00:10it's always really exciting to be at
00:13Stanford yeah I think you guys have it
00:15here and I've visited a few campuses
00:18around the world and every time I come
00:19back it's really awesome to be here so I
00:23just want to give you a little bit intro
00:25but before I do that I actually want to
00:27take a little bit of a poll so I can
00:30kind of have an idea on who you guys are
00:32and what you guys want to be so I'm an
00:35engineer and most review maybe aren't
00:36engineering and maybe you can you can
00:40take a look at this question and say
00:42what what do you want to do when you
00:45so just as a show of hand so the options
00:47are be a founder of a company stay in
00:49academia have a people management role
00:52in industry have some technical role in
00:54industry and no idea
00:57so just number one who wants to be a
01:00founder all right actually Wow
01:03he's half the class who wants to stay in
01:06academia okay well one or two who wants
01:11to have a people management role all
01:13right two three or four who wants to be
01:16have a technical role in industry two or
01:18three and no idea oh wow almost half the
01:22class all right great
01:24so that's actually a pretty pretty great
01:26mix so I actually was probably throw my
01:31career probably it ranged all five like
01:34when I started I was probably five when
01:36I was like undergrad and then it went to
01:39four and then it went to two and then it
01:43went to three and then one and then
01:45probably I met one now and probably
01:47wanna go back two maybe three eventually
01:49or maybe two so it'll change but I can I
01:53can go into it so just a few of the
01:56highlights I just want to talk about
01:57just my background where I came from
01:59talk about artists our company and then
02:02lessons learned along the way so about
02:08originally from Vancouver Canada and I
02:12got my undergrad degree in mechanical
02:14engineering at the University British
02:16Columbia I I was always interested in
02:18cars I was part of the the SAE kind of
02:25that we did the Formula one cars at the
02:27time or and and was always interesting
02:29actually from there I actually worked in
02:32industry and being from Canada I always
02:36thought that was the that was the best
02:38thing in the world you know working for
02:40these large companies I worked in
02:41Germany I worked in Canada and and being
02:44there as being part of engineering teams
02:45I realized that's not what I wanted to
02:47do it was it was political is
02:50bureaucratic who was slow there were
02:52processes and I realized if I were gonna
02:54move up anywhere in my career I had to
02:57do something different and at the time I
02:59decided hey I'm gonna apply it a few
03:01schools in the States and try to get a
03:03try to get a good spot and going to grad
03:06school and so I did and I got into the
03:11Aero Astro program I didn't have funding
03:14I when I arrived I had a and I also
03:19didn't have a plan of how I was gonna
03:21pay for school so when I arrived I I
03:23knocked on doors I literally knocked on
03:25doors to try to get ta za raised trying
03:28to find funding and I did I found a
03:32great professor and we came up with a
03:34business plan of how we're gonna like
03:35fund my research and fund what I was
03:37gonna do which was in Formula one cars
03:39so he was an Emmy because I tried
03:41knocked in all the Aero Astro doors and
03:43they were all closed so Emmy was a great
03:46opportunity and that's how I got started
03:49so I actually worked in got a funding
03:56through that Toyota Formula One team and
03:59we did a lot of research on aerodynamics
04:01of the front wheel assembly of the car
04:04and that took about few a few years I
04:07hadn't intended to do my PhD but it just
04:10kind of happened that way you know there
04:13were a lot of master students that that
04:14were there at the time that said well I
04:16probably won't do a PhD but that
04:18eventually changed and I kind of kinda
04:19got sucked into it and said okay well
04:21I'll stay a few more years ended up
04:23being four or five years but throughout
04:26that process I kind of got sucked into
04:28this whole entrepreneurship thing I mean
04:30there was there were programs or
04:32opportunities and and that's kind of how
04:34I mean I kind of how it all got started
04:36I actually took a took a was called pre
04:40at the time which was a program in
04:43ship which has changed it but and now
04:45there's all kinds of programs and then
04:47in 2011 I started arturis and that's
04:51that's the history so but before I get
04:55into what arturis is just a little
04:57background on how we got started I I was
05:00working in my little lab at the corner
05:03of campus really close to the circle of
05:05death and and somebody literally knocked
05:09on my door and through a mutual friend
05:12she was trying to look for somebody in
05:14physics who could simulate airflow
05:18through his upper airway because he had
