// transcript — 401 segments
0:00 Intro
0:09 all right well welcome everybody thank you all so much for coming it's so nice
0:13 to be able to do these events in person and see so many faces in the audience
0:17 i'm nina shadyan i'm a partner at index ventures and i'm so excited today to
0:22 have the opportunity to speak with alexander wang the founder and ceo of
0:28 scale ai alex welcome thanks so much for uh having me and nina by the way
0:31 congratulations on making it to the midas list brinklets oh thank you thank
0:36 you so much news justin came out yesterday thank you well you're too kind
0:40 well the the story today is really about the incredible journey
0:45 that scale ais had so tell us maybe just how did the whole story begin what was
0:51 were you thinking about this problem how did you come across this massive market
0:55 opportunity and just kind of give us a sense of the early days of the company
0:58 building part yeah so i always like to start this story with my background because i think
1:00 Background
1:04 that you know it's always hard to disintegral um your early background
1:09 from how you end up starting a company but i grew up in los alamos new mexico
1:13 both my parents were physicists at los alamos national labs and so i grew up
1:19 with this deep appreciation for science technology you know always thinking
1:22 about what were the what were the upcoming technologies how were those
1:25 going to affect the world what were the were like the inner workings behind
1:29 those technologies and so that was really like the the petri dish so to
1:33 speak that i was i grew up in and uh then when i when i uh when i went to
1:38 college and i was sort of like was studying computer science and
1:41 programming um there was this this is back in 2015 there's this crazy new
1:45 technology um called ai and in particular deep learning and uh you know
1:51 this was the early results were you know there was this famous um algorithm that
1:55 could tell whether or not there were cats in youtube videos um this was
2:00 probably 2013 i think um and so it was like this this technology that seemed
2:03 really really magical and when you know growing up as a programmer um you know
2:06 you despite how powerful computers are you realize how how limited they are you
2:13 know uh programming is is a lot of uh giving very very clear and simple
2:17 directions to computers to be able to accomplish incredible things but again
2:20 you're really limited it's very black and white and if as i thought about the
2:24 implications of artificial intelligence and deep learning it created this
2:28 potential for in judgment and intelligence and sort of like gray area decision making to be
2:34 added to computers which was very powerful so is this incredibly inspiring
2:39 technology um and then uh i was so they were studying it at mit very excited had
2:43 some incredible professors something in great detail and then i wanted to build
2:48 a side project so i wanted to build a camera inside my refrigerator that would
2:52 tell me when my roommates were stealing my food and uh yeah it's like a very college problem
2:58 you know thankfully that was not the startup idea but uh the next thing was a
3:01 startup idea and and i started working on it for just for a few weeks and i
3:04 realized there's just no way to get enough data to actually build it um and
3:09 uh and so it was like hmm this is gonna be a problem you know in what way there
3:12 was like too much footage for you to go through to figure out like what actually
3:16 was being taken when or yeah well if you if you think about the all the data you
3:20 need to build this algorithm effectively you need data about all the different
3:24 kinds of products that you might have in a fridge in all the different
3:27 configurations as well as data about all the different ways in which somebody
3:30 could go and reach in grab something maybe somebody just like tries to grab
3:34 something at first and you'll grab something else and you know there's all
3:37 these different configurations this sort of like very complex machine learning
3:40 problem and not unlike self-driving which i'm sure we're going to talk about
3:44 in a bit it just required just massive amounts of really high quality machine
3:48 learning data sets they're impossible for me to get as a as a developer and so
3:52 and i was lucky you know this was in 2016 and uh you know aws and stripe had already been
3:59 these massive successes really built upon this idea of how do you how do you
4:04 empower like the college dorm room developer with infrastructure that is is
4:09 vastly beyond sort of their their reach um and uh and so that was that was sort
4:13 of like this inspiring idea and applied to ai i realized you know how do you
4:17 empower the college a dorm room developer with just incredible ai
4:21 technology um and and data was really the first the first step of that and so
4:25 that was really the the genesis behind scales building you know the the
4:29 defining company focused around data for machine learning got it so how did the
4:30 Early Pitch
4:34 early pitch go with investors did people right away understand the idea and back
4:38 it because you were very early in really identifying this problem and being a
4:41 solution you know i actually remember there was this like one early pitch i won't uh i
4:48 won't name the vc but uh there's one early pitch it was it was uh uh you know
4:53 where i we had this whole pitch deck and talking about ai and the importance of
