so that it's a
different product in a sense.

124
00:08:19,565 --> 00:08:24,000
And so the non compete is not
valid. for this kind of thing?

125
00:08:24,000 --> 00:08:29,025
Melonee Wise: Yeah, because my
non compete was very specific to

126
00:08:29,025 --> 00:08:33,810
autonomous mobile robots, so
AMRs, and didn't apply to other

127
00:08:33,810 --> 00:08:37,000
types of mobile manipulation
technology.

128
00:08:37,000 --> 00:08:41,357
Audrow Nash: Okay, and so I
interviewed Jonathan Hurst a

129
00:08:41,357 --> 00:08:46,337
number of years ago, who is one
of, is he one of the founders of

130
00:08:46,337 --> 00:08:51,317
Agility? And so he's a professor
at Oregon State, right? Who was

131
00:08:51,317 --> 00:08:56,141
deeply involved. Yeah, one of
the founders and everything. and

132
00:08:56,141 --> 00:09:00,499
so when we talked, Agility
Robotics was really about the

133
00:09:00,499 --> 00:09:05,323
last. My last last hundred feet
of delivery this kind of thing

134
00:09:05,323 --> 00:09:09,914
so like going from the FedEx
truck to the curb I guess curb

135
00:09:09,914 --> 00:09:14,893
to doorstep or something. And I
understand from talking with you

136
00:09:14,893 --> 00:09:19,251
earlier, I think at ROSCon or
some other time that it's,

137
00:09:19,251 --> 00:09:24,075
pivoted just a little bit. Can
you tell me kind of what market

138
00:09:24,075 --> 00:09:28,744
you guys are going into and also
why not? Last little bit of

139
00:09:28,744 --> 00:09:31,000
delivery, this kind of thing.

140
00:09:31,000 --> 00:09:36,261
Melonee Wise: So, like all
startups, our product intentions

141
00:09:36,261 --> 00:09:41,612
have evolved. and when you look
at last mile delivery or the

142
00:09:41,612 --> 00:09:47,141
last hundred feet of delivery,
let's call it, there's a lot of

143
00:09:47,141 --> 00:09:52,313
challenges that go beyond the
robotics technology. Safety,

144
00:09:52,313 --> 00:09:57,753
compliance, Weather, many of
the, many of the challenges that

145
00:09:57,753 --> 00:10:03,371
you see, even in the autonomous
car market, Digit would have to

146
00:10:03,371 --> 00:10:08,722
face if it was going to do the
last hundred feet. And you're

147
00:10:08,722 --> 00:10:13,805
seeing today that, you know, a
lot of companies have been

148
00:10:13,805 --> 00:10:18,621
successful in indoor semi
structured environments. And

149
00:10:18,621 --> 00:10:24,150
before I even got to Agility,
they were already, converging on

150
00:10:24,150 --> 00:10:29,590
product market fit and product
alignment that was more geared

151
00:10:29,590 --> 00:10:34,316
towards indoor Material
Handling. And since I joined,

152
00:10:34,316 --> 00:10:40,023
I've been working with the team
to really get us focused down on

153
00:10:40,023 --> 00:10:45,463
a set of solutions that, really
play to Digit's strengths and

154
00:10:45,463 --> 00:10:50,635
are scalable and repeatable
within our customer sites. And

155
00:10:50,635 --> 00:10:56,075
this is, this is something that,
you know, many startups have

156
00:10:56,075 --> 00:11:01,515
done in the AMR space, Locus,
Fetch, Automotors, we all ended

157
00:11:01,515 --> 00:11:07,044
up getting into kind of a set of
repeatable workflows that our

158
00:11:07,044 --> 00:11:12,484
customers really were excited
about. And we're doing the same

159
00:11:12,484 --> 00:11:14,000
thing at Agility.

160
00:11:14,000 --> 00:11:18,065
Audrow Nash: Hell yeah, and so
that workflow for Agility is the

161
00:11:18,065 --> 00:11:20,000
moving totes. Is that correct?

162
00:11:20,000 --> 00:11:23,500
Melonee Wise: Yeah. And there's
tons of workflows in the

163
00:11:23,500 --> 00:11:26,000
warehouse that involve moving
totes. Oh,

164
00:11:26,000 --> 00:11:30,161
Audrow Nash: hell yeah. And
then, so, I mean, like just

165
00:11:30,161 --> 00:11:35,077
thinking about humanoids, why is
a, like, so a thing that strikes

166
00:11:35,077 --> 00:11:39,843
me is that perhaps. And I'd love
to hear kind of the, why, the,

167
00:11:39,843 --> 00:11:44,306
why Digit is a very good fit for
this, but like I imagine a

168
00:11:44,306 --> 00:11:48,694
humanoid is going to be more
expensive than a robot with a

169
00:11:48,694 --> 00:11:53,006
wheeled base, walking is a
challenge, probably limits the

170
00:11:53,006 --> 00:11:57,696
payload, um, why a humanoid Form
factor for this kind of task.

171
00:11:57,696 --> 00:12:02,310
Why not just a mobile base with
a big pinch, like, like a two

172
00:12:02,310 --> 00:12:07,000
foot pinch gripper that could
just grab the totes and move on.

173
00:12:07,000 --> 00:12:11,790
Melonee Wise: So as someone who
spent a lot of time building

174
00:12:11,790 --> 00:12:16,739
mobile robots, one of the things
that you'll find is there's a

175
00:12:16,739 --> 00:12:21,369
lot of companies, including
Fetch and Automotors who built

176
00:12:21,369 --> 00:12:26,558
base platforms and then got into
the business of building lots of

177
00:12:26,558 --> 00:12:30,789
accessories, right? You can see
that with Fetch, Geek

178
00:12:30,789 --> 00:12:35,898
Automotors, and even, Locus now
with, with their acquisition of,

179
00:12:35,898 --> 00:12:40,688
Waypoint, right? Whatever. but,
but the thing is, is that Is

180
00:12:40,688 --> 00:12:45,558
that when you look at, that the
accessories start to eat into

181
00:12:45,558 --> 00:12:50,108
the payload, right? So like,
when you add, so like if you

182
00:12:50,108 --> 00:12:54,499
look at a lot of autonomous
mobile robots, right? Like,

183
00:12:54,499 --> 00:12:59,209
everyone wants to put a hundred
kilograms or more on top of

184
00:12:59,209 --> 00:13:04,238
these things, but every bit of
shelving or cart or whatever you

185
00:13:04,238 --> 00:13:09,188
put on there reduces the payload
that you can put on. And then

186
00:13:09,188 --> 00:13:14,057
the next thing is, so say you
want to put a hundred kilograms

187
00:13:14,057 --> 00:13:19,007
of You know, e commerce goods on
there. Now it usually gets to

188
00:13:19,007 --> 00:13:24,036
some conveyor end point and they
typically want you to push the

189
00:13:24,036 --> 00:13:28,586
tote off onto the conveyor. But
now you have to build the

190
00:13:28,586 --> 00:13:33,376
conveyor, uh, you know, tool,
and typically they want to put

191
00:13:33,376 --> 00:13:38,325
in multi bins. And so now you
need like multi layer conveyors.

192
00:13:38,325 --> 00:13:43,035
And then you end up in this
situation where you're building

193
00:13:43,035 --> 00:13:47,586
all of this mechanism to do
these like really complicated

194
00:13:47,586 --> 00:13:52,216
things. And. all at the same
time, you're trying to battle

195
00:13:52,216 --> 00:13:56,846
the other problem of having a
small footprint because most

196
00:13:56,846 --> 00:14:01,556
warehouses are not meant to
have, you know, 10 foot aisles,

197
00:14:01,556 --> 00:14:06,745
they're meant to have relatively
narrow aisles because people are

198
00:14:06,745 --> 00:14:11,694
relatively narrow, and you're
also fighting against stability.

199
00:14:11,694 --> 00:14:16,803
So, as you put mass higher on a
very small platform, it wants to

200
00:14:16,803 --> 00:14:21,673
tip a lot more. And so one of
the nice things about legged or

201
00:14:21,673 --> 00:14:26,782
dynamically balancing systems is
as you reach higher, you can do

202
00:14:26,782 --> 00:14:31,891
other things with your stability
platform to enable you to reach

203
00:14:31,891 --> 00:14:36,841
and pull weight from, you know,
kind of outstretched positions

204
00:14:36,841 --> 00:14:41,471
back into your footprint. And
so. One of the advantages of

205
00:14:41,471 --> 00:14:46,500
having a dynamically balancing
system or a bipedal robot is you

206
00:14:46,500 --> 00:14:50,651
can have a relatively small
footprint, you can reach

207
00:14:50,651 --> 00:14:55,361
relatively high, and you can
carry a relatively competitive

208
00:14:55,361 --> 00:14:56,000
payload.

209
00:14:56,000 --> 00:15:00,032
Audrow Nash: Very cool. Yeah,
it's an interesting I guess the

210
00:15:00,032 --> 00:15:04,197
alternative of an AMR has a lot
of trade offs that you run into

211
00:15:04,197 --> 00:15:07,965
and so a humanoid is very
flexible in what it can do. You

212
00:15:07,965 --> 00:15:11,468
can change the dynamics. It's,
it's almost like these

213
00:15:11,468 --> 00:15:14,641
environments were made for
people, which is very

214
00:15:14,641 --> 00:15:18,740
interesting and so you can kind
of leverage that. What do you,

215
00:15:18,740 --> 00:15:22,706
what do you think about that
perspective where it's like one

216
00:15:22,706 --> 00:15:26,474
of the big benefits of a
humanoid form factor is that the

217
00:15:26,474 --> 00:15:30,639
Majority of the world and a lot
of the infrastructure was built

218
00:15:30,639 --> 00:15:34,869
around people. So the aisles and
the warehouses are thin so that

219
00:15:34,869 --> 00:15:38,571
a people can, a person can walk
through, not a big robot

220
00:15:38,571 --> 00:15:42,339
carrier, like with a huge
footprint. what do you think of

221
00:15:42,339 --> 00:15:43,000
that idea?

222
00:15:43,000 --> 00:15:47,256
Melonee Wise: Yeah. I mean,
that's just the reality of the

223
00:15:47,256 --> 00:15:51,658
world, right? Like. And the
thing is, is that it's very hard

224
00:15:51,658 --> 00:15:55,987
and costly to change over these,
these environments for the

225
00:15:55,987 --> 00:16:00,463
infrastructure, right? And our
customers are very targeted at

226
00:16:00,463 --> 00:16:04,939
return on investment in under
two years. And so remodeling an

227
00:16:04,939 --> 00:16:09,415
entire warehouse. For a robot,
typically puts you in the 5 10

228
00:16:09,415 --> 00:16:13,744
year return on investment time
frame. But if you can drop a

229
00:16:13,744 --> 00:16:18,000
robot right in, your return on
investment is pretty rapid.

230
00:16:18,000 --> 00:16:22,220
Audrow Nash: That's awesome. And
so, our how does Digit compare

231
00:16:22,220 --> 00:16:26,105
to people in terms of speed of
doing tasks or this kind of

232
00:16:26,105 --> 00:16:30,392
thing? Or what's, maybe it's not
even a fair comparison, but how

233
00:16:30,392 --> 00:16:32,000
do you think about that?

234
00:16:32,000 --> 00:16:36,567
Melonee Wise: Yeah, so,
typically it's not about Direct

235
00:16:36,567 --> 00:16:41,466
Speed Comparison, it's about
throughput comparison. Because

236
00:16:41,466 --> 00:16:45,949
what you'll see when people do
tasks, they do a lot of

237
00:16:45,949 --> 00:16:51,181
compound, rapid tasks, and then
they wait around a lot of time.

238
00:16:51,181 --> 00:16:55,000
A lot of time They do. people
are really good.

239
00:16:55,000 --> 00:16:57,838
Audrow Nash: So it's like the
tortoise and the hare Yeah. Kind

240
00:16:57,838 --> 00:17:00,493
of thing. So person goes and
does like six things at once.

241
00:17:00,493 --> 00:17:03,239
Yeah. And then they just like
take a 15 minute break kind of

242
00:17:03,239 --> 00:17:05,757
thing, I guess. Whereas the
robot can be doing like one

243
00:17:05,757 --> 00:17:08,000
every three minutes or something
like that. Yeah.

244
00:17:08,000 --> 00:17:13,153
Melonee Wise: You find, you find
if you go and watch the way

245
00:17:13,153 --> 00:17:18,477
people do. A standard activity
is it's very bursty, where they

246
00:17:18,477 --> 00:17:23,630
have lots of like activity and
then no activity. and so what

247
00:17:23,630 --> 00:17:28,868
our customers really care about
is total, total activity over

248
00:17:28,868 --> 00:17:34,279
some timeframe. The other thing
that you see is, is people take

249
00:17:34,279 --> 00:17:39,603
breaks, right? Like people over
the course of a, an eight hour

250
00:17:39,603 --> 00:17:44,842
day. Take at least one hour of
breaks. And so that's a lot of

251
00:17:44,842 --> 00:17:50,166
time for a robot to catch up in
as well, um, in terms of total

252
00:17:50,166 --> 00:17:53,000
utilization and total
throughput.

253
00:17:53,000 --> 00:17:57,309
Audrow Nash: What, on the order
of how does it compare? So if

254
00:17:57,309 --> 00:18:01,617
you consider a long period of
hours, what, what are, I guess,

255
00:18:01,617 --> 00:18:05,714
long enough so that the breaks
and everything average out,

256
00:18:05,714 --> 00:18:09,952
what's kind of the ratio of
throughput, for digit versus say

257
00:18:09,952 --> 00:18:12,000
a human doing a similar role?