05:20seen a doctor that was gonna do surgery
05:23on him and he thought that if he was
05:25going to go through surgery he didn't
05:26want to do that and maybe he thought
05:28there's a figure a better way to predict
05:30how that surgery would end up if if only
05:33that you could do some sort of
05:34simulation or fluid simulation so and I
05:37was the CFD expert at the time and
05:39that's how this idea to use simulations
05:42or just or to use kind of cloud
05:43computation kind of got started so so we
05:47worked on this idea and submitted it to
05:49this pre program and and that's kind of
05:52where it all got started and originally
05:55arturis was not about the this problem
05:59of trying to solve upper airway disease
06:00using CFD we actually tried that for
06:03several year or for a few months and and
06:07that was challenging because going
06:10through the whole fundraising process to
06:12trying to raise capital trying to find
06:15out how we're gonna sell it in
06:16healthcare it's challenging so we
06:18actually hit a lot of roadblocks and so
06:20in that year basically in 2012 we we
06:23learned a lot about the the industry in
06:25the business and the way we learned a
06:27lot is by going out and talking to
06:29doctors talking to a lot of people who
06:31knew the field I knew how this
06:33technology could be used and throughout
06:35that process we met two other physicians
06:37at the time we're at Stanford
06:39Albert Hsiao and tres vasana Walla who
06:42are other co-founders and that's really
06:44where it all got started and they were
06:46working on a technology at the time
06:47which was called 4d flow which is a way
06:49to naana basically measure blood flow
06:51using MRI and throughout that
06:54partnership the four of us decided to
06:58and that's how we started really from
07:00sleep apnea and really moving towards
07:02cardiovascular mr and now we do
07:04something even differently so with that
07:07what is arturis so the artists are our
07:10mission is to really could create an
07:12automated intelligent imaging platform
07:14to unleash the real world data to make
07:18healthcare accurate and data-driven so
07:19that's what we're all working towards
07:21and it's pretty broad and what really
07:25our focus has been in the last few years
07:27is really we've developed a zero
07:30footprint medical imaging analytics
07:31platform with deep learning that has FDA
07:34clearance that's what were that were the
07:35first company to do this in the world
07:37and and what is medical imaging
07:39analytics so what is a medical imaging
07:41platform mean well what we're doing is
07:43we're collecting data from from MRI
07:46scanners from from any scanner really in
07:48and we're pushing all that data to the
07:50cloud and we're allowing physicians
07:52senator users to log in and process this
07:55data in in real time and process this
07:57data further a clinical workflow so
07:59we're solving real clinical problems for
08:02them and and that's where the idea of
08:04machine learning or deep learning comes
08:05in you can you can get very intelligent
08:08and smart as to how you can automate a
08:10lot of their workflow by using by using
08:12these types of technologies just in
08:15terms of the the size of the opportunity
08:17or market it's the just the medical
08:18imaging market alone is a nice for
08:21post-processing is about 3.2 billion
08:22dollars so right now it's very local
08:25you have very legacy software in in
08:28these types of hospitals and it's it's
08:31pretty archaic I mean even the
08:32algorithms or everything that's that's
08:34that's out there so it's very far behind
08:36the Facebook's or the Amazons or the
08:39apples of the world and there's a reason
08:40for that it's primarily in regulation
08:42primarily in in you know the impact that
08:46that a potential software vulnerability
08:48can can have on a patient so there's a
08:50lot of roadblocks in healthcare or
08:52getting getting software out also
08:55there's other issues with regards to you
08:57know working with hospitals selling
08:59infrastructure to hospitals selling
09:00software to hospitals the the time it
09:03takes to get money or the sales cycles
09:07are very different from conventional
09:11so in terms of kind of in imaging today
09:14and and and what's the current status of
09:17the market it's it's actually moving
09:19from right now a paper use so physicians
09:22in the past have gotten paid for every
09:24procedure and it's moving to a
09:26value-based care model so what does that
09:28mean really physicians who are getting
09:30compensated for outcomes compensated for
09:32how well a patient you know gets gets
09:36better but and not really about I'm
09:38gonna make more money by doing this
09:40procedure so and that's that's that
09:42incentivizes everyone to in the right
09:43direction the field does suffer in
09:46general from lack of consistency it's
09:48difficult to track patients across time
