00:00:00
gecko robotics builds wall climbing
00:00:02
robots and enterprise software to
00:00:04
maintain and protect essential
00:00:06
infrastructure he is the CEO and
00:00:08
co-founder at Gecko robotics Jake thanks
00:00:11
for joining me today every hour you save
00:00:14
you're saving potentially millions of
00:00:15
dollars Hardware is hard be very careful
00:00:19
there's a lot of really important
00:00:20
problem to solve there's so much sex
00:00:23
appeal to building new things you got to
00:00:25
get the business model right and the
00:00:26
business model has to make a CEO or CFO
00:00:29
give a
00:00:31
there is a huge fire going on right now
00:00:33
at Philadelphia Energy Solutions oh my
00:00:35
gosh again look at this guys look at
00:00:38
this video right now today the Navy
00:00:40
remains a formidable fighting force but
00:00:43
even officers within the service have
00:00:45
questioned its Readiness at a missile
00:00:47
silo we visited time and frigid weather
00:00:50
had clearly taken their toll developing
00:00:53
right now gushing for hours with no end
00:00:55
in sight thousands of barrels of crude
00:00:58
oil spilling from a
00:01:00
uh the report does an estimate of what
00:01:02
the need is to bring the overall grade
00:01:04
up to a b which is what the society sort
00:01:06
of determines to be adequate and it's
00:01:08
like
00:01:09
$4.59 trillion
00:01:13
[Music]
00:01:19
[Music]
00:01:27
[Music]
00:01:34
[Music]
00:02:02
yeah all
00:02:03
right hi I'm Jake the founder and CEO of
00:02:07
gecko robotics a company that makes
00:02:09
robots and software to help diagnose the
00:02:12
health of the built world now it started
00:02:15
in a college dorm my college dorm is now
00:02:18
a company that manages over 500,000 of
00:02:21
the world's most important and critical
00:02:22
pieces of
00:02:24
infrastructure now the structures that
00:02:26
we use to power
00:02:28
civilization have reached their useful
00:02:30
life it's a huge problem and it's
00:02:33
getting way worse but it's a problem
00:02:35
that you probably don't think about very
00:02:36
much but you should in New York for
00:02:39
example there are over 177,000 Bridges
00:02:42
most of which are in New York City and
00:02:44
guess how many of those bridges are not
00:02:46
in need of immediate repairs only
00:02:51
six see maintaining things has always
00:02:54
been an afterthought but that
00:02:57
afterthought is now a 4.59 trillion
00:03:00
domestic
00:03:02
problem and by the way it's getting
00:03:05
worse it's holding us back for example
00:03:07
the military spends 40% of their budget
00:03:11
over $400 billion on
00:03:14
maintenance not on building new things
00:03:16
just keeping old things working and
00:03:19
Fortune 500 companies will lose $1 1.5
00:03:22
trillion every single year because of
00:03:26
catastrophic failures that were
00:03:27
unpredictable
00:03:30
and our best defense to stop that from
00:03:32
happening hasn't changed in over 60
00:03:35
years it's this this is Joe and Joe's on
00:03:41
a rope now Joe's armed with a handheld
00:03:43
sensor and what looks like an
00:03:45
excruciating
00:03:48
wedgie now Joe is our best chance to
00:03:52
ensure that pipelines don't explode that
00:03:54
Bridges don't collapse that dams don't
00:03:57
fail and that airplanes don't
00:03:58
disassemble mid-flight
00:04:00
it's an impossible job unfortunately for
00:04:04
Joe you see we obsess about how software
00:04:07
has changed everything for everyone it's
00:04:10
eating the world right it's important to
00:04:12
remember that for the guys behind me
00:04:15
it's actually never help them you see
00:04:18
the data that we need to prevent
00:04:20
catastrophes from happening in the built
00:04:21
World simply doesn't
00:04:24
exist and without data what can software
00:04:28
do now I became obsessed with this
00:04:30
problem in college I was studying
00:04:31
electrical engineering and my obsession
00:04:34
for energy took me to a local power
00:04:35
plant in Pennsylvania I wanted to se
00:04:38
power was made and so I decided to dive
00:04:41
in head first now I I actually dove in
00:04:44
head first straight to this hole and
00:04:47
when you got through this hole you got
00:04:48
into a 200 foot tall steel tubed box the
00:04:51
length and the width of a football field
00:04:54
this is a power plant boiler and the
00:04:55
boiler's job was to turn water into
00:04:57
steam by getting really hot see the
00:05:00
problem is as the plant manager Jeff
00:05:02
told me that 40% of the time this boiler
00:05:04
would be shut down because of pressure
00:05:06
tube
00:05:08
explosions it would cost them $2 million
00:05:10
every single day they were down this is
00:05:11
a small little power plant and so I
00:05:14
asked Jeff how do you stop this from
00:05:16
happening and he says well we send up
00:05:17