4:57 ai et cetera i go to the first slide and it's like ai needs lots and lots of data
5:01 something something along those lines was the text on the on the um text on
5:06 the the slide and then the uh one of the vcs interrupts me like i was one
5:11 sentence into the presentations like you know i heard from so and so stanford
5:15 professor that uh actually you don't need that much data for ai and i was
5:20 just and that was he was just like is this immediate confrontation and i was
5:24 like um i'm not sure you know there probably is like a lot of
5:27 context behind what they were saying but no i can assure you that like if you
5:30 want really high performing ai you need lots and lots of data and uh and so
5:33 there's a lot i just was immediately met with skepticism so um that's one of many
5:38 examples but the the the core just was like yeah early on when you're when
5:42 you're pitching a startup in the very early days and you have no proof points
5:46 behind what you're building um it's very easy for a lot of venture
5:51 capitalists to just say oh no that like this idea is bad or this idea is silly
5:55 or this idea is stupid and um and i think it's it's pretty natural honestly
5:59 because i think you know especially for a new technology like artificial
6:02 intelligence where frankly we've been discovering a lot about ai even over the past few years
6:08 there have been a number of major breakthrough results even in the past
6:11 year or two um as we're discovering more about it i think you know everybody understands more
6:16 about it but in the early days back in 2016 you know it wasn't that well
6:22 understood a technology and so um you know it's easy for people to say no so
6:26 needless to say i'm thankful to index uh for having believed in scale
6:31 even from the very early days um but uh but it was that was not everyone you
6:35 know it was a it was a tough tough journey to pitch it early on wow
6:37 Challenges
6:39 well after you got your initial funding and you had the money in the bank tell
6:42 us a little bit about what the first couple years at scale look like i mean
6:46 you must have interacted or actually encountered so many different
6:49 challenges when looking back what were some of those things where at the moment
6:53 you thought like this is insurmountable but now looking back it's amazing to see
6:58 you you know conquer those challenges yeah so you know at least
7:02 the way i think about is like early on in a company you basically have this
7:07 like you're you're constantly um fighting two very important battles one
7:11 is getting more customers or users depending on the product um and the
7:15 other one is getting more building your team and getting more
7:19 great people to join you in your mission and both of these things are incredibly
7:23 incredibly hard and for us for example we were trying to convince you know the smartest ai teams
7:28 in the world a lot of them were self-driving companies which had who again had
7:33 attracted some of the the most brilliant ai scientists and ai teams again in the
7:39 world to conv to use our product and really rely a very important part of
7:45 their stack on on scale and we were you know we were this very scrappy young
7:49 team you know at first it was just like you know we were a team of two pitching
7:52 these great ai scientists they were maybe we were team of four pitching this
7:55 great ai scientist but we were always this always a sort of like very very
8:00 small team trying to convince these like extremely well-funded companies to to
8:03 rely on us and those that was really really hard and i think
8:07 you know there are a number of there were a number of things that we did to
8:10 actually enable us to get over that hump but that was you know if i were to think
8:14 about like some of the hardest pitches i've had to make or the hardest um
8:18 like uh the hardest arguments i've had to make it's really in convincing these
8:22 very well resource organizations to to rely on a on a small startup and the
8:26 same thing is true on the on the employee side you know um when you're
8:29 when you're a very small company obviously you have huge ambitions and a big dream and a big
8:36 a big goal but it's a it's a big jump for uh someone to make to join the team
8:41 and actually you know be ready to ride in the trenches of a of a crazy
8:45 fast-moving startup and um particularly for scale we we got into a spot pretty
8:50 early on where thankfully we did pretty good job getting customers but that
8:53 meant that we were really behind in getting building a team and so we were always
9:00 way overloaded from a demand perspective relative to the team that we had and so
9:04 every time i would meet with a candidate they would be like wow you guys seem
9:06 really really uh tired and you seem like you're you're working
9:12 really hard you're like yes we must you're like hence the open job role yeah
9:16 yeah we have an open job role and if you have five friends you know bring them
9:21 please yeah um but uh but it was so that was just the that that sort of like
9:25 balance um was was always the trickiest part early on i mean it's truly
9:26 The Aha Moment
9:30 remarkable you must have been 21 22 a team of two to four people you were
9:33 going to these you know incredible companies that had world-class ai scientists like you
9:40 said what was the aha moment for them in the pitch where they're like all right