09:50across different institutions it's
09:52difficult to aggregate data and the
09:55intelligence of a single of the work
09:58that's going to be done on a particular
09:59patient is really limited to the to the
10:02user or to the physician that and
10:04whether they've seen that a certain
10:06issue or not and that data is very
10:08rarely shared across institutions across
10:10the world so there's a lot of collective
10:12intelligence that you could bring by
10:13having a centralized system and that's
10:15really what what we're doing there's
10:18also a need to make imaging more
10:20quantitative so think of the idea of
10:22trying to quantity quantify certain
10:24structures and images or and extract
10:26additional parameters from from these
10:28MRI machines you can extract so much
10:30more than just black and white Anatomy
10:32images there's a lot there for example
10:34blood flow and lastly you can combine
10:36different types of data for for
10:38biomarkers the ultimate goal really is
10:40to do predictive not predictive analysis
10:42to identify what are the what are the
10:46features that cause a certain disease so
10:50there's a big opportunity here for for
10:52change in the conventional kind of
10:55imaging market it's again local the
10:58interfaces are very old-school it like
11:02measurements and reports lack
11:03consistency or structure everything is
11:06is either oral it like dictated or you
11:09have these handwritten reports or it's
11:12all very scattered it's difficult to
11:14make any sense of this data and and
11:17again automation has been difficult and
11:19primarily because that they haven't had
11:24data to try to automate these these
11:25things how we're changing that game is
11:29really using the cloud and aggregating
11:32that data on that in that single place
11:33to be able to provide insights so we're
11:36providing very structured reports and
11:38quantified reports to make it consistent
11:40I think consistency in healthcare is
11:42very important so that you can have you
11:45know you can you can track my new
11:47changes over time and make sure that
11:49they're not biased by the reader that's
11:50that's very important and and what do we
11:54do or what are our main tasks so they're
11:56primarily around detection detection of
11:58what's in the images of tumors or other
12:00other types of things
12:01classification tracking segmentation
12:05these are fairly you know these are the
12:07building blocks of what of what our
12:09back-end does in terms of as I mentioned
12:14this local approach today is really CPU
12:17based not a lot of you know advanced
12:20rendering or quantification segmentation
12:23where we're moving to is this idea of
12:25pushing the data out to the cloud very
12:28similar to the way Tesla's doing with
12:29with their cars if you push the data out
12:30you can get so much smarter and and our
12:34back-end is it's fundamentally GPU based
12:37we usually pews everywhere in our stack
12:39from rendering quantification and
12:41segmentation the machine learning it's
12:43it's all has to be real-time and fast so
12:49just a few of the core technologies but
12:50behind our our system obviously the
12:52cloud computation the deep learning it's
12:55a zero footprint browser-based interface
12:58we have a unique piece of core
13:00technology that actually keeps all
13:02patient information in the hospital
13:03something that that that nobody has done
13:06successfully using the cloud before and
13:08of course we have the FDA and CE mark
13:11and several regulatory clearances to
13:13sell our product into those markets and
13:17as I mentioned what we're kind of
13:18changing the game in terms of how we're
13:20using this data to our advantage
13:22hospitals are just now really realizing
13:24the value of that data that's sitting
13:26there and and we want to leverage that
13:28data and we want to make it accessible
13:29not only for us but also for everyone
13:31else out there and and the way we do
13:34this is is again it's a collective
13:37we can we can take that data and get
13:40smarter over time so you could think of
13:42it as as people use use our platform we
13:44learn we learned what they're doing well
13:47we learned the mistakes we're making and
13:49then we're just constantly getting
13:50better releasing updates and and finally
13:54yeah every physician benefits from all
13:55the thousands of other experts in the
13:57field so it's really a great great use
14:02of making qual quality analysis or
14:05quality care and it's in this race
14:11especially it's important to have access
14:13to data and really the company that has
14:17the fastest can really accelerate and
14:19become the most intelligent and the way
14:21I think of this is let's say test on
14:23Google Tesla has as 200 million miles
14:25driven versus Google has an order of
14:28magnitude less and and they're they're
14:30ahead and they can stay ahead because
14:32they have that at the collective