humans on ropes looking for invisible
00:05:22
defects then he began to tear up and he
00:05:25
told me a story about how his best
00:05:27
friend fell and died the year before
00:05:31
doing one of these inspections and he
00:05:33
fell and died in the exact spot I was
00:05:35
standing so I was floored by the
00:05:38
story and so I had to do something about
00:05:40
it so I went back to my college storm
00:05:43
and started building the First wall
00:05:44
cling robot now I armed this robot with
00:05:47
ultrasonic sensors just like doctors use
00:05:49
for sonograms and I deployed that robot
00:05:52
into the boiler saving the plant manager
00:05:54
Jeff $30 million just that year and I
00:05:57
became absolutely obsessed
00:06:00
with how we understand the health of the
00:06:02
built structures that we use every
00:06:05
single day and so I started a
00:06:07
company and I boot shaed that company
00:06:10
for three years pouring my life savings
00:06:12
into it I slept on my best friend's
00:06:14
apartment floor and I was down to the
00:06:17
last
00:06:18
$100 two things happened within two
00:06:20
weeks first I got an offer to buy the
00:06:22
company from a company that makes power
00:06:25
plants and then second two partners from
00:06:27
a group called y combinator said that if
00:06:29
I stay poor and kept on building the
00:06:31
vision that one day gecko would change
00:06:33
everything that we knew about built
00:06:35
structures so I decided to stay poor and
00:06:38
keep building the vision so we launched
00:06:40
the company in
00:06:41
2016 and we began to deploy the
00:06:44
technology into the oil and gas
00:06:46
manufacturing public infrastructure and
00:06:48
even defense sectors we had to build
00:06:50
robots that could climb and Traverse and
00:06:52
get sensors into all different kinds of
00:06:54
surfaces geometries conditions and once
00:06:57
you have 500,000 assets you have to
00:06:59
climb around you begin to iterate your
00:07:01
robots really really well to be able to
00:07:03
handle these kinds of environments and
00:07:05
it became clear that contrary to popular
00:07:08
belief from the VCS at the time the
00:07:10
robots were the mo because they could
00:07:12
get sensors to places that could never
00:07:14
be gotten to before we can convert atoms
00:07:17
into bits and so I wanted to double down
00:07:20
on that moat and so we started building
00:07:21
robots that could fly swim crawl and
00:07:25
walk up any surface we begin to build
00:07:28
autonomous platform
00:07:30
to arm those robots to be a to go to
00:07:31
places so that humans didn't have to be
00:07:33
in dangerous environments we built and
00:07:35
became the best in the world of building
00:07:37
ultrasonic sensors electromagnetic
00:07:39
sensors as well as lasers to be able to
00:07:42
see and understand what was going on
00:07:43
inside of Steels composits and concrete
00:07:46
we built fixed sensors that could stream
00:07:49
live information and data sets to us
00:07:52
both the health but also the operational
00:07:54
conditions of those assets itself and we
00:07:57
built an API platform for robots called
00:08:00
so that other robotic companies could
00:08:02
actually be used on our platform and
00:08:04
streaming data and information live to
00:08:05
our customers and after 10 years of
00:08:08
collecting data on almost every
00:08:09
structure imaginable we launched
00:08:13
Canal our Ai and Robotics powered
00:08:16
operating platform to put those data
00:08:18
layers to use for our customers you see
00:08:20
when you start building software by
00:08:23
first starting out with the data layers
00:08:24
and then building up you're severely
00:08:27
advantaged because you can build
00:08:28
software from principles and our
00:08:31
ontology now was able to affect folks
00:08:33
from the ground level the guys on the
00:08:35
ropes all the way through to the
00:08:36
Executive Suite it was extremely
00:08:38
powerful but post is talking about it
00:08:41
let's actually dive into an example so
00:08:43
to do that I'm going to take you to
00:08:44
Georgia to a manufacturing facility that
00:08:46
me and you me and you use in the
00:08:48
bathroom every single
00:08:50
day now this facility has thousands of
00:08:53
assets and billions of dollars worth of
00:08:55
infrastructure and so they wanted us to
00:08:57
prove out over 50 assets what we could
00:08:59
actually do so I'm going to take you
00:09:01
through one of those assets today a
00:09:03
sulfuric acid
00:09:04
tank so first what we do is we gather
00:09:07
information about the asset by customers
00:09:09
sending us their metadata and we build
00:09:11
out a digital
00:09:13
representation of that asset inside of
00:09:15
Canever and then we send in our robots
00:09:19
first we use a drone you can see over
00:09:20
here now the Drone is armed with with