258
00:18:12,000 --> 00:18:17,589
Melonee Wise: Yeah, so we don't
have I would say a large enough

259
00:18:17,589 --> 00:18:23,000
data set to make any like claims
on the direct claims on that

260
00:18:23,000 --> 00:18:28,000
Audrow Nash: Give any any
guesses like is it 1 to 1 is it?

261
00:18:28,000 --> 00:18:33,058
Melonee Wise: Yeah 1 to 2 is I
mean, I would say that that as

262
00:18:33,058 --> 00:18:37,452
we Continue to go into the
field. We are at parity or

263
00:18:37,452 --> 00:18:42,427
slightly better over the long
run then Then in the use cases

264
00:18:42,427 --> 00:18:47,153
we're deployed in. So like,
obviously there are some high

265
00:18:47,153 --> 00:18:52,543
speed activities that Digit most
likely will not be doing anytime

266
00:18:52,543 --> 00:18:57,683
soon, but in the use cases that
our customers have us targeted

267
00:18:57,683 --> 00:19:01,000
at, we are at parity or slightly
better.

268
00:19:01,000 --> 00:19:05,408
Audrow Nash: That's amazing. Oh
yeah. And then how do you think

269
00:19:05,408 --> 00:19:09,256
of like, while we're on the
topic of people doing these

270
00:19:09,256 --> 00:19:13,594
jobs, I, so I hosted a space on
X and so we talked a bit about

271
00:19:13,594 --> 00:19:17,372
this as kind of like about
humanoids and like food for

272
00:19:17,372 --> 00:19:21,710
thought for this interview. one
of the concerns brought up was

273
00:19:21,710 --> 00:19:25,348
kind of the ethical one of
having people moved, like

274
00:19:25,348 --> 00:19:29,686
basically replacing people in
different jobs. And I brought up

275
00:19:29,686 --> 00:19:34,094
labor shortages and things like
this, but I'd love to hear your

276
00:19:34,094 --> 00:19:37,942
perspective on kind of the
ethics of robots, especially

277
00:19:37,942 --> 00:19:42,000
humanoid robots, and kind of the
complexities there. Yeah.

278
00:19:42,000 --> 00:19:46,868
Melonee Wise: So let, let's,
let's separate some of the

279
00:19:46,868 --> 00:19:52,356
concerns. So let's start with
just labor in general. So when I

280
00:19:52,356 --> 00:19:57,933
first started in this industry,
the logistics and manufacturing

281
00:19:57,933 --> 00:20:02,978
industry, robotics, kind of
focused. In, in the 2014 2016

282
00:20:02,978 --> 00:20:08,555
timeframe, there were about 600,
000 jobs available. So, and in

283
00:20:08,555 --> 00:20:14,132
that time, all of the robotics
companies that you know and love

284
00:20:14,132 --> 00:20:19,000
have been deploying robots. Into
that, that environment

285
00:20:19,000 --> 00:20:20,000
Audrow Nash: and as fast as they
can,

286
00:20:20,000 --> 00:20:25,579
Melonee Wise: as fast as they
can. I mean, I don't know if

287
00:20:25,579 --> 00:20:31,159
you've seen the latest numbers
from Amazon with their Kiva

288
00:20:31,159 --> 00:20:36,642
fleet. It's like 750, 000
robots. They've deployed. Yeah.

289
00:20:36,642 --> 00:20:42,414
Remarkable. So, so, since then,
the labor, the labor gap has

290
00:20:42,414 --> 00:20:47,801
grown to a million unfulfilled
jobs. So, so we have been

291
00:20:47,801 --> 00:20:53,861
throwing robots at the problem
as fast as we can. And the labor

292
00:20:53,861 --> 00:20:59,633
shortage has 000 jobs. So that,
that's something to food for

293
00:20:59,633 --> 00:21:03,000
thought. however. The, the thing
is

294
00:21:03,000 --> 00:21:06,362
Audrow Nash: 400, 000 and since
when did you say that? That's

295
00:21:06,362 --> 00:21:09,504
since 2014? Between 2016 and
today. That's bonkers. Oh my

296
00:21:09,504 --> 00:21:10,000
goodness.

297
00:21:10,000 --> 00:21:12,000
Melonee Wise: Yeah, just look at
the Bureau of Labor Statistics.

298
00:21:12,000 --> 00:21:15,759
Audrow Nash: I think it's the,
is it the, the baby boomers

299
00:21:15,759 --> 00:21:19,000
retiring? Is that what, one of
the forces on this?

300
00:21:19,000 --> 00:21:24,399
Melonee Wise: It's the aging
labor population. It's a, lack

301
00:21:24,399 --> 00:21:30,072
of interest in the jobs. it's a,
wage pressure problem. Right?

302
00:21:30,072 --> 00:21:35,928
and it's also just, you know, I,
I'm one of them and you are one

303
00:21:35,928 --> 00:21:41,327
of them. As our parents told us
that when we would grow up,

304
00:21:41,327 --> 00:21:47,092
yeah, we would grow up, we were
going to be special flowers and

305
00:21:47,092 --> 00:21:52,399
go to college and do whatever
our hopes and dreams were. I

306
00:21:52,399 --> 00:21:58,072
don't know about you, but my,
none of my hopes and dreams were

307
00:21:58,072 --> 00:22:01,000
to work in a warehouse. but, but

308
00:22:01,000 --> 00:22:04,706
Audrow Nash: that is, what a way
to put it. Yeah. But yes, very

309
00:22:04,706 --> 00:22:05,000
true.

310
00:22:05,000 --> 00:22:10,309
Melonee Wise: That's one aspect
of it. But I, I think, you know,

311
00:22:10,309 --> 00:22:15,120
on the flip side of it is as
time approaches infinity. And

312
00:22:15,120 --> 00:22:20,097
robots approach infinity. We
eventually are going to have to

313
00:22:20,097 --> 00:22:24,742
have a conversation about the
nature of certain jobs and

314
00:22:24,742 --> 00:22:29,803
whether people do those jobs,
right? Technology is continuing

315
00:22:29,803 --> 00:22:35,028
to prove that it can create more
jobs. Less than six months ago

316
00:22:35,028 --> 00:22:39,840
or nine months ago, at this
point, nine months ago, no one

317
00:22:39,840 --> 00:22:44,734
had ever heard of a prompt
engineer. And now it's like. One

318
00:22:44,734 --> 00:22:49,628
of the the more random jobs out
there for the kind of Quote

319
00:22:49,628 --> 00:22:54,190
unquote artificial intelligence
economy that we have So

320
00:22:54,190 --> 00:22:59,582
technology is creating jobs. The
problem is is whether the people

321
00:22:59,582 --> 00:23:04,559
who are currently doing These
other jobs have the skill set,

322
00:23:04,559 --> 00:23:09,702
the training, and the capability
to move into these new, these

323
00:23:09,702 --> 00:23:14,513
new labor positions. And the
problem is, is that's a socio

324
00:23:14,513 --> 00:23:19,076
political issue in large
respect. There is some onus on

325
00:23:19,076 --> 00:23:23,887
robotics professionals to
develop tools that are easier to

326
00:23:23,887 --> 00:23:28,781
use. and that can be cross
trained from someone who's doing

327
00:23:28,781 --> 00:23:34,007
a warehousing job, but we've got
a long way to go. I mean, most

328
00:23:34,007 --> 00:23:38,818
people struggle to use their
iPhones and web browsers. And

329
00:23:38,818 --> 00:23:43,878
we're talking about like robots
are basically iPhones and web

330
00:23:43,878 --> 00:23:48,938
browsers on steroids with legs
and arms and actually, and so,

331
00:23:48,938 --> 00:23:53,418
although, So, although
technology is going to continue

332
00:23:53,418 --> 00:23:58,063
to create more jobs, we have a
problem of helping people

333
00:23:58,063 --> 00:24:02,791
transition into those jobs long
term. And eventually, the

334
00:24:02,791 --> 00:24:08,017
definition of work is going to
change. And I, you know, this is

335
00:24:08,017 --> 00:24:13,160
a hard one for me because from
a, from a personal perspective,

336
00:24:13,160 --> 00:24:18,137
I believe in your universal
basic income, and I believe that

337
00:24:18,137 --> 00:24:23,363
the way we solve this problem is
changing the way we view work.

338
00:24:23,363 --> 00:24:28,506
and creating a social safety net
and then creating a basis for

339
00:24:28,506 --> 00:24:33,732
everyone to be, you know, live.
Yeah. To live. And, and if they

340
00:24:33,732 --> 00:24:38,792
want more, they can earn more.
Like we create an economy that

341
00:24:38,792 --> 00:24:43,272
allows for that. and that, that,
that doesn't preclude

342
00:24:43,272 --> 00:24:48,000
capitalism. It just, Yeah, it
enables a basis for living.

343
00:24:48,000 --> 00:24:52,208
Audrow Nash: It raises the
bottom but keeps, you can still

344
00:24:52,208 --> 00:24:56,850
be super ambitious. Yeah. yeah,
I know, I feel like that is such

345
00:24:56,850 --> 00:25:01,203
a complex issue and I don't, I
don't know where to go on it.

346
00:25:01,203 --> 00:25:05,338
Yeah. Because I've heard size
for everything. What do you

347
00:25:05,338 --> 00:25:09,546
think, so related to all the
prompt engineers and stuff, I

348
00:25:09,546 --> 00:25:13,898
feel Like, like we all thought
all the blue collar jobs were

349
00:25:13,898 --> 00:25:18,324
going to be automated and then
here comes ChatGPT and then it

350
00:25:18,324 --> 00:25:22,459
looks like a lot of the white
collar jobs are going to be

351
00:25:22,459 --> 00:25:26,594
automated. how, cause, and then
that would be interesting

352
00:25:26,594 --> 00:25:30,946
because the people that are
doing fairly sophisticated work,

353
00:25:30,946 --> 00:25:34,791
it's like, okay, now go learn
how to be a plumber and

354
00:25:34,791 --> 00:25:38,854
electrician from a lawyer or
something like this. there,

355
00:25:38,854 --> 00:25:43,134
there was a very funny South
Park on this a little bit ago.

356
00:25:43,134 --> 00:25:47,559
what do you think about that?
Like, what are your thoughts on

357
00:25:47,559 --> 00:25:52,202
this in general, all the AI and
knowledge workers and automation

358
00:25:52,202 --> 00:25:53,000
in general?

359
00:25:53,000 --> 00:25:58,049
Melonee Wise: Yeah, I, I think
that it has come a long way.

360
00:25:58,049 --> 00:26:03,355
It's very interesting. I think
that there is a class of worker

361
00:26:03,355 --> 00:26:08,661
out there today that is going to
struggle, specifically in the

362
00:26:08,661 --> 00:26:13,710
area of knowledge consolidation.
And so that's where you're

363
00:26:13,710 --> 00:26:19,016
seeing a lot of the pain right
now. in, in like, you know, you

364
00:26:19,016 --> 00:26:24,322
can go to ChatGPT and ask for
basically a travel schedule, you

365
00:26:24,322 --> 00:26:29,029
know, that challenges travel
agents. You can go ask for

366
00:26:29,029 --> 00:26:34,249
consolidation of legal law, you
know, like legal Opinions for

367
00:26:34,249 --> 00:26:38,871
law arguments, right? And
that's, in my mind, a lot of

368
00:26:38,871 --> 00:26:43,406
what ChatGPT is good at right
now is consolidation of

369
00:26:43,406 --> 00:26:48,370
knowledge. And so any job that
basically is that activity,

370
00:26:48,370 --> 00:26:53,334
yeah, there's a, there's a
problem, but what we are seeing

371
00:26:53,334 --> 00:26:57,869
is ChatGPT struggles with
creative tasks, algorithmic

372
00:26:57,869 --> 00:27:02,319
tasks in some ways, ironically
enough. Well, complex

373
00:27:02,319 --> 00:27:07,625
algorithmic tasks, anything that
needs, intuition or intuitive

374
00:27:07,625 --> 00:27:12,675
thought, anything that needs
specialized knowledge. And so,

375
00:27:12,675 --> 00:27:17,809
it's, yeah, it's hollowing out
the center, but I don't think

376
00:27:17,809 --> 00:27:23,115
that it is getting to the point
where it, It's challenging the

377
00:27:23,115 --> 00:27:28,250
specialized knowledge sets or
anything that requires a human

378
00:27:28,250 --> 00:27:32,529
interface or reasoning
contextually in this social

379
00:27:32,529 --> 00:27:37,749
framework, right? Like a lot of
like what lawyers and doctors

380
00:27:37,749 --> 00:27:43,055
and other people do is they're
making decisions not only based

381
00:27:43,055 --> 00:27:48,447
on case law. But also the social
context, the social framework,

382
00:27:48,447 --> 00:27:53,582
and the emotions of what is
happening at the time. You know,

383
00:27:53,582 --> 00:27:58,288
can you imagine, ChatGPT
diagnosing you with cancer and

384
00:27:58,288 --> 00:28:00,000
just reading it out?

385
00:28:00,000 --> 00:28:02,951
Audrow Nash: Like, oh my god,
right, that little prompt pops

386
00:28:02,951 --> 00:28:05,705
up and yeah, it's like,
congratulations! Like a sad face

387
00:28:05,705 --> 00:28:06,000
emoji.