14:33intelligence and and really it's the
14:34same thing in our industry as well with
14:38that I want to give you in a sense or a
14:41demo of our system and and how it works
14:43I think it just helps to put to put a
14:48picture to it to words so so this is
14:52this is our system we we again have a
14:55way to access images anywhere in the
14:57world where we have clearances and in
15:00Europe and United States but really
15:02we're deploying our solution across many
15:04different other countries so a physician
15:06logs in clicks on a particular study
15:08that they're interested in and then then
15:10they can look at this study anywhere
15:12anywhere in the world so I'm just going
15:16to reset my view here so what you're
15:18looking at is a 3d image of of the body
15:22so what I can do is just just natively
15:26in the browser I can look at about five
15:28gigabytes of data and the way we do this
15:30is we distribute this load across across
15:33it basically a GPU cluster of resources
15:37and we spin up and down dependent
15:39depending on the the load of the of the
15:43of the system so it's it's just 3d and I
15:47can just hit spacebar here and it's
15:49actually 40 in time as well so you see
15:51heart beating and there's extra
15:54dimensions as well so what i can do is i
15:56can turn on blood flow so well i can
15:59really in one simple 10 minutes can
16:01really see everything that's happening
16:03in your body and this is a this is a
16:06brand new piece of technology that's
16:08we've partnered with General Electric GE
16:10to to commercialize and to deploy
16:14eventually across all them and our
16:16machines in the world that's that's our
16:19really first application and it's not
16:21only visual it's also looking at you
16:24know you can quantify several things so
16:26here this comes back to my CFD or
16:30compeition fluid dynamic today's is
16:32looking at streamlines so looking at
16:35where the where the flow is going red
16:37here is showing areas of high high flow
16:40or high speed and and blue is is low
16:42speed I can also turn on vectors so
16:47vectors are it can tell me the direction
16:50of flow as well forwards or backwards
16:52its quantitative I can measure blood
16:55flow through any turning artery or vein
16:57so in this case I've made a region of
17:02interest and I can measure measure flow
17:05in this case it's 1 or 1.7 liters
17:08forward or net flow I could I can make
17:12new measurements as well so let's say I
17:14want to measure the flow and the the
17:17pulmonary artery the single click it
17:19attracts that pulmonary artery it finds
17:21it and it makes it makes new
17:23measurements so in this case one point
17:25six eight liters per minute and so these
17:27are the types of tools that radiologists
17:29want to use they want to make imaging
17:31quantitative and consistent and this is
17:33just one way to do that and it's all
17:36about being able to identify to track
17:38and classify what type of pathology is
17:42this in in oncology to trying to assess
17:45what type of tumor it is the looking at
17:47the texture there's a lot of parametric
17:49data out there that you can jack you can
17:51aggregate in terms so this is this is
17:56the 40 flow product we we also have the
17:58product of looking at other types of
18:02data so I can look at brain brain
18:05so this is a an example of lesion
18:10detection so this is a this is all for
18:12research do so only at the moment it's
18:13not on the market and what I can do is I
18:16can go to what you're looking at is
18:20several different types of sequences
18:23they're called so different types of
18:24scans and you can combine the different
18:26scans together - - - - - to get
18:30different types of information so if I
18:32go to lesions here I click on find
18:34lesions what about the system does is it
18:38it identifies where these solutions are
18:40and it identifies the enhancing region
18:43of the the lesion the core of the lesion
18:46as well as the the whole the whole tumor
18:49or and and it quantifies it so it does
18:52really football you metrics segmentation
18:53and you'll notice that I didn't have to
18:55do anything really the back end of all
18:58this is machine learning and deep
18:59learning and we have a production
19:01inference service where the data gets
19:02sent up it's in real time it's it's it's
19:06it's it's it's tracked segmented in it
19:08and pushed back to the browser and in
19:11seconds and from there obviously they
19:15have simple tools to be able to to
19:17annotate or to track or to change the
19:19rendering there's there's all kinds of
19:21specific rated weight radiology tools
19:23that you can use to to report on the
19:26patient ultimately a few other examples
19:29looking at chest CT or natl detection so
19:35what I can do here is I can just
19:38maximize this particular patient here so
19:42you're probably familiar with just
19:44looking at kind of chest x-rays so CT is