00:09:23
cameras that are doing a photog gometric
00:09:25
scan of the asset it enriches the asset
00:09:27
model itself being able to identify
00:09:29
different kinds of defects using Point
00:09:31
clouds so corrosion areas like over here
00:09:34
we're able to categorize and locate and
00:09:37
then dents and cracks as well enriching
00:09:39
the data asset model we want to go
00:09:41
further than that we incorporate other
00:09:43
sorts of components like piping and
00:09:45
pumps that you see here both in the
00:09:46
inlet and Outlet this is extremely
00:09:49
important and valuable because we keep
00:09:51
on adding data layers and the next one
00:09:54
is a
00:09:56
dog come here boy
00:10:01
good
00:10:02
dog
00:10:05
hey everyone say hi please yeah there we
00:10:08
go nice so let me let me pet him real
00:10:10
quick he likes to be pet yeah good
00:10:12
boy so this dog will walk around to
00:10:15
dangerous environments gathering
00:10:17
information about what's going on on the
00:10:19
infrastructure now what's important also
00:10:21
to understand is that because of an API
00:10:23
platform for robots we've built an
00:10:25
extensible way for a company like
00:10:27
anybotics one of our partners to be able
00:10:29
to gather information in data sets and
00:10:31
this robot is extremely exciting because
00:10:34
it's built to be explosion proof meaning
00:10:35
it can go inside of oil and gas
00:10:37
facilities nuclear facilities and Beyond
00:10:40
it's gathering information like you see
00:10:41
above thermal imaging to understand
00:10:43
what's going on with the asset all this
00:10:45
data is really important when we do
00:10:47
optimizations later now we want to begin
00:10:50
to continue to gather more information
00:10:52
about the asset so we send in
00:10:53
submersible robots these submersible
00:10:56
robots are looking at the deformation
00:10:58
because of the weight the liquid as well
00:11:00
as the health of the assets floor to
00:11:03
prevent things like that um oil tanker
00:11:06
leaking into rivers and once we've
00:11:08
gotten this and the customer is really
00:11:10
excited because we can do this W they're
00:11:12
online we then send in our robots and we
00:11:15
rise collecting information and data
00:11:18
while the C while the customer's tanks
00:11:20
are still in operation now these robots
00:11:23
that you see right here are armed with
00:11:24
ultrasonic sensors cameras and the IMU
00:11:27
on board to ensure that we can do this
00:11:28
autonomously Gathering terabytes of data
00:11:31
in 12 hours for this tank a process that
00:11:33
used to take about a month to do while
00:11:35
the asset was shut down costing millions
00:11:37
of dollars and we can do this in a way
00:11:39
as well that ensures we can localize
00:11:41
data points to begin to run
00:11:42
optimizations and in this case because
00:11:46
these assets are supposed to be reaching
00:11:48
their useful life or reached it we can
00:11:50
extend the useful life of the
00:11:52
infrastructure and so for this example
00:11:54
we can tell them what to fix in five and
00:11:57
10 years to extend the useful life so
00:11:59
that this asset continues to be able to
00:12:02
uh do its function opposed to having to
00:12:03
replace it for $8 million which was what
00:12:06
the plant thought they would have to do
00:12:08
so once we do that predictive model we
00:12:09
work with maintenance companies to
00:12:11
ensure that they actually take action on
00:12:12
that and we update the model to ensure
00:12:15
that a source of Truth
00:12:17
remains next we actually the customer
00:12:20
actually wants us to begin to do other
00:12:22
optimization so we use fix sensors like
00:12:24
this that'll stream information about
00:12:26
not just the
00:12:27
health but also the operating condition
00:12:30
of the facility you see when you're
00:12:33
running a let's say a big manufacturing
00:12:34
facility your goal is to figure out how
00:12:37
to make more product without having
00:12:39
stuff blow up U because of a new
00:12:41
operating condition now that's never
00:12:42
before been possible because the data
00:12:45
you've been able to work with has been
00:12:46
from Joe on a rope and so you don't know
00:12:49
if you change your throughput or make
00:12:51
more product if that'll destroy the
00:12:53
assets we're able to run
00:12:56
optimizations now I'm going to show you
00:12:57
that here so this customer was able to
00:12:59
because we're streaming information and
00:13:01
data from the pumps and from the asset
00:13:03
itself we were able to figure out how to
00:13:06
increase the throughput or make more
00:13:08
product by about 5% more while only
00:13:11
having to incur over 90 days an
00:13:13
accelerated um an accelerated damage of
00:13:15
the asset of about two months equivalent
00:13:17