388
00:28:06,000 --> 00:28:10,579
Melonee Wise: Yeah, you have
cancer. So I think that there,

389
00:28:10,579 --> 00:28:15,392
there is, there is a We're not
there yet. It would probably or

390
00:28:15,392 --> 00:28:20,126
might get there. It depends on
how we evolve as a society and

391
00:28:20,126 --> 00:28:24,706
how comfortable we get with
these kind of interactions. You

392
00:28:24,706 --> 00:28:29,130
can see some of that in some
cultures that are more robot

393
00:28:29,130 --> 00:28:33,244
facing or more technology
adopting. But I think where

394
00:28:33,244 --> 00:28:37,357
we're seeing it right now in a
lot of the pain is the

395
00:28:37,357 --> 00:28:42,014
consolidation of knowledge kind
of work group where Yeah, if

396
00:28:42,014 --> 00:28:46,206
that's what your job is, yeah,
ChatGPT can do it. It's

397
00:28:46,206 --> 00:28:49,000
basically a database searching
tool.

398
00:28:49,000 --> 00:28:53,369
Audrow Nash: It is, and it just
formats the information for you.

399
00:28:53,369 --> 00:28:57,398
I, I was reflecting on this
recently, and the way, the kind

400
00:28:57,398 --> 00:29:01,426
of conclusion, or The, I don't
know, the, the, what, what I

401
00:29:01,426 --> 00:29:04,976
arrived at was that I think
it's, it's, there's like

402
00:29:04,976 --> 00:29:09,209
knowledge, like how do you do
stuff? And there's what do I do?

403
00:29:09,209 --> 00:29:13,100
And it's not very good at the
what, like if I'm solving a

404
00:29:13,100 --> 00:29:17,402
thorny programming problem, it's
not much help if I describe it

405
00:29:17,402 --> 00:29:21,498
to it and ask, what can I do?
It's super generic advice. But

406
00:29:21,498 --> 00:29:25,663
if I, like, I'm having to use a
new, large framework at work,

407
00:29:25,663 --> 00:29:29,691
and I don't know how to do
things in it, but I can be like,

408
00:29:29,691 --> 00:29:33,651
I know exactly what I need to
do. What should I do? And it

409
00:29:33,651 --> 00:29:37,610
tells me very well. So I'm
getting up and running probably

410
00:29:37,610 --> 00:29:41,639
like 10 times faster than I
would before this. So it's like

411
00:29:41,639 --> 00:29:46,008
knowledge is cheap, but like the
intuition on how to proceed and

412
00:29:46,008 --> 00:29:49,627
understanding all the
complexities or nuances of what

413
00:29:49,627 --> 00:29:53,723
you're doing, and like common
sense checking it. Seems to be

414
00:29:53,723 --> 00:29:57,000
like, there's a ways to go
there, in my opinion.

415
00:29:57,000 --> 00:30:01,313
Melonee Wise: Well, and, and it,
it also, I think one of the

416
00:30:01,313 --> 00:30:05,986
other things is you have to make
sure to check it, right? Because

417
00:30:05,986 --> 00:30:10,371
it does storytell, and make up,
information. Actually, one of

418
00:30:10,371 --> 00:30:14,828
the funnier things that, my co
founders and I did was we asked

419
00:30:14,828 --> 00:30:19,141
it who founded Fetch Robotics,
ChatGPT, and it didn't get it

420
00:30:19,141 --> 00:30:23,311
right at all. And we were
wondering about this because two

421
00:30:23,311 --> 00:30:27,624
of my co founders are very.
Under the radar, they don't have

422
00:30:27,624 --> 00:30:31,650
much social media. They don't
have much about themselves

423
00:30:31,650 --> 00:30:36,178
online and it didn't even know
who they were. So, so it's, it's

424
00:30:36,178 --> 00:30:40,851
one of these things where it has
only the information that it has

425
00:30:40,851 --> 00:30:45,452
access to, right? And so. It is
still when, when I say knowledge

426
00:30:45,452 --> 00:30:49,549
consolidation, that's what it's
doing. It's consolidating

427
00:30:49,549 --> 00:30:53,000
knowledge that's available and
it has access to.

428
00:30:53,000 --> 00:30:57,630
Audrow Nash: Yeah, I think
that's a big point. I really

429
00:30:57,630 --> 00:31:02,681
like that, that way of phrasing
it. Knowledge consolidation.

430
00:31:02,681 --> 00:31:07,985
Let's see. Going back to Digit,
do you, Are there, are you guys

431
00:31:07,985 --> 00:31:12,531
attempting to put an LLM or
anything on Digit for like

432
00:31:12,531 --> 00:31:17,834
having it do, I don't know, or I
guess what's your, what do you

433
00:31:17,834 --> 00:31:22,885
think about LLMs and their use
in robotics, and are you guys

434
00:31:22,885 --> 00:31:26,000
interested in trying this at
Agility?

435
00:31:26,000 --> 00:31:31,777
Melonee Wise: Yeah, so Yes, we
have an innovation group within

436
00:31:31,777 --> 00:31:37,367
Agility, and we have done some
pretty interesting demos with

437
00:31:37,367 --> 00:31:43,051
LLMs, largely showing how, if
you assume that Digit has a set

438
00:31:43,051 --> 00:31:48,268
of skills, Like, walk, pick up,
place, and you have that

439
00:31:48,268 --> 00:31:53,672
interface of skills, then you
can start asking Digit from,

440
00:31:53,672 --> 00:31:59,356
through a large language model
to do arbitrary tasks, as long

441
00:31:59,356 --> 00:32:05,319
as they can be broken down into
those composite skills. So, very

442
00:32:05,319 --> 00:32:10,630
recently we did a demo day in
San Francisco, with a large

443
00:32:10,630 --> 00:32:16,686
group of Interested Parties, and
one of the things that we showed

444
00:32:16,686 --> 00:32:22,463
off was a large language model
demo and someone was like, take

445
00:32:22,463 --> 00:32:28,146
the box that is the color of
Darth Vader's lightsaber and put

446
00:32:28,146 --> 00:32:34,016
it on the, podium. Labeled, the
same number as the movie number

447
00:32:34,016 --> 00:32:39,793
that Darth Vader appeared in for
the first time and it did it.

448
00:32:39,793 --> 00:32:45,756
And like, like, there's so much
to impact there in so many ways,

449
00:32:45,756 --> 00:32:51,253
like, and, but all of that,
basically, because it basically

450
00:32:51,253 --> 00:32:56,378
Digit was in a, like an
environment with some boxes and

451
00:32:56,378 --> 00:33:02,247
some podiums that were numbered
and labeled and they had colors

452
00:33:02,247 --> 00:33:08,117
and other iconography on them.
And so, the large language model

453
00:33:08,117 --> 00:33:14,080
basically unpacked that all into
basic, like an action tree, you

454
00:33:14,080 --> 00:33:20,043
know, a behavior tree, and then
using the skills that Digit had,

455
00:33:20,043 --> 00:33:22,000
it went and executed.

456
00:33:22,000 --> 00:33:25,656
Audrow Nash: Mm hmm. That's
really, it's very interesting

457
00:33:25,656 --> 00:33:29,633
and it's so funny that like now
you can be like, you can give,

458
00:33:29,633 --> 00:33:33,738
you're trying to give your robot
instructions and you program it

459
00:33:33,738 --> 00:33:37,907
with like a riddle and it has to
solve it and this kind of thing.

460
00:33:37,907 --> 00:33:41,819
Like you can be like esoteric
and whatever. It's so funny. Do

461
00:33:41,819 --> 00:33:45,796
you what do you think about its
impact like LLMs? In robotics,

462
00:33:45,796 --> 00:33:49,901
do you think, do you think it's
going to be like very useful for

463
00:33:49,901 --> 00:33:53,685
high level control, or is it
kind of a flash in the pan and

464
00:33:53,685 --> 00:33:57,405
we can do some neat demos with
it? Or is it just like user

465
00:33:57,405 --> 00:34:01,190
interface changes where it's,
you can talk to it and it's a

466
00:34:01,190 --> 00:34:05,102
bit easier to then do action
selection from there? How do you

467
00:34:05,102 --> 00:34:06,000
think it sits?

468
00:34:06,000 --> 00:34:10,393
Melonee Wise: I think we'll
initially see it, you know,

469
00:34:10,393 --> 00:34:15,265
because one of the things that
we are building at Agility is,

470
00:34:15,265 --> 00:34:20,297
is kind of this skill framework
for Digit. So as we start doing

471
00:34:20,297 --> 00:34:24,450
more and more with our
customers, probably the first

472
00:34:24,450 --> 00:34:29,323
place you'll see us potentially
using these tools is for, our

473
00:34:29,323 --> 00:34:34,275
customers to describe in natural
language what they want. The,

474
00:34:34,275 --> 00:34:39,307
the robot to do from a, from a
workflow perspective. It's like,

475
00:34:39,307 --> 00:34:43,540
I want you to move this tote
from the put wall to the

476
00:34:43,540 --> 00:34:48,572
conveyor. If you have a, if the
tote has an error, I would like

477
00:34:48,572 --> 00:34:53,684
you. To put it in the, hospital,
quote unquote, hospital area of

478
00:34:53,684 --> 00:34:56,000
the warehouse, but, you know,

479
00:34:56,000 --> 00:34:59,313
Audrow Nash: So it's super fast
templating of actions

480
00:34:59,313 --> 00:35:02,000
effectively, like building a
behavior tree.

481
00:35:02,000 --> 00:35:06,500
Melonee Wise: You're, you're
taking natural language of like

482
00:35:06,500 --> 00:35:10,925
someone who talks in business
logic. And making it easy for

483
00:35:10,925 --> 00:35:14,975
them to describe that without
them having to know like

484
00:35:14,975 --> 00:35:19,775
specifically, okay, digit has to
walk over here and identify the

485
00:35:19,775 --> 00:35:24,575
tote. And then digit has to grab
the tote. Like all of that will

486
00:35:24,575 --> 00:35:29,000
be basically derived from the
natural language description.

487
00:35:29,000 --> 00:35:33,261
Audrow Nash: That's super cool.
Yeah, I do think that kind of

488
00:35:33,261 --> 00:35:36,614
thing I could see speeding up
robotics adoption.

489
00:35:36,614 --> 00:35:40,735
Significantly, because they,
because you don't have to have

490
00:35:40,735 --> 00:35:45,065
someone who's an expert in the
field per se, translate it into

491
00:35:45,065 --> 00:35:49,326
like an expert in manufacturing
or logistics, go and take all

492
00:35:49,326 --> 00:35:53,726
the words, relate them to what
it means, then program it in the

493
00:35:53,726 --> 00:35:58,197
robot. So you need someone who's
good at both or can communicate

494
00:35:58,197 --> 00:36:02,667
with teams that do both. And so
now it's like, ah, you just tell

495
00:36:02,667 --> 00:36:06,718
the robot and it does it and it
translates it. That's very

496
00:36:06,718 --> 00:36:10,839
interesting. Okay, so that's
kind of the short term. Do you

497
00:36:10,839 --> 00:36:14,821
think like, eventually it's
going to be a bunch of robots

498
00:36:14,821 --> 00:36:19,291
that are all using LLMs for high
level control, and they'll just

499
00:36:19,291 --> 00:36:23,692
have like action sets that they
can do, and you'll get the high

500
00:36:23,692 --> 00:36:27,952
level commands? from the LLMs
and they'll just do it? Or what

501
00:36:27,952 --> 00:36:29,000
do you imagine?

502
00:36:29,000 --> 00:36:33,242
Melonee Wise: I'm, I'm guessing
that what you'll have is like a

503
00:36:33,242 --> 00:36:37,214
natural language description
that you can say, and then you

504
00:36:37,214 --> 00:36:41,186
basically evolve it until it's
what you want, especially in

505
00:36:41,186 --> 00:36:45,562
manufacturing and logistics. And
then you say freeze and then run

506
00:36:45,562 --> 00:36:49,602
forever. because People want
repeatable, deterministic labor

507
00:36:49,602 --> 00:36:53,642
from their robots. They, they
don't want every time you give

508
00:36:53,642 --> 00:36:57,547
it an instruction, it runs the
LLM every time and is like,

509
00:36:57,547 --> 00:37:01,855
well, today I decided I'm gonna
take a tour around the warehouse

510
00:37:01,855 --> 00:37:03,000
before I do this.

511
00:37:03,000 --> 00:37:07,260
Audrow Nash: I'll move this box
to over here every single time

512
00:37:07,260 --> 00:37:11,176
it has to decide it. This kind
of thing, that'd be pretty

513
00:37:11,176 --> 00:37:12,000
silly. Okay.

514
00:37:12,000 --> 00:37:16,270
Melonee Wise: But you can
consider, you could believe that

515
00:37:16,270 --> 00:37:20,982
that would be a normal thing in
like a home robot, right? And so

516
00:37:20,982 --> 00:37:25,399
it's really a domain specific
behavior and way of looking at

517
00:37:25,399 --> 00:37:29,963
the problem. It's just in our
domain, we care about stability,

518
00:37:29,963 --> 00:37:34,528
determinism, repeatability, you
know, throughput, reliability,

519
00:37:34,528 --> 00:37:36,000
all of those things.

520
00:37:36,000 --> 00:37:40,480
Audrow Nash: Yeah, for sure.
Let's see. So going back to

521
00:37:40,480 --> 00:37:45,600
Digit, um, one thing that I saw
that was very interesting on the

522
00:37:45,600 --> 00:37:49,840
website is you guys have
recently opened like a robot

523
00:37:49,840 --> 00:37:54,640
factory, like a big humanoid
robot factory. Will you tell me

524
00:37:54,640 --> 00:37:56,000
a bit about that?