19:47just a kind of a 3d chest x-ray so what
19:50I can do is you know I'm here I'm just
19:54rendering the full chest cavity and
19:57depending on that on the way of how you
20:00how you render this data you can you can
20:03actually try to identify tumors or
20:06nodules so here I click on find nodules
20:09and it'll automatically track and
20:11identify where these nodules are
20:13something that's very time consuming
20:14today using conventional radiology so
20:19looking at thousands of images and it's
20:21time-consuming and they can miss them so
20:23this is a perfect tool for for machine
20:25learning so with that I have all kinds
20:30of demos but I'll stop there and if
20:33there's time at the end we can go
20:34through some some other cool stuff but I
20:36just wanted to give you an idea of of
20:38practically what what we do and how it
20:40impacts the the clinical care so so so
20:46with that our first our first kliklak a
20:48clinical application is cardiac mr.and
20:51and why is that important
20:54it's it's it has enormous value for kind
20:56of the pain points specifically in
20:58trying to diagnose structural heart
20:59disease so what's happening inside
21:01someone's heart whether their valves are
21:03leaking whether they have a hole in
21:05their hearts whether they have a single
21:07ventricle there's there's really
21:08structural issues that a lot of people
21:10have and it's difficult to diagnose
21:12these using conventional approaches like
21:14echo or echo echocardiography you have
21:17to be really an expert to try to
21:18identify what's what's in these images
21:20cardigan MRIs and specifically this
21:22technique is different in that you just
21:24need to acquire a box and a technician
21:26can just acquire that box very simply so
21:28there's advantages there it also
21:31provides an additional metric of the
21:33flow data something that in imaging
21:35today just is not there it's a in terms
21:39of kind of what are what are kind of the
21:43whatever value add here is that we're
21:45providing something that's a 10x
21:46difference and it's important to have
21:48really a very important 10x product I
21:52call it because it's difficult to just
21:54sell something that is incremental in
21:56hospitals there's a big burden to
21:58actually get something sold into a
21:59hospital and it's that's why you need to
22:01have kind of your Trojan horse to get
22:03into the door provides something really
22:05of a clinical unmet need and then expand
22:08from there and as I mentioned so this
22:12this is our this is our system here so
22:15that the conventional way of doing this
22:16is very manual it takes it takes on the
22:18order of half an hour to an hour to try
22:20to extract for example volumes of your
22:23ventricles and we can do that in seconds
22:26now so it's really that's why it's a
22:29paradigm shift compared to their manual
22:32using very very simple tools to just
22:35click a click a few buttons and get get
22:37all your answer very quick very quickly
22:41these are a couple examples of of how we
22:46use the deep learning and how he's
22:48probability so on the right you see the
22:49probability maps there for the long
22:51again on the bottom you see the
22:53probability of where we think the
22:55nodules are and we take that information
22:58then extract contours out of that so
23:01that these contours are editable and
23:03physicians can edit those contours and
23:05those edits get fed back into the
23:07training data and that's how we get
23:08smarter and learned over time as I
23:13mentioned we are partnering with with GE
23:17healthcare and it's that that's a great
23:19partnership because we can develop the
23:21product and they have the sales force
23:23and and distribution to be able to get
23:24it out onto the market something that's
23:26very expensive for us to to just to
23:30build out ourselves so if that's why
23:31it's important to have both direct sales
23:33channels but also indirect sales
23:34channels and and leverage the you know
23:37the the larger companies out there so
23:40it's it's a very very important
23:42relationship that we have and also that
23:44they they're great at marketing and
23:46great at by pushing the product out and
23:47it's it's a great partnership that way
23:50so we we showed it that this big
23:52radiology conference at the end of last
23:54year and it was really was that one of
23:58the highlights of the whole conference
23:59of those 60,000 people there and artists
24:03radiology software and best new
24:05radiology company so we were very proud
24:08about that we really have a great team
24:11it's just a little kind of history about
24:13the company what's very important for us
24:16in terms of milestones is getting FDA
24:18clearance and getting regulatory
24:20clearance to sell our product that's
24:21really out the way our barometer power
24:24how we're measured we we incorporated
24:27our waste we really raised our first