to two months and so we proved that we
00:13:19
could do that by actually lowering the
00:13:21
fill Heights in the tank and increasing
00:13:23
the acid concentration level you can see
00:13:25
the optimization being run right behind
00:13:27
me now this was significant
00:13:29
because of the ability to not just
00:13:31
extend the useful life but actually
00:13:33
produce more while not having the
00:13:35
potential risk of a catastrophic failure
00:13:38
something never before possible for
00:13:39
these
00:13:40
companies now let's talk about the
00:13:42
outcomes for the customer on average
00:13:43
over the 50 assets we extended the
00:13:45
useful Life by 10 years this affects
00:13:47
their p&l and their margin right away
00:13:49
because of ability to extend your
00:13:50
depreciation models and then we created
00:13:52
$15 million um of value by being able to
00:13:56
reduce safety risks as well as
00:13:57
Environmental as well as being able to
00:14:00
reduce the amount of CBS that customers
00:14:01
needed to spend and the estimated from
00:14:03
the customer was a 4% impact to their
00:14:05
margin now all of this optimization and
00:14:08
information coming in doesn't just help
00:14:09
the customer it also helps can lever be
00:14:12
exceptional in a compounding way at
00:14:14
running facilities more efficiently and
00:14:16
so now you have an ability to have an
00:14:19
unfair Advantage from from companies
00:14:21
that are not utilizing technology like
00:14:22
this so not just are robots cool but
00:14:25
they're actually solving a business
00:14:27
problem
00:14:29
so this has been flying off the shelf as
00:14:30
you can imagine since we launched
00:14:31
Canever this year the 12th largest oil
00:14:34
and gas company in the world for example
00:14:36
determined that they have 100,000 tanks
00:14:39
and that we could provide $122,000 of
00:14:41
Roi per tank now initially we signed a
00:14:43
$30 million contract it's exciting it's
00:14:45
going to extend to 100 million but it
00:14:48
shows how if you adopt technology in a
00:14:50
way like this it's
00:14:54
unfair now on the defense side we're
00:14:57
working with the uh Air Force on $130
00:14:59
billion modernization program now they
00:15:02
have to modernize over 400 nuclear
00:15:04
missile silos and the best way to
00:15:06
determine how to modernize or what the
00:15:08
scope was to improve the missile silos
00:15:11
was I kid you not Joe on a rope with a
00:15:14
hammer who was listening to the sound
00:15:15
that the silos made when he hit
00:15:19
it so now gecko is helping to improve
00:15:23
what the modernization scope actually
00:15:24
should be and it points out something
00:15:26
interesting those that are determining
00:15:28
the scope and size of these
00:15:30
modernization Pro projects are the same
00:15:32
ones incentivized for that amount of
00:15:34
dollars to be as high as it can
00:15:37
be now on the Navy side one of the
00:15:39
biggest problems is only a third of our
00:15:40
ships are available to patrol and deter
00:15:43
conflict around the world and the reason
00:15:44
why is because of Maintenance
00:15:47
Cycles so we worked with uh the Navy to
00:15:50
improve in this case it was Joe on a
00:15:53
skateboard on his belly over a flight
00:15:55
deck looking at different different
00:15:56
areas trying to gather information and
00:15:58
data set
00:15:59
we improve that to be able to reduce
00:16:01
labor by 85% and improve the turnaround
00:16:03
times for Flight deex Alone by about a
00:16:05
month so now we're doing tens of
00:16:07
Millions with a navy on flight decks and
00:16:08
we're extending that to ballast tanks
00:16:10
hulls as well as commercial Maritime
00:16:13
it's really exciting and then on the
00:16:15
energy manufacturing sector that's where
00:16:16
most of our 78 Fe accounts lie with big
00:16:18
contracts like xon BP and
00:16:21
Beyond now one thing that's extremely
00:16:24
exciting is that it turns out if
00:16:26
pipelines explode or when oil leaks into
00:16:30
rivers it's pretty bad for the
00:16:32
environment so the studies show that by
00:16:35
2030 in the US you can reduce emissions
00:16:38
by about 18% if you can stop those kind
00:16:40
of things from happening so technology
00:16:42
is available today to make a drastic
00:16:45
impact on Net
00:16:47
Zero and then it turns out as well if
00:16:50
you're the best in the world of
00:16:51
understanding the health of built
00:16:52
structures you're actually very
00:16:53
advantaged in building new things and so
00:16:55
that's what the Admiral in charge of
00:16:56
$132 billion nuclear um submarine
00:17:00
project called the Columbia class
00:17:02
determined so now we're helping to
00:17:03
create the most advanced submarine in
00:17:05
the world from the beginning to the
00:17:07
production end and it gives you a peak
00:17:09
into what's