525
00:37:56,000 --> 00:38:01,490
Melonee Wise: Yeah. So we've, we
haven't opened it yet. We've

526
00:38:01,490 --> 00:38:07,070
kicked off construction of it.
You know, it's, it's opening in

527
00:38:07,070 --> 00:38:12,110
the spring. and that will Be,
you know, it's designed to

528
00:38:12,110 --> 00:38:17,690
produce long term, up to 10, 000
robots in anticipation of, of

529
00:38:17,690 --> 00:38:23,000
the, the customer contracts that
we currently have. Mm hmm.

530
00:38:23,000 --> 00:38:27,655
Audrow Nash: That's awesome. And
so what parts of the making the

531
00:38:27,655 --> 00:38:32,164
robot will be done there? Will
it be a lot of assembly or will

532
00:38:32,164 --> 00:38:36,818
it be Okay, where do you get the
parts from otherwise? Like the,

533
00:38:36,818 --> 00:38:41,400
the motors, the sensors, where,
where are they all coming from?

534
00:38:41,400 --> 00:38:43,000
I guess it's a diverse

535
00:38:43,000 --> 00:38:47,626
Melonee Wise: Yeah, it's a
diverse set of suppliers and

536
00:38:47,626 --> 00:38:52,000
fabricators, um, in largely
North America. Wow. Hell

537
00:38:52,000 --> 00:38:56,769
Audrow Nash: yeah. Is that, I
guess that's, that's probably by

538
00:38:56,769 --> 00:39:00,000
choice to have it mostly in
North America?

539
00:39:00,000 --> 00:39:05,696
Melonee Wise: You know, we're
making sure that we are choosing

540
00:39:05,696 --> 00:39:11,024
suppliers and vendors that are
not in, I guess, conflicted

541
00:39:11,024 --> 00:39:16,444
geopolitical regions. We're
trying our best to do that. and

542
00:39:16,444 --> 00:39:21,864
we're trying to make sure that
we have a good, diversity of

543
00:39:21,864 --> 00:39:27,375
suppliers so that we're not
single sourced. but a lot of the

544
00:39:27,375 --> 00:39:32,612
North American focus is, is
because our initial market is

545
00:39:32,612 --> 00:39:35,000
North America. And so, um,

546
00:39:35,000 --> 00:39:39,785
Audrow Nash: Very interesting.
Do you, can you tell, I mean, we

547
00:39:39,785 --> 00:39:44,266
talked about it in, I think our
last interview, about just.

548
00:39:44,266 --> 00:39:48,823
There's actually a good amount
of manufacturing here, one of

549
00:39:48,823 --> 00:39:53,684
the, here in the US or in North
America. one of the things that,

550
00:39:53,684 --> 00:39:58,316
another space, so I was talking
to Rgs and he was saying that

551
00:39:58,316 --> 00:40:02,570
the US is really good for, and
if I, if I am mistaken or

552
00:40:02,570 --> 00:40:06,747
anything, I apologize in
advance. But what I understood

553
00:40:06,747 --> 00:40:11,684
was the US is very good at, very
specialized. Like high precision

554
00:40:11,684 --> 00:40:16,241
manufacturing, but when you get
more to volume, it's tricky.

555
00:40:16,241 --> 00:40:20,570
cause it doesn't seem that
there's that much of that, but

556
00:40:20,570 --> 00:40:23,000
what's your perspective on this?

557
00:40:23,000 --> 00:40:27,363
Melonee Wise: Yeah. I mean, I
think that that is relatively

558
00:40:27,363 --> 00:40:31,949
true. except for in automotive,
I think if you, if you look at

559
00:40:31,949 --> 00:40:36,682
it, it really comes down to the
cost of, of bringing the product

560
00:40:36,682 --> 00:40:41,341
back into the United States and
in what volume and in what kind

561
00:40:41,341 --> 00:40:45,261
of import tariffs and things
like that that you might

562
00:40:45,261 --> 00:40:49,477
encounter. It for very large
high value goods for smaller

563
00:40:49,477 --> 00:40:53,618
electronics, high value
electronics, you know, things in

564
00:40:53,618 --> 00:40:58,130
the one 5, 000 range. One of the
things that we basically run

565
00:40:58,130 --> 00:41:02,567
into in the United States is
just the labor pool. Like we're

566
00:41:02,567 --> 00:41:06,930
already talking about logistics
and manufacturing, having a

567
00:41:06,930 --> 00:41:11,368
million open jobs right now in
the United States, right? Can

568
00:41:11,368 --> 00:41:15,953
you imagine producing the iPhone
in the United States? Like, I

569
00:41:15,953 --> 00:41:20,539
saw, like, an estimate that you
would have to basically build,

570
00:41:20,539 --> 00:41:25,272
like, a whole town with millions
of people to support the iPhone

571
00:41:25,272 --> 00:41:29,413
production, in the United
States. And so Like one of the

572
00:41:29,413 --> 00:41:33,259
big reasons that actually
precludes us from building

573
00:41:33,259 --> 00:41:37,401
certain things, especially at
high volumes in the United

574
00:41:37,401 --> 00:41:41,542
States, is we just don't have
the infrastructure and the

575
00:41:41,542 --> 00:41:46,054
people to do it. which is, is
different in other parts of the

576
00:41:46,054 --> 00:41:50,713
world. They actually have, like,
if you look at Shenzhen in, in

577
00:41:50,713 --> 00:41:55,150
China, Like, they produce so
much there, and you could like,

578
00:41:55,150 --> 00:41:59,514
walk down the street and go to
like 14 injection molders in

579
00:41:59,514 --> 00:42:03,655
like, the, like the span of like
5 city blocks, which is

580
00:42:03,655 --> 00:42:07,501
something you would never find
in the United States,

581
00:42:07,501 --> 00:42:12,308
potentially. and so there's just
this contrast of, of, suppliers.

582
00:42:12,308 --> 00:42:16,450
The density of them and the
sheer amount of labor that's

583
00:42:16,450 --> 00:42:20,000
available to support some of
that manufacturing.

584
00:42:20,000 --> 00:42:24,005
Audrow Nash: Yeah, where, so
where are most of your

585
00:42:24,005 --> 00:42:28,874
Manufacturers in North America,
like where in the U S are they

586
00:42:28,874 --> 00:42:33,979
in Mexico? Maybe some in Canada,
like where, where are you seeing

587
00:42:33,979 --> 00:42:35,000
most of them?

588
00:42:35,000 --> 00:42:39,822
Melonee Wise: I can't speak for
all of that because that's not

589
00:42:39,822 --> 00:42:44,722
my like, job at, at agility, but
we do use, several large North

590
00:42:44,722 --> 00:42:49,389
American machine shops and other
vendors for making our, our

591
00:42:49,389 --> 00:42:53,744
machine parts and things like
that. And they're, They're

592
00:42:53,744 --> 00:42:56,000
located in the United States.

593
00:42:56,000 --> 00:42:59,982
Audrow Nash: Gotcha. do you
like, one thing that has been

594
00:42:59,982 --> 00:43:04,383
interesting is the US I think is
investing pretty significantly

595
00:43:04,383 --> 00:43:08,156
in like, reshoring, a lot of
manufacturing. So I think

596
00:43:08,156 --> 00:43:12,347
there's like. I don't know,
trillions of dollars or billions

597
00:43:12,347 --> 00:43:16,329
of dollars, like some huge
amount of money in flight. And

598
00:43:16,329 --> 00:43:20,731
that infrastructure is going to
take, I don't know, 10 years or

599
00:43:20,731 --> 00:43:25,062
something to spin up. But how do
you imagine this changing in,

600
00:43:25,062 --> 00:43:29,533
say, 10 years, 15 years, if you
have opinions on kind of what's,

601
00:43:29,533 --> 00:43:31,000
what's been going on?

602
00:43:31,000 --> 00:43:36,524
Melonee Wise: I think that in
order to achieve that, we have

603
00:43:36,524 --> 00:43:41,680
to reestablish our trade class
as a country. and I think

604
00:43:41,680 --> 00:43:47,296
that's, that's something that a
lot of people are working on.

605
00:43:47,296 --> 00:43:53,188
But, I think that there are not
strong incentives right now. for

606
00:43:53,188 --> 00:43:58,805
people to go become tradesmen,
to become skilled electricians

607
00:43:58,805 --> 00:44:04,605
or skilled mold makers, welders,
welders. Yeah. Yeah. And so it

608
00:44:04,605 --> 00:44:10,497
kind of creates this vacuum. It,
even if we wanted to reshore it

609
00:44:10,497 --> 00:44:16,113
all, could we reshore it all?
Because we have tended towards.

610
00:44:16,113 --> 00:44:21,637
over the last 20 to 50 years,
you know, towards a higher and

611
00:44:21,637 --> 00:44:26,517
higher, higher education
population, that is trending

612
00:44:26,517 --> 00:44:32,041
away from trade based jobs.
which leaves a gap for trying to

613
00:44:32,041 --> 00:44:36,000
reshore manufacturing in the
United States.

614
00:44:36,000 --> 00:44:39,939
Audrow Nash: Yeah. It's such an
interesting thing. So like I

615
00:44:39,939 --> 00:44:44,076
talked with. on the Sense Think
Act Podcast, I talked with like

616
00:44:44,076 --> 00:44:47,424
AMT, Association for
Manufacturing Technology and a

617
00:44:47,424 --> 00:44:50,904
few other manufacturing
organizations and some of the

618
00:44:50,904 --> 00:44:54,778
things that stuck out to me,
it's like, Actually, community

619
00:44:54,778 --> 00:44:58,520
college and, like community
college is a wonderful way to

620
00:44:58,520 --> 00:45:02,525
get a trades education and you
can get like, so you train for

621
00:45:02,525 --> 00:45:06,596
six months or something fairly
short and you immediately get a

622
00:45:06,596 --> 00:45:10,535
high paying job for that. Like
not CEO high paying, but like

623
00:45:10,535 --> 00:45:13,884
enough to raise a family,
probably a couple hundred

624
00:45:13,884 --> 00:45:15,000
thousand or more.

625
00:45:15,000 --> 00:45:19,271
Melonee Wise: Yeah, because,
because there's so few trades

626
00:45:19,271 --> 00:45:23,542
people, right? And they're
trying more and more to get the

627
00:45:23,542 --> 00:45:28,107
word out. I know AMT is, I mean,
they're really trying. A3 is,

628
00:45:28,107 --> 00:45:32,230
they're trying to pull more
people into being, you know,

629
00:45:32,230 --> 00:45:36,796
robotic operators, for example.
That's a very high paying job,

630
00:45:36,796 --> 00:45:41,140
but just getting young people
interested in it, them seeing

631
00:45:41,140 --> 00:45:45,190
the value in it, and seeing it
as a career path that is

632
00:45:45,190 --> 00:45:46,000
meaningful.

633
00:45:46,000 --> 00:45:50,176
Audrow Nash: Do you think, so a
thing that I have been aware of

634
00:45:50,176 --> 00:45:53,954
is there's, and I don't, this
doesn't, definitely doesn't

635
00:45:53,954 --> 00:45:57,666
speak for everyone, but I've
been seeing a lot of people

636
00:45:57,666 --> 00:46:01,908
thinking universities are not a
terribly good deal for this kind

637
00:46:01,908 --> 00:46:05,818
of thing. So it's like some
people are saying things like I

638
00:46:05,818 --> 00:46:09,729
won't have my kids go to college
and whatever because maybe

639
00:46:09,729 --> 00:46:13,839
trades and this kind of thing
could be a good Alternative, and

640
00:46:13,839 --> 00:46:17,816
a lot of, a lot of people just
end up getting a lot of debt.

641
00:46:17,816 --> 00:46:21,991
But, do you think that kind of
society's going to start having,

642
00:46:21,991 --> 00:46:25,769
like, will we move back into a
better balance with trades

643
00:46:25,769 --> 00:46:29,746
versus everyone goes to a four
year college with not as many

644
00:46:29,746 --> 00:46:32,000
exceptions and this kind of
thing?

645
00:46:32,000 --> 00:46:36,589
Melonee Wise: I don't know. I
doubt it, if I were to guess

646
00:46:36,589 --> 00:46:41,178
because The United States, we
have a very manifest destiny

647
00:46:41,178 --> 00:46:45,925
approach to life, right? What
does it mean? Of, well, if you

648
00:46:45,925 --> 00:46:50,910
have a desire, you should go get
it, right? And so we raise Our

649
00:46:50,910 --> 00:46:55,578
children and our cultural bias
is towards achieving what we

650
00:46:55,578 --> 00:47:00,088
want, living our dreams.
Passion, passion, passion. Yeah.