24:29round in 2013 we just had a few people
24:32initially and then for a few years we
24:34developed the product and then we raised
24:37our second or second round or a round in
24:402016 we're grew to about 15 people and
24:46around 40 people and didn't raise our
24:48next round we we have a very rich
24:54pipeline of new products in apps coming
24:56out and we were developing both the
24:58platform and a new applications so
25:00that's in a nutshell just in terms of
25:04team as I mentioned others we have four
25:06co four co-founders and we have the
25:10different groups for regulatory quality
25:14machine learning software products so we
25:17have a you have did a lot of different
25:19teams and sales and and we work very
25:23very effectively with with one another
25:26in terms of status we've we're we are
25:29active in about 40 40 research sites and
25:32have helped over 7,000 patients so it's
25:36and what's great is because we can count
25:39we know exactly what how many patients
25:41are coming in and and how people are
25:43using our software something that's very
25:44unique in this industry yeah
25:48so with that I'll I'll go into some of
25:51the lessons I learned along the way I
25:52think when I started I had no idea what
25:56I was gonna but about to start and I can
25:59I can kind of I can imagine myself in
26:02your shoes today and thinking what what
26:05do I want to do and hopefully some of
26:07these lessons will will he'll help and
26:10it's just a random mix of lessons but it
26:13just it just helps helps us try to think
26:15about that so so the first thing I'll
26:17say is find the right partners you trust
26:20I think doing things by yourself yes you
26:25can do it the companies have done it
26:27before but it just helps having a
26:28partner having it having a co-founder it
26:32just you can bounce ideas off somebody
26:34and and it's just it just relieves some
26:36of the stress and some of the
26:37responsibilities and and it's just
26:41extremely helpful even my my co-founder
26:42says the same thing he started it by
26:44himself he start his first company by
26:46himself and he said it's just so much
26:48easier just having having a partner the
26:51issue with that though is you have to
26:53think about whether you trust this
26:59and whether they're committed to this
27:02also in terms of like how are we gonna
27:04hang our we gonna spend on this idea
27:06because a lot of you if you have an idea
27:08you may not be working on it full time
27:10you're maybe you're still in school and
27:11you may feel resentful if somebody's
27:14working harder than others and what's
27:16interesting is that what you'll do is
27:18you may try to get fancy with how you
27:20split split the equity in terms of how
27:23how involved someone is to start off and
27:26I will say keep it to keep it simple I
27:29think just because somebody is working
27:32full-time for a week on a particular
27:36project doesn't mean that they should
27:38get more revenue or want more equity
27:40split because it hopefully you're
27:41building a great business and a week
27:43isn't gonna make a difference and just I
27:47would say structure in a way that makes
27:49it successful long term and and everyone
27:52is incentivized it's not easy no it's
27:55not easy to do but it's it's it's really
27:57important with that leverage the
28:01incubators that are around here if you
28:03have an idea there's now so many
28:07incubators so just to give you an idea
28:08the artist has been through now for and
28:10growing so we started at startx
28:12in 2011 which is a fantastic program it
28:16gives you access and accelerates your
28:19business so fast especially initially
28:23you just don't have the funding to to
28:25just incorporate or to you know think
28:29about payroll think about lawyers legal
28:32contracts infrastructure servers
28:35hardware there's so much just initial
28:38investments that maybe you need just to
28:40provide a proof of concept before you
28:41get friends and family around so
28:44startups can definitely help you there
28:46and if you don't get into startx there's
28:48other incubators all the way around here
28:49we after startx went to the hattery
28:52which is a mini beta up in the city and
28:55we then move to UCSF we were part of the
28:59qb3 incubator and we now just started at
29:02TM CX which is an incubator at the
29:04Houston lead at the medical school there
29:08so it's again leverage leverage all the
29:12kind of the resources that you have
29:13around you super valuable
29:16the next is bored so you probably won't
29:19think about this early on but as you
29:22raise your you're you're kind of first
29:25round if you can do it with friends and
29:28family your own personal money and
29:29that's not an issue great but if you do
29:32get investment from from a VC or some
29:36other investment vehicle think about
29:38your investors I think it's really
29:40important to to to to trust them and