00:17:11
coming now this is why it
00:17:14
matters you see I'm not crazy building
00:17:18
robotics Material Science AI software
00:17:21
sensor company it's really freaking hard
00:17:23
um but I had no choice you see the
00:17:26
promise of AI from AI compan
00:17:29
to make impacts in these industries have
00:17:32
gone empty for years and years and years
00:17:34
and it's no wonder why they're building
00:17:37
their foundational models off of Joe's
00:17:41
data data that looks like
00:17:44
this this is a real report from one of
00:17:48
our customers before they used gecko
00:17:50
it's no wonder that AI hasn't made the
00:17:53
impact in of the promise that it was
00:17:55
supposed
00:17:56
to so this is why we built geeko and why
00:17:59
I believe because of software being
00:18:01
commoditized that first order data
00:18:03
companies will dominate the next 10 and
00:18:05
20 years in
00:18:07
software and my journey through the rust
00:18:09
has given me both a pragmatism and
00:18:11
optimism about the future a future where
00:18:13
understanding how things work helps you
00:18:16
build new things understanding how to
00:18:18
use Ai and Robotics in these real
00:18:21
practical
00:18:23
ways a reality where we can understand
00:18:26
the health of the built structures all
00:18:27
around us just as as well as we
00:18:29
understand our own
00:18:30
health and you begin to see robots of
00:18:33
course in normal society but these
00:18:36
robots won't be built for doing back
00:18:38
flips or folding laundry they're going
00:18:41
to be built to help realize the impact
00:18:44
of AI for the built world with systems
00:18:47
like Canever thank
00:18:52
you thank you that was awesome
00:18:57
yeah come here puppy oh David Zach
00:18:59
showed up
00:19:01
everybody doggy hey Z that's a robotic
00:19:04
dog it's kind you're kind of dog
00:19:08
oh go to sax go to sax go give him a
00:19:13
kiss oh there we go SX is very
00:19:16
affectionate so you can pet him if you
00:19:18
want David you can pet him s here where
00:19:21
where do you pet him exactly there you
00:19:24
go that was a lot of love yeah I am
00:19:27
experiencing oh
00:19:29
companionship from this
00:19:32
excited yeah it's definitely it's
00:19:35
definitely a nice dog yeah nice
00:19:38
dog doesn't
00:19:41
bitey I think one thing that would be
00:19:43
great based on the kinds of customers
00:19:45
you have can you tell us a little bit
00:19:47
about the sales life cycle and the type
00:19:50
of deals you do I mean it's so
00:19:52
interesting to is it like an Enterprise
00:19:55
software type sale and you know when
00:19:57
you're going in and doing
00:19:59
physical work I mean I didn't know where
00:20:01
to sit were you worried about the dog no
00:20:03
this was it's a very poorly organized
00:20:06
conference yeah let's talk about that
00:20:08
yeah we we only told you where to sit
00:20:10
five times in the last four minutes but
00:20:12
I don't understand these images I have
00:20:14
you know literally we go in there after
00:20:16
and like Sergey comes he does his first
00:20:18
thing and I'm like Ser oh yeah do you
00:20:20
have any food I go out to the food and
00:20:21
it's just like rubber conference
00:20:23
chicking in the VIP speaker area and I'm
00:20:25
like free birg can we just get some
00:20:27
sushi from noou we should go through all
00:20:29
the details yeah no tell us about the
00:20:32
sales life cycle so what are the kinds
00:20:33
of deals you not talking about rubber
00:20:36
chicken right now we've got a panel uh
00:20:38
but yeah tell us about the sales life
00:20:39
cycle the juice has been really good by
00:20:40
the way I like thank you um yeah so so
00:20:42
the life cycle it's been it's been it's
00:20:44
been wild so you know gecko actually
00:20:46
became profitable in 2017 right after
00:20:48
YC's launch in 2016 and so you were a YC
00:20:51
company yes
00:20:52
2016 and um what was interesting about
00:20:55
that was you know we decided to build a
00:20:57
company very for deployed so instead of
00:20:59
building robots and Labs actually funny
00:21:01
story one of the VCS you had here last
00:21:04
year offered a bunch of money at YC um
00:21:08
for us not to leave and go back to
00:21:10
Pittsburgh and do this forward deployed
00:21:12
motion of building robots but instead
00:21:13
build it in a lab and I turned that down
00:21:16
because I just fundamentally didn't
00:21:17
believe in that way of
00:21:18
building and so um but so we decided to
00:21:22
launch into the um and build robots like
00:21:25
I've literally soldered in these
00:21:26
environments um before and and just
00:21:29
figure out how to make the robots work
00:21:30
in reality in the real world and so the
00:21:33
sales motion was basically we would go
00:21:34
to the plant managers sometimes I'd call
00:21:36