651
00:47:00,088 --> 00:47:04,598
And until we start reframing,
um, some of these things as

652
00:47:04,598 --> 00:47:09,108
something that you can be
passionate I mean, engineering,

653
00:47:09,108 --> 00:47:13,776
for example, has had a really
bad, bad rap for a long time,

654
00:47:13,776 --> 00:47:18,602
right? Right? You go and you
look at TV and every engineer is

655
00:47:18,602 --> 00:47:23,429
homely and weird and doesn't
have a girlfriend and, you know,

656
00:47:23,429 --> 00:47:27,938
sits at home and He's on their
computer all the time, and

657
00:47:27,938 --> 00:47:32,765
everyone's been told that they
have to be the smartest person

658
00:47:32,765 --> 00:47:37,591
in the world. And what have we
had for, you know, the last 20

659
00:47:37,591 --> 00:47:42,180
years? A dearth of engineers.
And so, you know, we've been

660
00:47:42,180 --> 00:47:46,690
trying to change it as a
community, trying to help people

661
00:47:46,690 --> 00:47:51,200
understand engineering isn't
just about being smart, it's

662
00:47:51,200 --> 00:47:55,947
about being creative, it's about
problem solving. You're not

663
00:47:55,947 --> 00:48:00,774
going to be Sheldon. Maybe you
want to be, but, you know, you

664
00:48:00,774 --> 00:48:05,758
don't have to be Sheldon if you
don't want to be. And, and that

665
00:48:05,758 --> 00:48:09,952
is something that like,
societally, we have to change

666
00:48:09,952 --> 00:48:15,015
about any of these roles that we
want people to go into. Whether

667
00:48:15,015 --> 00:48:19,921
it's engineering, or a trades
job, or, or something like that,

668
00:48:19,921 --> 00:48:24,431
but if, if young people don't
believe that they're either

669
00:48:24,431 --> 00:48:29,178
qualified for the role, in the
case of engineering, or it's,

670
00:48:29,178 --> 00:48:34,004
something to aspire to, In the
case of some trades, you know,

671
00:48:34,004 --> 00:48:38,989
how do we convince people that
they should aspire to it? And so

672
00:48:38,989 --> 00:48:43,815
a lot of it is reframing what,
what people should be aspiring

673
00:48:43,815 --> 00:48:48,484
to and raising our children to
believe that it's okay to do

674
00:48:48,484 --> 00:48:53,389
these things. But if you spend
all your time telling your kids

675
00:48:53,389 --> 00:48:56,000
the best thing, change the
world.

676
00:48:56,000 --> 00:48:58,000
Audrow Nash: Yeah, these kinds
of things.

677
00:48:58,000 --> 00:49:01,471
Melonee Wise: Yeah, or go to
university, then they're going

678
00:49:01,471 --> 00:49:04,000
to feel like they're failing if
they don't.

679
00:49:04,000 --> 00:49:07,952
Audrow Nash: Yeah, it's such an
interesting problem in a sense.

680
00:49:07,952 --> 00:49:11,779
What do you think would be, do
you have any ideas what a good

681
00:49:11,779 --> 00:49:15,480
solution is? It's reframe it,
but how would you reframe it?

682
00:49:15,480 --> 00:49:18,930
and I know clearly we've been
struggling with this as a

683
00:49:18,930 --> 00:49:21,000
community, but what do you
think?

684
00:49:21,000 --> 00:49:26,964
Melonee Wise: I think that's a
hard one because I think it, it,

685
00:49:26,964 --> 00:49:32,456
It really comes down to what
people value, right? Like, do

686
00:49:32,456 --> 00:49:38,325
they value stability? Do they
value, you know, wealth? Do they

687
00:49:38,325 --> 00:49:44,101
value career growth? and. Those
are stories you have to tell,

688
00:49:44,101 --> 00:49:50,254
but I also think that one of the
reasons people have historically

689
00:49:50,254 --> 00:49:56,219
not wanted to adopt trade jobs
because they have limited career

690
00:49:56,219 --> 00:50:02,089
growth at some point and limited
wage growth, for example. And

691
00:50:02,089 --> 00:50:07,864
this comes back to some of the
question before, which is, you

692
00:50:07,864 --> 00:50:12,882
know, universal basic income.
What does it mean to be

693
00:50:12,882 --> 00:50:18,941
successful? You know, how do you
create, how do you go and reach

694
00:50:18,941 --> 00:50:25,000
beyond and, and get more? and we
don't have great stories there.

695
00:50:25,000 --> 00:50:28,818
Audrow Nash: Yeah, so coming up
with some sort of good story so

696
00:50:28,818 --> 00:50:32,515
that people can, you can just
chill and you can do a job, and

697
00:50:32,515 --> 00:50:35,970
it can be a meaningful one.
That's say a trade, or if you

698
00:50:35,970 --> 00:50:39,424
want to go really big and you
like. Like it's going to be

699
00:50:39,424 --> 00:50:43,182
like, I mean, you with Fetch,
for example, I imagine that was,

700
00:50:43,182 --> 00:50:47,000
that was a hard path to choose
and it was a great, like you did

701
00:50:47,000 --> 00:50:50,576
very well and everything is
great because of it. But it was

702
00:50:50,576 --> 00:50:53,000
also probably very difficult, I
imagine.

703
00:50:53,000 --> 00:50:57,538
Melonee Wise: And it was high
risk, right? Yeah. High risk.

704
00:50:57,538 --> 00:51:02,077
But I, I think the other thing
that, that has changed. Very

705
00:51:02,077 --> 00:51:06,769
much in the last 50 years or 60
years is the disappearance of

706
00:51:06,769 --> 00:51:11,308
the pension has also had a very
big impact I think on trade

707
00:51:11,308 --> 00:51:12,000
adoption.

708
00:51:12,000 --> 00:51:15,684
Audrow Nash: I don't, so the
pension, pensions are like you,

709
00:51:15,684 --> 00:51:19,000
the retirement accounts that
your company contributes.

710
00:51:19,000 --> 00:51:24,205
Melonee Wise: No, no, so
pensions were The last salary

711
00:51:24,205 --> 00:51:30,084
that you made in the last year
of work, you got 80 percent of

712
00:51:30,084 --> 00:51:35,000
it for the rest of your life. Oh
wow. Oh wow. Yeah.

713
00:51:35,000 --> 00:51:39,337
Audrow Nash: So why did they
disappear? They're expensive. I

714
00:51:39,337 --> 00:51:41,000
guess that makes sense.

715
00:51:41,000 --> 00:51:44,219
Melonee Wise: Yeah. Go look up
like what happened with GM. Like

716
00:51:44,219 --> 00:51:47,387
they had a very large pension
program and it almost bankrupted

717
00:51:47,387 --> 00:51:48,000
the company.

718
00:51:48,000 --> 00:51:51,819
Audrow Nash: So it's kind of
like, I don't know, you hear

719
00:51:51,819 --> 00:51:54,901
about like. The fall of
different countries or

720
00:51:54,901 --> 00:51:58,787
something. And they eventually
are like way over leveraged

721
00:51:58,787 --> 00:52:02,672
financially or something to
their citizens because of like

722
00:52:02,672 --> 00:52:06,558
generous retirement plans, like
this kind of happened with

723
00:52:06,558 --> 00:52:10,645
pensions. And so we stepped back
with it, but it removes some

724
00:52:10,645 --> 00:52:14,196
incentive to go do trade jobs
that had good pensions.

725
00:52:14,196 --> 00:52:15,000
Interesting.

726
00:52:15,000 --> 00:52:19,571
Melonee Wise: I mean, there's a
couple of jobs still left in the

727
00:52:19,571 --> 00:52:23,929
world that still get pensions.
Teachers tend to get pensions.

728
00:52:23,929 --> 00:52:27,786
Government jobs get pensions,
but they used to be that

729
00:52:27,786 --> 00:52:30,000
tradespeople got pensions. huh.

730
00:52:30,000 --> 00:52:35,172
Audrow Nash: Would the, it's the
pension, the pension is paid by

731
00:52:35,172 --> 00:52:40,102
the employer, correct? So yeah,
that does sound expensive for

732
00:52:40,102 --> 00:52:44,790
this kind of thing. And also
it's interesting because like

733
00:52:44,790 --> 00:52:49,720
teachers, like, I don't know, I
have family who, are teachers

734
00:52:49,720 --> 00:52:54,407
or, have been teachers and
they're not paid very fairly or

735
00:52:54,407 --> 00:52:59,337
not very much. So it's like you
give them a pension, but like

736
00:52:59,337 --> 00:53:04,106
you're only paying them, I don't
know, a third of what they

737
00:53:04,106 --> 00:53:08,712
probably should be paid at
least. Yeah, you get the thing

738
00:53:08,712 --> 00:53:13,885
and you get 80 percent of that,
but it's still super, super low.

739
00:53:13,885 --> 00:53:18,653
what an interesting thing. I
wonder, um, so. Back to robots

740
00:53:18,653 --> 00:53:23,664
with Digit. So one thing that
struck me from looking at Digit,

741
00:53:23,664 --> 00:53:28,109
in some of the videos, it
doesn't really have hands. It

742
00:53:28,109 --> 00:53:32,958
has like flippers for where I,
where we have all the fingers

743
00:53:32,958 --> 00:53:37,969
that it can use to pull in and I
guess grab things and then it

744
00:53:37,969 --> 00:53:42,656
probably pinches. I don't, I
don't remember seeing thumbs.

745
00:53:42,656 --> 00:53:45,000
No. tell me a bit about that.

746
00:53:45,000 --> 00:53:53,194
Melonee Wise: So, that's one
minute. Instantiation of Digit's

747
00:53:53,194 --> 00:53:54,000
hands.

748
00:53:54,000 --> 00:53:57,000
Audrow Nash: Yeah, very modular,
I suppose.

749
00:53:57,000 --> 00:54:02,154
Melonee Wise: Yeah. You know,
when you look at where we're

750
00:54:02,154 --> 00:54:07,664
going with the end effectors of
Digit, Digit, um, as we evolve

751
00:54:07,664 --> 00:54:12,997
the design, we'll have kind of,
interchange point, like all.

752
00:54:12,997 --> 00:54:18,596
Robots where we will be able to
change out the end effectors of

753
00:54:18,596 --> 00:54:24,372
the robot based on the task It's
it's what industrial robots have

754
00:54:24,372 --> 00:54:29,438
been doing for a long time We've
tried to focus on having

755
00:54:29,438 --> 00:54:34,237
relatively simplistic grippers
to start or I guess end

756
00:54:34,237 --> 00:54:39,569
effectors I should say that
solve A large swath of problems,

757
00:54:39,569 --> 00:54:44,990
with the simplest design. And
that's one, approach that we've

758
00:54:44,990 --> 00:54:49,434
taken for doing tote
manipulation. And it's fairly

759
00:54:49,434 --> 00:54:54,944
robust for some of the, the, the
tasks that we've been focused

760
00:54:54,944 --> 00:55:00,009
on, but we are right now going
to add a different type of

761
00:55:00,009 --> 00:55:05,519
gripper to digits repertoire,
for handling totes. so it's, you

762
00:55:05,519 --> 00:55:11,118
know, We're trying to create MVP
products, right? And so we, we

763
00:55:11,118 --> 00:55:16,539
are not trying to solve. And
swallow all of the complexity at

764
00:55:16,539 --> 00:55:21,782
once. Digit has the ability to
have other end effectors. We

765
00:55:21,782 --> 00:55:27,115
will make other end effectors
for Digit. But our priority is

766
00:55:27,115 --> 00:55:32,358
not to make high dexterity
hands. Because honestly, like, I

767
00:55:32,358 --> 00:55:36,890
haven't seen a problem yet.
That, that we need high

768
00:55:36,890 --> 00:55:42,311
dexterity hands for digit for
the set of use cases that we're

769
00:55:42,311 --> 00:55:44,000
tackling right now.

770
00:55:44,000 --> 00:55:48,717
Audrow Nash: Gotcha. So, it's
just unnecessary complexity and

771
00:55:48,717 --> 00:55:53,356
you can get most of the way
there with just very simple like

772
00:55:53,356 --> 00:55:58,228
pinch in grippers kind of thing.
Yeah. What do you think, so To

773
00:55:58,228 --> 00:56:02,868
me, what it seems is, and this
kind of goes back even before

774
00:56:02,868 --> 00:56:07,817
you were involved with agility,
it seems like it's like the min,

775
00:56:07,817 --> 00:56:12,147
as you said, an MVP, it's a
minimum product that, can do

776
00:56:12,147 --> 00:56:16,554
something useful and can find
market fit and this kind of

777
00:56:16,554 --> 00:56:21,117
thing. what do you think of?
There's been several companies

778
00:56:21,117 --> 00:56:26,066
entering the humanoid space and
they seem to be making like full

779
00:56:26,066 --> 00:56:30,705
fledged humanoids with not, to
my knowledge, they don't have

780
00:56:30,705 --> 00:56:35,499
too much of an application in
mind for them and maybe they do,

781
00:56:35,499 --> 00:56:39,907
but what are your thoughts on
some of the other humanoid?

782
00:56:39,907 --> 00:56:43,000
Robot companies or humanoid
initiatives.

783
00:56:43,000 --> 00:56:45,493
Melonee Wise: Yeah, so if you
look at 'em, some of them are

784
00:56:45,493 --> 00:56:47,986
very impressive and have been
around for a long time. Like,

785
00:56:47,986 --> 00:56:49,000
look at Boston Dynamics.

786
00:56:49,000 --> 00:56:53,200
Audrow Nash: Mm-Hmm. . Oh yeah.
Yeah. I mean, super impressive.

787
00:56:53,200 --> 00:56:57,000
some of them are some back folks
and stuff, just bonkers.