9:43 we'll sign on the dotted line yeah so one of the one of the the secrets to scale
9:50 and i think this applies across almost every industry but um the the problem
9:54 that we saw that we originally were solving which is around building really
9:57 high quality data sets was something that most machine learning teams knew was
10:01 very important and so most people agreed that was very important but it wasn't
10:06 necessarily the sexiest problem um that uh that you know every ai scientist
10:09 wanted to spend their days and nights working on and this is like collecting
10:14 images labeling them yeah collecting images and labeling them and making sure
10:17 that was done very very high quality was a very you know frankly very operational
10:20 problem and we've built a lot of technology um uh to support it but again
10:24 it's at its core it's a very operational problem and you know there's one there's
10:29 one article that that was pretty seminal for me early on which was um
10:33 it was an article by paul graham or an essay by paul graham i should say called
10:38 schlep blindness and uh you know i encourage everyone to read it if you get
10:42 a chance but um basically the idea was that like you know most people avoid thinking about you
10:49 know the really sort of just like hairy ugly difficult annoying problems that
10:53 exist in the world but they're really really important and he's actually used
10:57 a stripe as one of the examples in his in his essay but um there's there's
11:01 these problems are everywhere you know the sort of like ugly hairy problems
11:05 that um that uh everyone knows are important but like aren't sexy to work
11:09 on and if you can identify what those problems are then those generally make
11:14 really exciting um startup ideas so um so that was kind of uh that was a lot
11:18 of the the the original pitch was like you know this is important but um
11:24 but you probably aren't the most excited to work on it and then we were just you
11:28 know i think we even very early on our whole team was super duper scrappy and so i think a lot
11:34 of teams and a lot of customers that we work with like saw our velocity and our
11:38 product velocity and how fast we were moving and and and thought to themselves
11:42 like hey yeah these these guys even if they don't have the perfect part today
11:45 they're going to get to a product that we're going to be able to rely on really
11:49 really quickly yeah i mean it's truly incredible you said you started the
11:54 company in 2016 and the company today a short you know a couple of years after
11:59 that is worth 7.5 you know seven and a half billion dollars which is incredible
12:03 maybe just some advice for the entrepreneurs that are sitting in the
12:07 audience i mean how did you do it how did you stay focused um and anything you
12:12 know now that you wish you knew when you started the journey
12:16 yeah you know it's it's um there was this uh there's a there's a thing that i that i
12:22 realized um uh early on in starting the company which was going to be a
12:26 uh you know the art of being a good of a good founder or a good ceo is being able to
12:35 constantly reinvent yourself and and be able to tackle the new challenges of the
12:39 company at every stage of the company and you know i think that's really
12:42 really true the challenges that we had when we were sort of um very early on
12:46 just getting our first few customers were very very different from our
12:49 challenges when we were sort of building the initial team up to like 30 people
12:52 which are very different from the challenges to get from that to
12:56 100 people which are very different from the challenge of getting that scale and
13:00 sort of a single product company to building multiple products and expanding
13:03 to new verticals and so there's sort of been this um constant cycle of of of
13:10 reinvention required at the company i think one thing that you know i don't
13:13 know if this would if i would have been happier or sadder to know this very
13:16 early on in the company but it's just gonna it's gonna be this um you know if
13:20 you have the right mindset it's gonna be this incredibly rewarding journey of of
13:25 sort of like learning and constantly improving and constantly identifying how
13:29 to like be the best at every stage of the company but it's also um it's also a
13:34 lot of work and you're gonna have to um you're never gonna feel
13:37 like oh i've like got this down i'm like finally like i finally feel like i've
13:44 got a handle on the problems you're always gonna feel on the back foot and
13:47 like you're you're trying to keep up with um with the company you know one of
13:51 the expressions that someone told me very early on when we were in this mode
13:56 where we had way more demand than than uh the team to support it is that you
14:00 know he said like wow you guys have a tiger by the tail and i think that like
14:03 truly has been how i felt even today at scale which is like you
14:09 know you if if you're doing this i should say if uh the part of the
14:12 rewarding part about scale has been this feeling of having a tiger by the tail
14:16 and sort of feeling like we're always trying to catch up to what our customers want or what we know
14:22 we can do or what we know our potential is and and i think that that's an
14:26 incredibly rewarding feeling um but uh but it it can be it's it's a real it's a real journey and
14:33 so i maybe the one big thing i would say you know what i we we found this we