and be like hey my pizza guy like you
00:21:38
know where's the where's the plan
00:21:39
manager can I talk to him and I'd get
00:21:41
figure out how to get to the plan
00:21:42
managers and then I'd uh convince them
00:21:44
to let us work with them in their
00:21:46
facilities and so started out that way
00:21:49
by selling to the folks who need this
00:21:52
the most yeah and so but now I'm talking
00:21:54
to um obviously CTO and and CFOs because
00:21:57
our products are actually very Financial
00:21:59
um it helps with depreciation models it
00:22:01
helps with optimizations but we started
00:22:03
by just selling to the folks in the
00:22:05
ground and building the robot um by
00:22:07
failing a bunch of times there and
00:22:10
fixing it live but now we have uh a
00:22:12
great platform and so now when customers
00:22:14
buy gecko the only way they can buy it
00:22:16
is through software so they buy a can
00:22:17
lever and they bought they buy an
00:22:19
implementation of the software which is
00:22:21
the robots getting the data and then
00:22:23
they pay for a license um for the
00:22:25
software and we try to make data
00:22:27
refreshes which is basically robots
00:22:28
going out and collecting more
00:22:29
information free is there a custom is
00:22:32
there a custom deployment in every one
00:22:33
of these because they've all got to have
00:22:34
different facilities and how hard is it
00:22:37
to kind of customize or do you have
00:22:38
standard standardization now in each
00:22:40
deployment they kind of do a Chinese
00:22:42
menu type selection we try so so it's a
00:22:46
great question we we started in the
00:22:48
beginning by letting the customers pick
00:22:49
what kinds of data layers they want they
00:22:51
want so data layers basically mean what
00:22:53
kind of robots now we actually don't
00:22:55
allow them to do that we follow all the
00:22:57
standards whatever like API which is
00:22:59
like these governing bodies about how to
00:23:01
take care of your infrastructure um but
00:23:02
then we go way beyond that because I
00:23:04
want to create an incredible user
00:23:05
experience that they cannot revert back
00:23:07
from Jak there's um there's all kinds of
00:23:10
crippling infrastructure problems around
00:23:12
the world that are not necessarily tied
00:23:14
to some of the obvious Industries like
00:23:16
oil and gas so I'll give you two
00:23:18
examples one was what happened in
00:23:19
Baltimore where you know this who knows
00:23:23
how it happened but basically the bridge
00:23:25
just collapsed in a situation that and
00:23:26
maybe it was supposed to be and and it
00:23:29
did another example was a few years ago
00:23:31
in genua in Italy an entire slab of a
00:23:33
bridge just collapsed and it fell on top
00:23:36
of uh um an environment and killed a
00:23:39
bunch of innocent civilians
00:23:41
so there's I think a public safety
00:23:45
requirement here which is like some of
00:23:46
this stuff was either designed poorly or
00:23:48
designed very quickly how much of that
00:23:51
is observable by these kinds of robots
00:23:53
and how do you convince folks that
00:23:56
Beyond depreciation and Financial
00:23:58
motivations there's a you know a real
00:24:00
need to make sure that this public
00:24:01
infrastructure is safe and you guys can
00:24:03
secure it um great question so the
00:24:07
answer to of how can we actually get
00:24:09
information on those types of incidents
00:24:11
is yes um we we like you can look at a
00:24:13
concrete bridge and say hey there's some
00:24:15
Decay here or there's something that's
00:24:18
happening in the giring here and you can
00:24:19
recognize that and learn and be able to
00:24:23
say wait a minute you need to send
00:24:24
inspectors or shut the bridge down or
00:24:27
stop and figure this out you first want
00:24:29
to do so what the robots are really good
00:24:30
at is getting a crap ton of data about
00:24:32
the assets and then you can pinpoint
00:24:34
exactly where to put fixed sensors in
00:24:36
specific locations that would be
00:24:37
indicative but then also because of our
00:24:40
of of our you know we have this like
00:24:42
really interesting data set that tells
00:24:44
us because of so many different types of
00:24:46
U situations um what kinds of potential
00:24:50
issues are occurring that we can
00:24:52
extrapolate out to these types of
00:24:54
situations that we might be not as
00:24:55
familiar with so we'll put fixed sensors
00:24:57
on to give us indication
00:24:58
and give us an ability to help
00:25:00
prioritize spending and so um you know
00:25:03
we just actually signed a a contract
00:25:05
with Governor Shiro to do this for
00:25:07
bridges in elany county in Pennsylvania
00:25:09
where Pittsburgh is and um we're helping
00:25:12
to modernize um Bridge maintenance and
00:25:15
prioritization of budget because what