788
00:56:57,000 --> 00:57:01,513
Melonee Wise: Yeah, and well,
and, and remember they, they

789
00:57:01,513 --> 00:57:06,027
started from a very different
place, right? They started a

790
00:57:06,027 --> 00:57:10,930
long time ago as the Petman and
they were testing hazmat suits,

791
00:57:10,930 --> 00:57:15,054
right? Yeah. And they're
hydraulic based. And so they

792
00:57:15,054 --> 00:57:19,412
have different challenges.
Superpower. Yeah. But I think

793
00:57:19,412 --> 00:57:23,692
that when you look at a lot of
the stuff that sprung up

794
00:57:23,692 --> 00:57:28,205
recently, a lot of it is very
startup heavy. they're still

795
00:57:28,205 --> 00:57:33,108
even just trying to figure out
why they're building it. many of

796
00:57:33,108 --> 00:57:37,933
them haven't declared a market
or, or shown my impression too,

797
00:57:37,933 --> 00:57:42,835
or shown working robots. I mean,
let's be honest, there's a lot

798
00:57:42,835 --> 00:57:47,427
of videos, not a lot of reality,
and, and I'm not trying to

799
00:57:47,427 --> 00:57:51,862
criticize, I'm just saying that
this is, you know, I went

800
00:57:51,862 --> 00:57:56,454
through this in the early days
of the AMR market. There was

801
00:57:56,454 --> 00:58:01,123
like five companies who were
actually building hardware, and

802
00:58:01,123 --> 00:58:05,870
everyone else was showing really
cool videos of hardware that

803
00:58:05,870 --> 00:58:10,695
they were going to build. Right.
and, and we're, we're in that

804
00:58:10,695 --> 00:58:15,052
stage right now with mobile
manipulation robots. and the

805
00:58:15,052 --> 00:58:19,799
thing that I find interesting
though, is, is some of the, the

806
00:58:19,799 --> 00:58:24,624
players that are getting into
the game is recently are kind of

807
00:58:24,624 --> 00:58:29,449
funny. They're just throwing a
lot of money at the problem and

808
00:58:29,449 --> 00:58:33,807
just hiring any engineer they
can. And they're like, you

809
00:58:33,807 --> 00:58:38,632
roboticists, you guys are taking
too long. It's like, what the

810
00:58:38,632 --> 00:58:43,534
hell? As a community, I've been
working on these problems for a

811
00:58:43,534 --> 00:58:47,814
very long time. And like,
magically, like thinking that

812
00:58:47,814 --> 00:58:52,795
you can throw a ton of engineers
at it and like get good results

813
00:58:52,795 --> 00:58:57,775
is I don't know. It's a mythical
man month, basically, and We'll

814
00:58:57,775 --> 00:59:02,289
see. But I, I think that there
are some really interesting

815
00:59:02,289 --> 00:59:07,113
competitors out there. I, you
know, and I'm excited about what

816
00:59:07,113 --> 00:59:11,860
they're building because, you
know, there's plenty of room. I

817
00:59:11,860 --> 00:59:16,607
mean, look at what happened in
the AMR market, you know. Many

818
00:59:16,607 --> 00:59:21,432
companies, yeah. Yeah, there's
Locus, Auto, Mirror, Fetch. You

819
00:59:21,432 --> 00:59:24,000
know, we all did very, very
well,

820
00:59:24,000 --> 00:59:28,317
Audrow Nash: So you're thinking
of it like AMRs, I suppose,

821
00:59:28,317 --> 00:59:32,488
seeing all these humanoid
companies. I was, I was feeling

822
00:59:32,488 --> 00:59:36,293
like it was a bit like
autonomous vehicles with, all

823
00:59:36,293 --> 00:59:40,756
the, like the, the feeling that
I'm getting with the investor

824
00:59:40,756 --> 00:59:45,293
interest and kind of the hype
cycles around it. It feels a lot

825
00:59:45,293 --> 00:59:49,244
like the early days of
autonomous cars in like 2014 or

826
00:59:49,244 --> 00:59:53,927
something when people were like
We're going to have self driving

827
00:59:53,927 --> 00:59:57,000
cars in four years or something
like that.

828
00:59:57,000 --> 01:00:00,646
Melonee Wise: They haven't put
enough money in for you to

829
01:00:00,646 --> 01:00:04,421
believe it's like autonomous
cars yet. It's more like AMRs.

830
01:00:04,421 --> 01:00:07,875
Gotcha. Yeah, like the total
amount of money in mobile

831
01:00:07,875 --> 01:00:11,330
manipulation right now in
humanoids, quote unquote, is

832
01:00:11,330 --> 01:00:15,168
closer to the AMR space still.
Like the total dollars put in

833
01:00:15,168 --> 01:00:16,000
from venture.

834
01:00:16,000 --> 01:00:19,650
Audrow Nash: How much would that
be, out of curiosity, if you

835
01:00:19,650 --> 01:00:23,000
have a good guess? Is it like
tens of billions or is it?

836
01:00:23,000 --> 01:00:28,702
Melonee Wise: It's under five to
ten billion. That's like. That's

837
01:00:28,702 --> 01:00:33,000
AMRs and then over 5 to 10
billion is automotive.

838
01:00:33,000 --> 01:00:35,442
Audrow Nash: Over 5 to 10
billion. How much, how much more

839
01:00:35,442 --> 01:00:37,000
over? Is it like 50 billion or
is it?

840
01:00:37,000 --> 01:00:42,598
Melonee Wise: I don't know. I
don't, I, I, I'm guessing it's

841
01:00:42,598 --> 01:00:48,102
in that, in that frame range
because like just look at the,

842
01:00:48,102 --> 01:00:53,700
what wasn't there a, like an
autonomous car company recently

843
01:00:53,700 --> 01:00:59,297
that got some massive, insane
amount of money. I don't know.

844
01:00:59,297 --> 01:01:05,268
It was like, what was it? Well,
was it, I thought it was 80, no,

845
01:01:05,268 --> 01:01:09,000
8 billion or something insane
like that?

846
01:01:09,000 --> 01:01:15,000
Audrow Nash: That's quite a lot.
I wonder which company that was.

847
01:01:15,000 --> 01:01:19,000
Melonee Wise: Yeah, I don't
know, let me see

848
01:01:19,000 --> 01:01:22,527
Audrow Nash: if I can Look if
you like. Yeah, we have time.

849
01:01:22,527 --> 01:01:25,875
Yeah. One of the wonderful
things about these long forum

850
01:01:25,875 --> 01:01:29,462
Things doesn't really matter.
and I'll cut it out if we wait

851
01:01:29,462 --> 01:01:30,000
too long.

852
01:01:30,000 --> 01:01:37,000
Melonee Wise: Yeah, I'm trying
to find it.

853
01:01:37,000 --> 01:01:41,910
Audrow Nash: so we couldn't find
the company that had the high

854
01:01:41,910 --> 01:01:46,741
valuation, but, so thinking more
about humanoids. What do you

855
01:01:46,741 --> 01:01:51,809
imagine is a timeline for them?
so you're going to, you guys are

856
01:01:51,809 --> 01:01:56,798
working on them in a logistics
and manufacturing space. what do

857
01:01:56,798 --> 01:02:00,995
you kind of imagine the
progression? Like when will I

858
01:02:00,995 --> 01:02:06,064
see one in a grocery store kind
of thing or like, like when will

859
01:02:06,064 --> 01:02:10,815
I see them in my day to day life
or will it be, will they be

860
01:02:10,815 --> 01:02:15,250
relegated mostly to like
manufacturing and logistics for

861
01:02:15,250 --> 01:02:19,843
a while? Okay, so I mean, but
there's a bunch of, it seems

862
01:02:19,843 --> 01:02:24,516
like there's a lot of excitement
where they're like, in two

863
01:02:24,516 --> 01:02:28,000
years, there's going to be one
in your home.

864
01:02:28,000 --> 01:02:34,164
Melonee Wise: Yeah. Okay,
Audrow, let's reflect. Yes. In,

865
01:02:34,164 --> 01:02:40,870
in 2004, 2004, they said that
everyone will have an autonomous

866
01:02:40,870 --> 01:02:47,250
car in 10 years. Oh, man. Oh,
man. So, now, in 2004, if you

867
01:02:47,250 --> 01:02:53,523
derated that and you said
autonomous cars would be part of

868
01:02:53,523 --> 01:03:00,012
your everyday life, do you feel
that's true today? Because I

869
01:03:00,012 --> 01:03:06,284
don't. Okay. So now, today,
we're starting with, with, you

870
01:03:06,284 --> 01:03:12,665
know, humanoid bipedal mobile
manipulation robots, right? I

871
01:03:12,665 --> 01:03:19,479
don't think that that'll be part
of your daily life for another

872
01:03:19,479 --> 01:03:25,535
20 or 30 years. I think we'll
spend the next 10 years in

873
01:03:25,535 --> 01:03:31,916
industrial, light industrial
environments. There's a lot of

874
01:03:31,916 --> 01:03:38,837
safety work that has to be done
to get them out of the warehouse

875
01:03:38,837 --> 01:03:41,000
and into your house.

876
01:03:41,000 --> 01:03:46,679
Audrow Nash: And just, I feel
like I have to ask, This doesn't

877
01:03:46,679 --> 01:03:52,450
get super accelerated because of
LLMs, or what are you thinking

878
01:03:52,450 --> 01:03:53,000
there?

879
01:03:53,000 --> 01:03:57,542
Melonee Wise: no, because it
doesn't make the hardware any

880
01:03:57,542 --> 01:04:02,084
cheaper. It doesn't make the
controls any easier. It makes

881
01:04:02,084 --> 01:04:07,096
the programming of them a little
bit easier, but it doesn't make

882
01:04:07,096 --> 01:04:11,482
some of the more fundamental
problems. Like do LLMs make

883
01:04:11,482 --> 01:04:15,946
autonomous cars go faster? So
why would they make bipedal

884
01:04:15,946 --> 01:04:19,000
navigating robots go any faster
either?

885
01:04:19,000 --> 01:04:24,009
Audrow Nash: Yeah, now, thought
experiment to just, what if, so

886
01:04:24,009 --> 01:04:29,177
if OpenAI reveals that they have
made super intelligent something

887
01:04:29,177 --> 01:04:33,630
or another and it's like
Einstein for every single field

888
01:04:33,630 --> 01:04:38,560
all at once, how does that work?
Like, does, we still have all

889
01:04:38,560 --> 01:04:43,330
these hard problems and the
timeline kind of stays unchanged

890
01:04:43,330 --> 01:04:45,000
or what do you think?

891
01:04:45,000 --> 01:04:49,307
Melonee Wise: I think the
timeline still changes. It

892
01:04:49,307 --> 01:04:54,276
remains unchanged. And maybe
that's naive of me, but like, I

893
01:04:54,276 --> 01:04:58,666
mean, technology has been
constantly progressing. You

894
01:04:58,666 --> 01:05:03,552
know, and although ChatGPT is
super interesting and showing

895
01:05:03,552 --> 01:05:08,190
very interesting progress, how
much has it fundamentally

896
01:05:08,190 --> 01:05:12,000
changed your day to day in the
last 12 months?

897
01:05:12,000 --> 01:05:15,000
Audrow Nash: It's a better
Google to me is what it's been.

898
01:05:15,000 --> 01:05:20,424
Melonee Wise: It's not, I mean,
yes, people are very excited

899
01:05:20,424 --> 01:05:26,210
about it. It's very powerful in
very specific ways, but it's not

900
01:05:26,210 --> 01:05:31,544
the, You know, singularity
moment that, that everyone, it's

901
01:05:31,544 --> 01:05:36,788
nowhere near that, right? It's,
it's like you said, it's a

902
01:05:36,788 --> 01:05:42,393
fancier Google right now. and I,
I think though that like, you

903
01:05:42,393 --> 01:05:47,998
know, some of the thoughts that
Bill Gates put, Bill Gates put

904
01:05:47,998 --> 01:05:53,242
out about like personalized
agents is kind of interesting.

905
01:05:53,242 --> 01:05:58,485
but there's a lot of other
things that we have to, to deal

906
01:05:58,485 --> 01:06:04,000
with. Or I'm like, what if, what
if you had your own ChatGPT?

907
01:06:04,000 --> 01:06:06,719
Audrow Nash: Hmm, like they had
a history and this kind of

908
01:06:06,719 --> 01:06:07,000
thing.

909
01:06:07,000 --> 01:06:10,945
Melonee Wise: That you know,
that was like it took all of

910
01:06:10,945 --> 01:06:14,751
your data, all of it. It has
everything about you, your

911
01:06:14,751 --> 01:06:18,558
medical records, everything,
everything you ever wrote,

912
01:06:18,558 --> 01:06:22,503
everything you ever did, all of
it. And was like your own

913
01:06:22,503 --> 01:06:26,932
personal agent and could be your
the business version of you and

914
01:06:26,932 --> 01:06:31,223
your home version of you and did
all of these things and could

915
01:06:31,223 --> 01:06:35,445
manage the complexities of your
life. Super cool, right? Like

916
01:06:35,445 --> 01:06:39,805
that's probably the next thing
that might happen with this kind

917
01:06:39,805 --> 01:06:44,235
of technology. It would be super
cool. Yeah. But there's so many

918
01:06:44,235 --> 01:06:48,595
problems we have to figure out
in order for Someone like me and

919
01:06:48,595 --> 01:06:53,025
potentially you to even want to
give it all of our data And then

920
01:06:53,025 --> 01:06:56,969
what it what do we do when
people want to advertise to us

921
01:06:56,969 --> 01:07:01,191
with that data? And how do we
set our own boundaries? And how

922
01:07:01,191 --> 01:07:05,136
do we deal with the complex
social interactions that come

923
01:07:05,136 --> 01:07:09,496
out of that? Like, okay, so you
and I have personal assistants.