14:37 ended up doing this but not because we were like brilliant about it early on
14:40 but i would say it's really really important for yourself your co-founders
14:45 your early team um and and everyone who gets involved in the company early on to
14:48 really be in it for quite a journey um because it is going to be
14:54 if everything goes well like this incredibly rewarding but also also
14:58 incredibly challenging very difficult you're gonna have to work really hard
15:02 you're gonna have to like you know constantly be uh constantly fall down
15:05 have to get back up like it's it's a real um it's a real track it's a real journey
15:11 and so i think that making sure that you and everyone else working on it are prepared for that is
15:16 Mission of Scale
15:16 really important thank you for sharing that i think you touched on this a little bit but we you
15:22 know as investors in the company have been so incredibly surprised to the
15:27 upside at just how massive the opportunity is for scale and you have touched already so many
15:34 different industries right you started with transportation now you've helped
15:38 the logistics supply chain e-commerce government tell us a little bit about
15:43 how you can traverse so many huge industries and really how you frame the mission of
15:48 scale today yeah so so early on even going back to how i started the company we like the
15:57 we knew that data was very very important for ai and you know one of the
16:02 one of the ways that um i explain sort of the strategy of scale to to new hires
16:07 to join the company is you know if you think about it ai is an extremely
16:11 general purpose technology it has the ability to be applied in every major
16:17 industry in the world um but what's but so so our mission is to
16:21 accelerate development ai applications but it's really to how do we enable ai
16:26 across every single major industry in every transformative use case there is
16:29 um if you were to think about how do you design a company to be able to do that um you know you
16:36 need to solve the key bottlenecks in that process of applying ai in every
16:40 single industry and that bottleneck again if you look at it just industry
16:43 after industry is really all around the data and the data sets and so the
16:48 the way in which we built scale and the strategy of scale has really been hey
16:52 first we're going to focus on how do you build the best possible technology that
16:56 enables data sets to be built as high quality with as high volume as quickly
17:01 as possible and then once we have that as a sort of core technology of the
17:05 company and a core infrastructure platform then it's really easy to go and
17:09 apply that to every single major industry so we applied it initially to
17:13 autonomous vehicles transportation as you mentioned autonomy automotive sector
17:18 then we applied it to government and defense um and that's been a uh that's
17:23 been a really exciting area for us to work on and something that i care a lot
17:27 about deeply as an american um then we applied it to uh to e-commerce
17:32 so we work with you know like instacart and etsy and we're applying it to sort
17:36 of in this like massive revolution of e-commerce globally uh then we applied
17:40 it to financial services so we work with companies like brex and empowering a lot
17:45 of the next-gen financial services features like enabling you know better
17:48 expense management for example that's all happening through partnership with
17:52 uh with braxton enabling them with the best in class ai functionality there's
17:57 working with flexport on global logistics and you know logistics is
18:01 still a uh global shipping is still a very paper and manual process and that the ability
18:07 to transform that entire industry through digitizing and enabling powering
18:12 all this uh data with with artificial intelligence to enable more automated
18:15 and intelligent decision making is really powerful and so what we've done
18:19 is you know again by starting at the core around how do we enable companies
18:23 with the best data that enables us to tackle every industry super super efficiently and super
18:30 quickly and our aspirations at this point you know we we don't think we
18:33 don't think we're anywhere close to being done we think we're gonna you know
18:38 our goal is to really help every single major industry in the world from
18:42 healthcare to transportation to logistics to financial services to
18:46 insurance you know every single huge industry uh and empower them with
18:51 War in Ukraine
18:51 transformational ai use cases i mean it's it's incredible it seems
18:56 like the opportunity is endless but i do want to highlight one thing that we're
19:00 incredibly proud of and i know you and the company and your whole team is incredibly proud
19:05 of which is the role that actually your technology has played in saving lives in
19:09 the war in ukraine and i was wondering if you could share a little bit about
19:13 how scale has played such an important role in that conflict yeah no and this
19:18 is something again that um i'm uh i'm extremely proud of the team
19:21 as part of our technology as well as like i think i think it's incredibly
19:26 important and so you know in the in the um russia ukraine conflict um
19:31 intelligence is is one really really important element of that you know being able to
19:35 know what's happening and be able to make good decision making as a part of