00:25:17
you can see here is you don't
00:25:18
necessarily need to rebuild stuff and in
00:25:20
some ways that's not even practical but
00:25:21
you can figure out where to deploy
00:25:23
Capital um and then by way did you see
00:25:26
this video on X where somebody was going
00:25:27
through the Lincoln Tunnel and it looked
00:25:29
like it was about to burst there was
00:25:31
like water creaking in it was really
00:25:34
disconcerting yeah but I think it was
00:25:37
more of a design feature to actually
00:25:38
like alleviate when times when the water
00:25:40
levels were super high but my point is
00:25:43
there's all of the stuff that we
00:25:44
interact with that it would be good to
00:25:45
know that there's a you know a service
00:25:48
out there looking for it and is is there
00:25:50
a world where you could also then
00:25:51
theoretically ingest like the actual
00:25:53
architectural orad of these things and
00:25:56
then also be able to do diffs and VAR
00:25:58
and be able to tell people hey hold on a
00:26:00
second this is not conforming to how we
00:26:02
thought it should be behaving yes we do
00:26:05
we do pull those in as much as we can
00:26:07
but it's important to remember that most
00:26:08
of the infrastructure that I'm talking
00:26:09
about is like 60 years old um now on the
00:26:12
new build side like for the new Columbia
00:26:15
class submarines for example um there's
00:26:17
an issue where like there's not a
00:26:18
digital thread you have like 5,000
00:26:20
different contractors that are trying to
00:26:21
make us most powerful sub in the world
00:26:23
and they're handing paper to each other
00:26:25
um as they build the submarine and so
00:26:28
causes one a big a bunch of delays and
00:26:30
issues which we're seeing with a lot of
00:26:31
our ability like China for example can
00:26:33
outbuild Us by 232 times uh submarines
00:26:36
that is or or new chips and it's a a big
00:26:39
part of like our our like we have to be
00:26:41
able to figure out how to be smarter
00:26:43
when we manufacture and so one of the
00:26:44
ways you can do that is digital threads
00:26:47
all the way through the manufacturing
00:26:48
process so that we're not like delayed
00:26:50
by handing paper to each other that may
00:26:52
or may not be incorrect and for the
00:26:54
customers that we work with you know
00:26:56
most of them you know you're looking at
00:26:57
drawing that are 60 years old they have
00:27:00
never been converted um I we even try to
00:27:02
get asset lists from customers and they
00:27:04
like we don't have it so we have to go
00:27:06
out and actually build that for them how
00:27:08
should we build a submarine just off
00:27:09
topic how should we build it yeah so we
00:27:11
don't have 5,000 contractors at well 1%
00:27:15
or 2% the speed of China it's a good
00:27:17
it's a good question I think we should
00:27:18
Orient our most efficient um way of
00:27:20
building as many components in one place
00:27:21
as we can but you have to remember as
00:27:23
well you know Congressional members have
00:27:25
their own constituents to advocate for
00:27:27
and so they want to bring jobs to their
00:27:29
to the local communities and so in a in
00:27:31
a democracy it's really tough actually
00:27:33
MH Jake how do you see
00:27:36
um where your Bots your robots and let's
00:27:40
say the more traditional generalized
00:27:42
humanoid robots intersect when when they
00:27:43
meet how do you think about that problem
00:27:45
oh it's a great question um I am so
00:27:47
excited to buy as many Optimus robots as
00:27:50
possible yeah you're a customer you'll
00:27:52
be a customer 100% you're not going to
00:27:53
be a competitor no no no I mean um look
00:27:57
at anybotics right here so this is a
00:27:58
this is a swed it's a company based in
00:28:01
Switzerland that have an incredible
00:28:03
robot and their data doesn't know where
00:28:05
to go and so the idea of robots this is
00:28:08
what I firmly believe is that taking
00:28:09
them home today is that um you know they
00:28:12
get sensors to places that are really
00:28:14
hard to get sensors to and so that
00:28:16
information has to be funneled um
00:28:18
somewhere to drive some large business
00:28:20
outcome so I really don't think like you
00:28:22
know I I'm not of the belief I guess
00:28:25
that when Elon talks about you know two
00:28:27
times the amount of um robots as humans
00:28:29
that you'll see them in society actually
00:28:32
as much as you maybe you'd think I think
00:28:34
they're actually going to be found
00:28:35
mostly in these like really dangerous
00:28:37
behind the scenes industrial settings
00:28:40
which I in my opinion that's like where
00:28:41
they should start for sure because
00:28:43
you'll have to you one you have to those
00:28:45
are really complex tasks but two they're