924
01:07:09,496 --> 01:07:13,718
That are these agents, right?
Yeah. And you, you say, you say

925
01:07:13,718 --> 01:07:17,594
to your agent, hey, see if
Melonee wants to have dinner.

926
01:07:17,594 --> 01:07:21,401
Okay. And my agent is like,
well, Melonee's already got

927
01:07:21,401 --> 01:07:25,415
plans with friends that we are
both friends with. Oh yeah,

928
01:07:25,415 --> 01:07:29,498
yeah. But you weren't invited,
right? And so do you want my

929
01:07:29,498 --> 01:07:33,997
agent to go to your agent and be
like, well, you know, and so how

930
01:07:33,997 --> 01:07:38,426
do you even keep secrets between
agents? How do you define that?

931
01:07:38,426 --> 01:07:42,717
How do we like, and so I think
it's a super interesting space,

932
01:07:42,717 --> 01:07:46,870
but I think there's a lot of.
Social fallout that we haven't

933
01:07:46,870 --> 01:07:51,092
thought about. And it's the same
with all of HRI, right? Like

934
01:07:51,092 --> 01:07:55,452
there's all this contextualized
social interaction and it's not

935
01:07:55,452 --> 01:07:59,397
just, you know, in quotes,
social interaction. It's, it's

936
01:07:59,397 --> 01:08:03,272
highly contextualized to our,
you know, region, our, our

937
01:08:03,272 --> 01:08:07,425
cultural backgrounds, things
like that. And. I don't know if

938
01:08:07,425 --> 01:08:11,301
we're ready or we're like, we're
at that point where the

939
01:08:11,301 --> 01:08:14,000
technology has that
sophistication yet.

940
01:08:14,000 --> 01:08:16,725
Audrow Nash: that's a good
point. Yeah, I guess. Yeah,

941
01:08:16,725 --> 01:08:19,853
there's a whole bunch of things
that need to be worked out and

942
01:08:19,853 --> 01:08:22,679
you really have to see how
people react. So this kind of

943
01:08:22,679 --> 01:08:25,000
thing can't really be rushed, I
would imagine.

944
01:08:25,000 --> 01:08:27,950
Melonee Wise: Well, I mean,
they'll always be the people at

945
01:08:27,950 --> 01:08:30,600
the forefront and they'll be
learning all the painful

946
01:08:30,600 --> 01:08:31,000
lessons.

947
01:08:31,000 --> 01:08:34,982
Audrow Nash: Yep. Yeah, that
distribution of like the early

948
01:08:34,982 --> 01:08:39,368
adopters and this kind of thing.
What do you, so one thing that's

949
01:08:39,368 --> 01:08:43,552
been interesting to me, there's
again, from a recent space, we

950
01:08:43,552 --> 01:08:47,736
had someone come on, that's like
just about to graduate with a

951
01:08:47,736 --> 01:08:51,853
degree in computer science where
they, feel like it's a hard.

952
01:08:51,853 --> 01:08:55,969
Like, I think the feeling is
because of things like chat GBT,

953
01:08:55,969 --> 01:08:59,816
that there's no reason to go
into something like computer

954
01:08:59,816 --> 01:09:03,730
science because you're just
going to get in a thing and be

955
01:09:03,730 --> 01:09:07,712
automated right away for this
kind of thing. I think that's

956
01:09:07,712 --> 01:09:11,693
the feeling. I wanted to hear
your thoughts on it. Like, do

957
01:09:11,693 --> 01:09:15,000
you think it's a good time to be
a roboticist or,

958
01:09:15,000 --> 01:09:20,037
Melonee Wise: I think it's a
great time to be a roboticist.

959
01:09:20,037 --> 01:09:25,159
This is the, probably the next
50 years are going to be like

960
01:09:25,159 --> 01:09:30,280
the heyday of robotics. Like of,
of mobile robotics, mobile,

961
01:09:30,280 --> 01:09:35,317
mobile manipulation robotics. I
think from 1960 to 2000 was

962
01:09:35,317 --> 01:09:40,012
probably the heyday of
industrial robotic arms. 2000 to

963
01:09:40,012 --> 01:09:45,390
2015 was probably the heyday of
collaborative robotic arms. you

964
01:09:45,390 --> 01:09:50,000
know, 2014 till now and going
forward is the heyday of

965
01:09:50,000 --> 01:09:54,951
autonomous mobile robots. You
know, I, I think that. And I

966
01:09:54,951 --> 01:09:59,561
think that, that now is the
time. I mean, I think that

967
01:09:59,561 --> 01:10:04,341
Willow Garage really kicks
something off and we're like,

968
01:10:04,341 --> 01:10:09,805
we're in it. So yeah, you should
definitely become a roboticist.

969
01:10:09,805 --> 01:10:11,000
It's our time.

970
01:10:11,000 --> 01:10:15,249
Audrow Nash: That's my feeling
too. I have a feeling that it is

971
01:10:15,249 --> 01:10:19,633
probably the best time so far to
be a roboticist. What do you, so

972
01:10:19,633 --> 01:10:23,343
for someone who's feeling lost,
especially with all the

973
01:10:23,343 --> 01:10:27,053
advancing technology. What
advice do you have for them?

974
01:10:27,053 --> 01:10:30,965
Like, how, how do they, how do
they get a foothold in this

975
01:10:30,965 --> 01:10:34,000
world? And I don't know, how do
they do well?

976
01:10:34,000 --> 01:10:39,000
Melonee Wise: Learn to, yeah,
learn to program, learn ROS.

977
01:10:39,000 --> 01:10:43,245
Audrow Nash: Hell yeah. The, the
thing that I was thinking, it's,

978
01:10:43,245 --> 01:10:47,163
It's almost like everyone gets a
bunch of junior programmers

979
01:10:47,163 --> 01:10:51,212
beneath them, for working. And
it's like with JGPT, you get an

980
01:10:51,212 --> 01:10:54,804
assistant for this kind of
thing. It's like everyone is

981
01:10:54,804 --> 01:10:58,461
kind of their own CEO, and you
have a bunch of people to

982
01:10:58,461 --> 01:11:02,576
delegate to that are all AIs, is
how I've been thinking of it a

983
01:11:02,576 --> 01:11:06,690
bit. It's not so much. Like you
still have a lot of autonomy in

984
01:11:06,690 --> 01:11:10,673
how you choose to move forward
and you don't have to do a lot

985
01:11:10,673 --> 01:11:14,592
of the grunt work, from these
things. And like, I mean, I'm,

986
01:11:14,592 --> 01:11:18,445
I'm using Angular, which is a
complex web framework made by

987
01:11:18,445 --> 01:11:22,363
Google in work for Intrinsic.
And, I am up and running super

988
01:11:22,363 --> 01:11:26,216
fast because of ChatGPT. It's
like, that's a, that's a real

989
01:11:26,216 --> 01:11:27,000
power there.

990
01:11:27,000 --> 01:11:32,667
Melonee Wise: Yeah. And I think
the other thing is, is. I don't

991
01:11:32,667 --> 01:11:37,885
know what your early career
space was like as an engineer,

992
01:11:37,885 --> 01:11:43,372
but there, there was this time
early in my career where I was

993
01:11:43,372 --> 01:11:48,950
always afraid to ask a question,
you know, and like, go bother

994
01:11:48,950 --> 01:11:54,437
someone who knew more than me.
And now you don't have to, you

995
01:11:54,437 --> 01:11:59,655
can bother ChatGPT, you know,
like, so your ability to ask

996
01:11:59,655 --> 01:12:05,142
questions and to fail. You know,
or, or ask dumb questions is

997
01:12:05,142 --> 01:12:10,000
totally enabled by having this
knowledge consolidator.

998
01:12:10,000 --> 01:12:13,607
Audrow Nash: Yeah, I think
that's a good point. I do think

999
01:12:13,607 --> 01:12:17,152
there's an interesting case
there where it's actually not

1000
01:12:17,152 --> 01:12:20,822
very good at answering questions
about the bleeding edge of

1001
01:12:20,822 --> 01:12:24,864
things. And so It's very good at
answering everything that's kind

1002
01:12:24,864 --> 01:12:28,782
of in the like very well known
space, but I wonder, I feel like

1003
01:12:28,782 --> 01:12:32,763
that's something to be a little
wary of, where if you're like, I

1004
01:12:32,763 --> 01:12:36,370
don't know, how do I do this
super hard thing and it gives

1005
01:12:36,370 --> 01:12:40,350
you a super generic answer, like
talking to you or someone who's

1006
01:12:40,350 --> 01:12:43,957
like actually done very hard
things is going to be so much

1007
01:12:43,957 --> 01:12:47,502
more informative than just
asking these same questions to

1008
01:12:47,502 --> 01:12:48,000
ChatGPT.

1009
01:12:48,000 --> 01:12:51,904
Melonee Wise: Sure, sure, but
that's when you're in the mid

1010
01:12:51,904 --> 01:12:55,676
cycle of your career. I was
speaking like early stages of

1011
01:12:55,676 --> 01:12:59,912
your career You just got out of
university and you're struggling

1012
01:12:59,912 --> 01:13:03,750
with something that you know You
feel like would be almost

1013
01:13:03,750 --> 01:13:06,000
wasting your mentor's time,
right?

1014
01:13:06,000 --> 01:13:10,630
Audrow Nash: Yes, that is a very
good point. Yeah, you could just

1015
01:13:10,630 --> 01:13:14,548
bug ChatGPT. Yeah, what an
interesting thing. So we are

1016
01:13:14,548 --> 01:13:18,893
Coming to the end of the time
one of the things that I wanted

1017
01:13:18,893 --> 01:13:23,523
to ask about And we've just been
talking about other things, but,

1018
01:13:23,523 --> 01:13:27,868
so. Agility was doing Amazon
trials, with their robots. Yeah.

1019
01:13:27,868 --> 01:13:32,000
How did that go? And tell me a
bit about that. It's great.

1020
01:13:32,000 --> 01:13:36,373
Melonee Wise: We, you know,
there were a whole bunch of,

1021
01:13:36,373 --> 01:13:41,059
news media articles put out.
There's a bunch of videos. It's

1022
01:13:41,059 --> 01:13:46,135
going really well. so we've been
working with Amazon for. Quite a

1023
01:13:46,135 --> 01:13:50,899
while. in, in some different
applications, you saw one of the

1024
01:13:50,899 --> 01:13:55,585
applications in some of the
videos that were highlighted, as

1025
01:13:55,585 --> 01:14:00,349
part of their demo day. And now
we're moving on to some other

1026
01:14:00,349 --> 01:14:04,956
phases of the project, which are
really exciting. and we're

1027
01:14:04,956 --> 01:14:10,032
continuing to work with them on,
on these projects and excited to

1028
01:14:10,032 --> 01:14:13,000
start deploying more robots with
them.

1029
01:14:13,000 --> 01:14:17,370
Audrow Nash: Yeah, that's so
cool. Are they, how does it, I

1030
01:14:17,370 --> 01:14:22,111
suppose you have other customers
too, but they're one of the big

1031
01:14:22,111 --> 01:14:26,630
ones and they're one of the,
like, I mean, you said they have

1032
01:14:26,630 --> 01:14:31,296
750, 000 robots, if I remember
correctly, Kiwa robots, which is

1033
01:14:31,296 --> 01:14:35,519
bonkers. So, I mean, there's
clearly lots of potential to

1034
01:14:35,519 --> 01:14:40,037
scale. I think that's really
awesome, but you're, you're also

1035
01:14:40,037 --> 01:14:44,185
working with other companies
too. Oh, yeah. And all very

1036
01:14:44,185 --> 01:14:48,407
same, similar use case at the
moment, which is picking up

1037
01:14:48,407 --> 01:14:53,074
those totes. Okay. Super cool.
I'm glad that's going well. It's

1038
01:14:53,074 --> 01:14:57,444
so cool. I, I really, it's
exciting to see a humanoid robot

1039
01:14:57,444 --> 01:15:02,185
that's doing a really. Practical
job in a sense and making a lot

1040
01:15:02,185 --> 01:15:06,556
of sense for the ROI of these
like justifying itself with a

1041
01:15:06,556 --> 01:15:11,148
good return on investment for
the companies that are investing

1042
01:15:11,148 --> 01:15:15,741
in it because I think that was
my big skepticism for the space

1043
01:15:15,741 --> 01:15:20,185
was I think it'd be hard to get
a good ROI for a lot of the.

1044
01:15:20,185 --> 01:15:23,000
More complex ones, at least
initially.

1045
01:15:23,000 --> 01:15:26,066
Melonee Wise: Yeah, but
customers wouldn't be working

1046
01:15:26,066 --> 01:15:29,364
with us if they didn't believe
that there was a return on

1047
01:15:29,364 --> 01:15:30,000
investment.

1048
01:15:30,000 --> 01:15:34,784
Audrow Nash: Hell yeah. Let's
see. So what, what do you think

1049
01:15:34,784 --> 01:15:39,255
is the future? Like tell me the
next two, five years. For

1050
01:15:39,255 --> 01:15:42,000
agility, where are you guys
headed?