19:39 that and um one really critical component of intelligence is is satellite data and uh
19:47 so there already is there are you know i think around 2000 satellites um circling
19:51 the earth there's constant collection of satellite data and satellite imagery
19:56 from the rush ukraine region but one thing that's actually really big
19:59 bald link is being able to analyze this satellite data really effectively and
20:03 it's actually a problem that is you know very applicable for ai to be
20:07 able to automatically look at satellite imagery and sort of tell you what's
20:11 going on and so what one thing that we did in addition to um
20:16 in in close partnership with sort of government end users is we built uh we
20:21 basically looked at all the satellite data in the rush ukraine region and are
20:25 continuing to do so and analyzed it for a number of key things one is damage
20:32 assessment so in key areas like kiev kharkiv nepro what is the was the damage
20:37 from the start of the conflict to now to basically enable uh
20:41 a strong humanitarian response and and enable a proper allocation of resources
20:46 and and people to be able to help you know provide aid and provide assistance
20:51 um as well as analyzing you know changes and movements in you know uh tanks or
20:56 planes or other sort of military equipment and these these things you know despite the fact
21:02 that so they're great problems for ai to solve and they can make a huge
21:05 difference if you can react really really quickly um in the moment in
21:10 theater and uh and so we've analyzed over 2000 square kilometers we're
21:13 constantly analyzing it even now and it's something that that
21:16 we're really proud of we think it's having an impact you know our government
21:19 end users are telling us having impact and um we're really passionate about it
21:23 so uh incredible wow that is thank you that's truly um outstanding and and i
21:30 Selling into government
21:35 i'd love to dig a little bit because you said it's something that you're
21:38 passionate about about being an american of just the journey of actually selling
21:42 into government so i can't imagine if it was hard to sell into you know a
21:46 self-driving car or a group of scientists how difficult it must have
21:50 been to then sell into a government and a federal regulated body tell us a
21:54 little bit about the behind the scenes how did you get a first meeting did you
21:57 have to get security clearance how long did it take to sign something with a you
22:02 know federal entity yeah so so um one thing i will say is is
22:08 uh the federal the members of you know the us dod as well as all members of the
22:13 us federal government they first off know just how important ai
22:18 technology is to the future of america and they i think they they all recognize
22:23 that a lot of the processes that exist today and a lot of the sort of
22:26 procurement processes or or a lot of the ways in which the government works with
22:31 the private sector need to improve and so um that's a that's a constant topic
22:35 and i think that like for anyone who's passionate about i encourage you all to
22:38 sort of outreach to your representatives and sort of and talk a lot about this
22:42 issue so they they definitely know things need to change but um it's it's a
22:45 really difficult process and this is one of the reasons why i think it's so rare
22:50 for startups to work with the the us government and work with you know the department of defense
22:55 or intelligence agencies because it's just it's it's you know it's a tough
23:03 divide and um it's it's uh it's not something that has historically improved
23:05 america if you think about a lot of the great technologies today a lot of them
23:08 were built out of partnership between silicon valley and the us government historically but that
23:15 you know these sort of two um communities have drifted solely apart
23:19 and so that you know early on to get meetings we would just we would have to constantly be
23:24 networking and we'd figure out like third and fourth and fifth degree
23:27 connections to be able to get to the right person um i've i have security
23:32 clearance you know and i've had to you know we've had to like figure out how do
23:36 we how do we bridge this gap we really want to work on these government
23:39 problems like i i went on a trip to china actually with a fellow index uh index
23:46 partner a few years ago and we visited um some of the the chinese ai companies
23:51 and you know i just realized it was going to be really really critical in
23:55 the united states for us to have the best in class ai companies work
23:58 closely with the us government otherwise you know we were gonna we're gonna fall behind
24:03 and so that really spurred you know our inspiration uh in in
24:07 working with with the us dod and i think that um at the end of the day it was really
24:13 you know our strength of mission and our strength of sort of conviction that like
24:16 you know the united states needs great ai technology for defense intelligence
24:20 that allowed us to ultimately bridge that gap and um and uh and help them and
24:24 it was sort of you know probably all in it was and this is extremely fast for
24:29 for government work but it was probably you know a year and a half maybe from
24:32 like first meeting to be able to get our first large production contract and that
24:36 is you know one of our investors is founders fund who um was a is a big