00:28:47
like very beneficial for Humanity um so
00:28:50
like is there is there a generalized
00:28:52
platform that you've built that allows
00:28:53
you to solve for these different use
00:28:54
cases or do you find that there's a lot
00:28:56
of application specific engineering
00:28:58
that's required good question so we we
00:29:01
we built an API platform for robots
00:29:04
where um companies can try their systems
00:29:07
out and we can because we have a go a
00:29:09
market we can now test if that robot is
00:29:11
producing something valuable from a data
00:29:12
side right and I'm am not actually as
00:29:15
interested in robots I can weld or
00:29:17
robots I can clean right now mostly I'm
00:29:19
just interested in just like what kind
00:29:20
of information and data can we build
00:29:21
better operating platforms and systems
00:29:23
on and um so that's that's where I'm
00:29:26
starting and we'll begin to add more
00:29:27
robots that can do different kinds of
00:29:29
jobs but um you know I think it's you
00:29:32
have I think this is where first ordered
00:29:34
data sets and and and software companies
00:29:36
are become more and more you know
00:29:37
powerful this is why maybe you know um
00:29:40
my opinion on on like the the Standalone
00:29:43
SAS model is like I think it's going
00:29:45
away um because the companies that are
00:29:47
so advantaged with first order data you
00:29:49
can you know just build your own
00:29:51
software have a capital markets last
00:29:54
question have Capital markets kind of
00:29:55
embraced the story or things
00:29:58
you can invest okay yeah um yeah they
00:30:01
they really are I think um because
00:30:04
there's a lot of this like hesitation
00:30:05
around deep Tech and Hardware
00:30:06
historically but you've obviously got an
00:30:08
incredible software layer and great
00:30:10
recurring business so you seem to be
00:30:12
pretty differentiated in terms of a lot
00:30:13
of the Long Haul build cycles that I
00:30:16
think we see out there and it's yeah in
00:30:20
Hardware in our case it's very sticky
00:30:22
yeah because you know once you convert
00:30:24
from you know paper and you're now not
00:30:26
using binders you're using can
00:30:28
yeah um it's it's really hard to go back
00:30:31
yeah I just yeah I think it's we've
00:30:33
talked before and I think what you're
00:30:34
doing is so inspiring because it feels
00:30:36
to me like this technology you know that
00:30:39
we've seen over the last 20 years in
00:30:41
iPhones in EVS in sensors um is now
00:30:45
allowing us to move from being reactive
00:30:47
and trying to figure out what happened
00:30:49
to then saying hey let's be proactive
00:30:51
what are the opportunities here to
00:30:53
extend life to say you know extend the
00:30:55
life of these assets um and avoid
00:30:58
tragedies and I just you know I think
00:31:01
the work you're doing is a real
00:31:03
Grinder's work but it's going to save
00:31:05
lives and it's going to really save
00:31:07
taxpayers money that could be deployed
00:31:08
in other places for beautiful things and
00:31:10
so I just want to commend you on doing
00:31:12
something that is so essential um in
00:31:15
many times I meet a founder and they are
00:31:18
doing something and I just think God I I
00:31:21
don't know if this is going to work or
00:31:22
not but I know that this founder is
00:31:25
going to figure out a way to make it
00:31:26
work and I I think it's just really rare
00:31:28
that somebody cares so much about
00:31:30
something and then executes as hard as
00:31:32
you have um and I just want to tell you
00:31:34
I personally very much appreciated it
00:31:36
because you know when a bridge collapses
00:31:39
I know the one in Italy it's it was a
00:31:40
very big tragedy I think it killed 50
00:31:42
odd people and it was privately owned
00:31:44
yeah and there's a you know very very
00:31:46
wealthy family that owned it and
00:31:48
ultimately sadly no repercussions yeah
00:31:52
same with dams in Brazil same with dams
00:31:53
in Brazil right so you know as public
00:31:56
infrastructure becomes private
00:31:57
infrastructure then the profit motive
00:31:58
supersedes a safety motive you're going
00:32:00
to get all these things unless there's
00:32:02
um um some sort of check and balance so
00:32:05
long way of saying we very much
00:32:06
appreciate this hard work that you've
00:32:07
dedicated last question yeah you're
00:32:09
you're you know you you've been in the
00:32:10
YC Community for a long time just really
00:32:13
you don't have to say yes or no but have
00:32:15
you ever been forced to do any founder
00:32:18
mode did you feel pressure have you just
00:32:21
just blink twice Jake just tell us the
00:32:23
truth Jake thank you for joining us Jake
00:32:29
that's awesome that was awesome that was
00:32:31
excellent