1051
01:15:42,000 --> 01:15:46,649
Melonee Wise: Yeah. So over the
next couple of years, we're

1052
01:15:46,649 --> 01:15:51,061
going to be very focused on
expanding our, let's call it

1053
01:15:51,061 --> 01:15:55,552
skill set. So, you know, as I
was telling you, we look at

1054
01:15:55,552 --> 01:16:00,043
Digit as a platform that has
composable skills. and as we

1055
01:16:00,043 --> 01:16:04,850
start working with more and more
customers, we're going to be

1056
01:16:04,850 --> 01:16:09,813
expanding the set of skills that
Digit has, so that would be in

1057
01:16:09,813 --> 01:16:14,226
the areas of. You know, tote
manipulation, but also tote

1058
01:16:14,226 --> 01:16:18,796
stacking, destacking, tote, tote
wrangling, those types of

1059
01:16:18,796 --> 01:16:23,523
things. But then also moving
into other types of containers,

1060
01:16:23,523 --> 01:16:27,699
like corrugate boxes,
palletization, depalletization.

1061
01:16:27,699 --> 01:16:32,663
And so it's just Looking at the,
the space of, of activities in

1062
01:16:32,663 --> 01:16:37,075
the warehouse and slowly
branching out across a swath of

1063
01:16:37,075 --> 01:16:41,803
similar activities within the
warehouse. So there's a lot of

1064
01:16:41,803 --> 01:16:46,609
processes that require taking
some kind of container, whether

1065
01:16:46,609 --> 01:16:51,336
it's a tote or a box, um, from a
shelf or to a shelf. From a

1066
01:16:51,336 --> 01:16:56,300
conveyor or to a conveyor, from
a cart or to a cart. and so now

1067
01:16:56,300 --> 01:17:00,949
you've got all these skills.
It's like, okay. And then now,

1068
01:17:00,949 --> 01:17:05,361
if you know how to take
something to and from a conveyor

1069
01:17:05,361 --> 01:17:10,404
and to, and from a shelf and to,
and from a cart, now you can go

1070
01:17:10,404 --> 01:17:15,525
from a shelf to a conveyor, from
a conveyor to a cart. Right. And

1071
01:17:15,525 --> 01:17:20,647
so we look at it as, Building up
of a composable skill space that

1072
01:17:20,647 --> 01:17:25,690
then eventually can be deployed
into different applications. And

1073
01:17:25,690 --> 01:17:30,496
eventually, Digit has all the
skills to form the basis for an

1074
01:17:30,496 --> 01:17:35,145
app store for labor. And then
you start looking at the, the

1075
01:17:35,145 --> 01:17:39,636
workflows and the tasks that
Digit can do. And, you know,

1076
01:17:39,636 --> 01:17:43,969
building out the next thing
based on the skill set that

1077
01:17:43,969 --> 01:17:48,539
Digit already has. And so as you
start gathering all these

1078
01:17:48,539 --> 01:17:53,345
skills. It's like any person.
The more skills you can do, the

1079
01:17:53,345 --> 01:17:55,000
more jobs you can do.

1080
01:17:55,000 --> 01:17:58,379
Audrow Nash: I like that. Yeah.
It's a, it's a cool thing. You

1081
01:17:58,379 --> 01:18:01,213
keep building the capabilities
that keeps opening up

1082
01:18:01,213 --> 01:18:04,591
applications that keeps letting
you grow your market. And then

1083
01:18:04,591 --> 01:18:07,643
it's just, it's like a nice
flywheel in a sense. And you

1084
01:18:07,643 --> 01:18:10,749
said app store at some point
like that. Yeah. Yeah. It's,

1085
01:18:10,749 --> 01:18:14,074
it's so interesting that app
stores are, like it's kind of, I

1086
01:18:14,074 --> 01:18:15,000
guess maybe where

1087
01:18:15,000 --> 01:18:17,000
Melonee Wise: we'll call it a
skill store. Fine.

1088
01:18:17,000 --> 01:18:21,386
Audrow Nash: Skill store. Yeah.
Yeah. Skillstore. But it's just,

1089
01:18:21,386 --> 01:18:25,634
it's so interesting because I,
in a lot of these interviews, I

1090
01:18:25,634 --> 01:18:29,746
talk with them and then it's
like the long term vision is to

1091
01:18:29,746 --> 01:18:34,063
get to something like a, an app
store or a skill store for this

1092
01:18:34,063 --> 01:18:38,106
kind of thing. And it makes a
lot of sense because then you

1093
01:18:38,106 --> 01:18:41,326
have the diversity of
application that, like it

1094
01:18:41,326 --> 01:18:45,506
becomes generally useful. Do
you, do you think that, Humanoid

1095
01:18:45,506 --> 01:18:49,618
robots are going to become like,
will they be like the silly

1096
01:18:49,618 --> 01:18:53,935
metaphor, but will they be like
the spreadsheet of the computer

1097
01:18:53,935 --> 01:18:58,046
age where it's like, you have
something flexible enough that

1098
01:18:58,046 --> 01:19:02,158
it justifies things, places
getting one. And then from that,

1099
01:19:02,158 --> 01:19:06,338
like you get it, it does its
core application and now a bunch

1100
01:19:06,338 --> 01:19:10,244
of people are buying it, but
then you can also add other.

1101
01:19:10,244 --> 01:19:13,602
Programs on it, that also
provide some value like

1102
01:19:13,602 --> 01:19:16,000
spreadsheets, justifying
computers.

1103
01:19:16,000 --> 01:19:20,726
Melonee Wise: I think so. I
think, I think the thing is, is

1104
01:19:20,726 --> 01:19:25,532
in the industrial landscape,
it's harder to do that because.

1105
01:19:25,532 --> 01:19:30,338
The primary motivator for a
return on investment is, is like

1106
01:19:30,338 --> 01:19:35,144
the primary work task. but I
think when you look at like, if

1107
01:19:35,144 --> 01:19:39,229
you look at like, retail
applications or storefront

1108
01:19:39,229 --> 01:19:44,436
applications, you know, Forgoing
the fact that Pepper had limited

1109
01:19:44,436 --> 01:19:49,322
utility because it had limited
utility, if Pepper was a fully

1110
01:19:49,322 --> 01:19:53,968
capable humanoid robot, like,
you could believe Then yeah,

1111
01:19:53,968 --> 01:19:58,614
then you would have kind of
Pepper's primary activity, but

1112
01:19:58,614 --> 01:20:03,420
there was probably a whole bunch
of other things that Pepper

1113
01:20:03,420 --> 01:20:07,825
could have done where LLMs,
ironically enough, would be

1114
01:20:07,825 --> 01:20:12,551
perfect. Yeah, for sure. Like
imagine walking into a retail

1115
01:20:12,551 --> 01:20:17,117
store and saying, Hey Pepper,
can you help me find a pink

1116
01:20:17,117 --> 01:20:22,244
blouse? Or, you know, then that
becomes very interesting in that

1117
01:20:22,244 --> 01:20:26,649
space. But I think when the
return on investment In the

1118
01:20:26,649 --> 01:20:31,295
industrial application is very.
You know, task oriented as

1119
01:20:31,295 --> 01:20:35,781
opposed to like a retail or
grocery space or a, hardware

1120
01:20:35,781 --> 01:20:40,587
store, for example, you know,
like if you ever tried to find

1121
01:20:40,587 --> 01:20:45,473
someone to help you find the
screw you're looking for at Home

1122
01:20:45,473 --> 01:20:50,279
Depot or Lowe's. It's, it's like
searching for a needle in a

1123
01:20:50,279 --> 01:20:51,000
haystack.

1124
01:20:51,000 --> 01:20:53,605
Audrow Nash: And then they're
walking somewhere and now you're

1125
01:20:53,605 --> 01:20:56,000
following them. Yeah, this kind
of thing. Yeah, for sure.

1126
01:20:56,000 --> 01:20:59,729
Melonee Wise: And so, but
imagine if, if the utility of

1127
01:20:59,729 --> 01:21:03,458
those types of robots is, is
more of what we're talking

1128
01:21:03,458 --> 01:21:07,322
about, where maybe the, the
original application that you

1129
01:21:07,322 --> 01:21:11,254
maybe have is, restocking for
the robot, but when it's not

1130
01:21:11,254 --> 01:21:15,458
restocking, an individual can
walk up to it and ask it to find

1131
01:21:15,458 --> 01:21:16,000
a screw.

1132
01:21:16,000 --> 01:21:21,202
Audrow Nash: Yeah. Or even while
it's restocking, just like the

1133
01:21:21,202 --> 01:21:26,321
people that are working there.
Okay, very interesting. do you,

1134
01:21:26,321 --> 01:21:30,532
let's see, like, I guess,
wrapping up, what are you

1135
01:21:30,532 --> 01:21:34,000
excited about in robotics now,
in general?

1136
01:21:34,000 --> 01:21:39,160
Melonee Wise: Hmm, what am I
excited about in robotics? I

1137
01:21:39,160 --> 01:21:44,411
don't know. I, I think I'm, I
think I'm most excited maybe

1138
01:21:44,411 --> 01:21:50,024
about, the, the growing interest
in making robots usable. and,

1139
01:21:50,024 --> 01:21:55,456
and I, I, I think that that's,
that's something that's still

1140
01:21:55,456 --> 01:22:01,159
going to take us a long time as
a community, but I'm, I'm very,

1141
01:22:01,159 --> 01:22:06,772
excited by the progress we're
making there. I, I somewhat wish

1142
01:22:06,772 --> 01:22:11,571
there was like. an academic
version of an, industrial

1143
01:22:11,571 --> 01:22:17,274
conference that was more like an
academic conference where like

1144
01:22:17,274 --> 01:22:22,615
companies could go and just
present their HRI work. I think

1145
01:22:22,615 --> 01:22:27,957
it would be very interesting
because one of the things that

1146
01:22:27,957 --> 01:22:33,389
you, when you look out into the.
Or their usability work for

1147
01:22:33,389 --> 01:22:38,911
robotics is there aren't a lot
of places for us to talk about

1148
01:22:38,911 --> 01:22:44,796
it as a community and, and a lot
of the research has limited data

1149
01:22:44,796 --> 01:22:49,865
sets that are limited to, to
like university students or

1150
01:22:49,865 --> 01:22:54,664
whatever users they could
scrounge up on a Sunday. as

1151
01:22:54,664 --> 01:23:00,367
opposed to some of the companies
that have, you know, thousands

1152
01:23:00,367 --> 01:23:05,708
of hours of interactions with
hundreds of people at a time,

1153
01:23:05,708 --> 01:23:11,231
kind of data sets. and, and I'd
really, I. I wish that we had

1154
01:23:11,231 --> 01:23:16,482
more of a community and an
opportunity to talk about and a

1155
01:23:16,482 --> 01:23:22,185
venue for talking about kind of
how do we advance usability for

1156
01:23:22,185 --> 01:23:23,000
robotics.

1157
01:23:23,000 --> 01:23:27,500
Audrow Nash: Mm hmm. Why do you
think someone's not doing that?

1158
01:23:27,500 --> 01:23:31,643
Or is it a new idea? Because I
think it seems like a great

1159
01:23:31,643 --> 01:23:32,000
idea.

1160
01:23:32,000 --> 01:23:36,493
Melonee Wise: probably some of
it is proprietary work. Like, I

1161
01:23:36,493 --> 01:23:40,913
will admit that I had a very
strong interest in it, but Fetch

1162
01:23:40,913 --> 01:23:45,551
never showed any of its UI ever.
It like, like, there's very few

1163
01:23:45,551 --> 01:23:47,000
videos of it online.

1164
01:23:47,000 --> 01:23:51,033
Audrow Nash: Yeah, gotcha.
Everyone's holding their cards

1165
01:23:51,033 --> 01:23:55,420
close to their chest for that
kind of thing. Yeah. Okay. Well,

1166
01:23:55,420 --> 01:23:59,736
uh, do you have any links or
contact info you'd like to share

1167
01:23:59,736 --> 01:24:02,000
with our watchers and listeners?

1168
01:24:02,000 --> 01:24:06,152
Melonee Wise: I don't know. I'm
on Twitter, Twitter or X,

1169
01:24:06,152 --> 01:24:10,669
whatever it's called these days.
And, at Melonee Wise, I mean,

1170
01:24:10,669 --> 01:24:13,000
all my handles are Melonee Wise.

1171
01:24:13,000 --> 01:24:16,482
Audrow Nash: Okay. Hell yeah.
And I'll put a link to Agility

1172
01:24:16,482 --> 01:24:20,254
in the episode. Hell yeah. Okay.
Well, it's been great talking to

1173
01:24:20,254 --> 01:24:24,027
you. And hearing your opinion on
a lot of things. it's an awesome

1174
01:24:24,027 --> 01:24:26,000
perspective and I really value
it.

1175
01:24:26,000 --> 01:24:29,750
Melonee Wise: Awesome. it's nice
seeing you and hopefully we'll

1176
01:24:29,750 --> 01:24:31,000
grab a beer sometime.

1177
01:24:31,000 --> 01:24:35,220
Audrow Nash: Hope so. All right.
See ya. That's it. I, for one,

1178
01:24:35,220 --> 01:24:39,038
had my opinions changed on
humanoids from this interview,

1179
01:24:39,038 --> 01:24:43,057
but what did you think? Do you
agree with Melonee that we're

1180
01:24:43,057 --> 01:24:46,742
not going to see humanoids
outside of manufacturing and

1181
01:24:46,742 --> 01:24:50,627
logistics for 10 years or so?
What other low hanging fruit

1182
01:24:50,627 --> 01:24:54,646
might humanoids be used for? If
you're not already, consider

1183
01:24:54,646 --> 01:24:58,665
subscribing to never miss an
interview and I'll see you next

1184
01:24:58,665 --> 01:24:59,000
time.