24:41 investor in palunder and andrew and a number of iconic uh companies and they
24:45 said that was the fastest they'd ever seen to get to a meaningful production
24:50 contract that's amazing wow that's right [Applause] so um one thing that i've heard you say
24:53 AI needs humans
24:58 in the past is that you believe that ai needs humans which is something that a
25:01 lot of people in silicon valley don't say they say that you know ai is going
25:06 to replace humans so just curious why do you have that belief and
25:09 how does it really organize the mission of scale ai yeah so so
25:15 we i think that like fundamentally if you look at the the core
25:20 uh building blocks of air the core setup of ai um ai needs a lot of data and then
25:25 data requires lots of humans to be able to to help produce and and ensure that
25:29 that data is correct and accurate and even in you know one of the things even
25:33 if you look at cases where um uh people say there's no humans involved it's
25:36 things where like you know they'll train off of all of the reddit data where's
25:40 where's reddit come from it comes from people typing on their computers and
25:45 commenting in their free time and so um fundamentally all the all the data used
25:50 to to train ai systems is going to come from humans in some some way uh shape or
25:56 fashion and um and this is something that like not only i think we at scale understand you know
26:02 the the most innovative companies in the world the ones who uh are able to do the most things
26:07 automatically frankly also really deeply believe this we work really closely with
26:12 meta and microsoft for example two of the again the most innovative companies
26:15 in the world and this is a fact that they embrace and they actually realize
26:19 that like hey making sure that humans are really deeply involved in the
26:24 process of building ai systems both today and long into the future is going
26:27 to be how we actually ensure that these ai systems are performing um in line
26:32 with human values and in line with human intentions and we're not going to have
26:36 these sort of weird doomsday scenarios where nai is doing something that humans
26:40 wouldn't expect or humans wouldn't be aligned with and so um we really we've always had the belief
26:46 that you know humans and ai are going to need to work together humans and the
26:48 algorithms are going to need to work together to build entirely new consumer
26:52 experiences entirely new business processes and ultimately build sort of a
26:57 much more powerful and and sort of innovative economy and that's been a
27:00 core belief at scale i think we're actually one of the few ai company ai
27:04 startups that really embraces this fact and um and i think it's something that
27:08 uh that we really passionately believe and we don't think is ever gonna go away
27:12 that's amazing and maybe just as a final question um you've described this the
27:17 mission of scale ai as being the world's ai partner tell us a little bit about
27:21 what you see um this space turning into in the next five to ten years
27:26 yeah you know so i think if you at the core if you look at the the massive waves of
27:33 technology you know take it you zoom out a lot there was this massive wave of
27:36 originally the the sort of like mainframe computer being used a lot of
27:39 business applications sort of ibm was the first wave um then there's sort of
27:42 personal computing as a second wave that sort of apple was one of the companies
27:45 and microsoft and all these companies were involved in that in that wave then
27:50 you had the internet really really spring into to life then you had mobile
27:57 and then software sort of like software in general has been this and software as
28:00 a service has been this massive wave of sort of productivity throughout the
28:03 throughout the world over the past you know two decades um these are these are
28:06 sort of like the massive trends of technology and i think as a as a world
28:12 we're kind of at this point where um each of these giant s curves are
28:15 starting to slow down i think that like you know the gains of software eating
28:21 the world as as mark andreessen put it are are beginning to slow the tailwinds
28:24 of software in the world are beginning to slow and and the next wave of value i
28:29 we really deeply believe is going to be artificial intelligence and the the
28:33 potential for ai to transform every major industry again whether it's agriculture or
28:39 defense or health care or financial services or life sciences you know it's just
28:44 absolutely astronomical but that's only going to happen if we're
28:48 able to build the infrastructure to support that so our view is that
28:53 you know frankly we believe that if you look a decade from now
28:56 a lot of the value and a lot of the innovation in technology is going to be
29:00 from artificial intelligence and ai entering every major industry and our
29:04 goal is to be a partner to as many of those industries as possible and
29:07 enabling that well alex thank you so much for sharing your extraordinary journey and vision
29:13 with all of us today congratulations on an incredible ride so far we continue to
29:17 be so excited about what you're doing and the massive markets and industries
29:22 and even humanitarian impact that you guys are doing are having so thank you
29:26 for sharing your thoughts and perspective with us today thank you so
29:30 much nina always a pleasure thank you thank you everyone