29:43make sure that you have control and and
29:45you're all aligned there's been stories
29:48endless stories about you know boards
29:51not agreeing with the founders and and
29:53it's really important when you talk to
29:54investors make sure that there's a
29:56mutual fit because that is I can't I
30:00can't say that how important that is and
30:02for us we've been extremely lucky we
30:04have a we have a great board and we're
30:05definitely all all aligned lastly is is
30:10the network I call it the Hoover tower
30:12syndrome the idea is everyone has sticks
30:15around here and everyone helps them
30:17helps in a different way and it's and
30:19it's extremely it's a useful tool that
30:22you should leverage the everyone here is
30:24involved in some forms with Stanford
30:26through a through a contact or your
30:29university or just through VCS or
30:32there's there's so many people that know
30:34something here so they grow your network
30:36and and and leverage the resources that
30:38you have is specifically around Stanford
30:40and Stanford is is very unique I can
30:42tell you we probably could not have done
30:44what we've done so far if we were up in
30:46Canada or if we were up in Europe it's
30:49such an advantage to be here in Silicon
30:51Valley yeah you have no idea we have
30:53very similar companies that are trying
30:54to do the same thing in Quebec and it's
30:57just they're just not getting traction
30:58and when they when they try to pitch to
31:01VCS here it's just not the same so again
31:04having headquarters having founders
31:05local is really really helpful and not
31:10it's not to say that then you can expand
31:12and grow your business in those other
31:13countries so we are headquartered here
31:15in San Francisco but we have offices in
31:18Canada and in France as well and you can
31:20expand the next lesson I've learned is
31:25the product market fit
31:26and you've probably heard of this and if
31:28you haven't the kind of the first
31:30Business School class will talk about
31:32this and specifically for us in
31:35healthcare it's really about fitting
31:37into a clinical unmet need so you really
31:39have to identify a really a pain point
31:42or an issue that's people are willing to
31:43solve and people really need that's
31:47really important I can't emphasize that
31:49enough what I'll say before you have a
31:52great technology a great idea just spend
31:55a little bit of time doing the market
31:57analysis to see if it's if it's a
32:01worthwhile market because I can tell you
32:03there's so much technology out there but
32:05if it's really not about if it's not a
32:07big enough market or it's not clear how
32:09you're gonna sell it or make money
32:10that's it's just not gonna work out so
32:13that is very important I would say
32:17before you start read a lot of
32:18post-mortems and if you go to hacker
32:20news or just Google postmortem
32:23I start up there are so many stories of
32:25other companies that maybe are in your
32:28space that have failed and you can just
32:30learn so much about what they want what
32:32they've done wrong and just learn from
32:34that and then and then move forward I'd
32:37also say with that is don't be a what's
32:42called a Walking Dead there's actually a
32:43lot of companies that are just
32:45scratching or just hanging on by one
32:48last thread and maybe just have an have
32:51been doing that for the last 5-10 years
32:53and and that's not something that you
32:55want to do and so it's really important
32:57to know when to pivot especially early
33:01in the in the process don't be afraid to
33:04pivot if it's not working out and and
33:07and that's what we have to do we really
33:08started in sleep apnea which it was a
33:10tough market for that particular market
33:13you just didn't do CT scan on a regular
33:17basis so that helps tremendously it's
33:19just you know having the courage to say
33:23hey it's not working let's just find it
33:25find a new route and lastly it's big
33:28market if you're pitching to investors
33:30and the market size doesn't have a
33:32billion in it it's challenging to say
33:36now that's not to say that you it'll
33:40but investors nowadays are looking to
33:41invest in overs the googles of the world
33:43and they have limited time and bandwidth
33:45to invest in smaller markets so you
33:48really have to have an engaging business
33:51opportunity for them and an upside for
33:53them so market size is obviously very
33:56very important the the other lesson I
34:01learned is around revenue and what I'll
34:04say it's it's fairly simple to try to
34:07get to a million bucks in revenue but
34:08passing the 10 million 10 million dollar
34:10mark is is challenging and the reason
34:13why I say it's challenging it's because
34:15of this idea of crossing the chasm this
34:19is another picture that I'd that's
34:21that's that's very common is just that
34:22you have this idea of of the innovators
34:24the innovators who are there people that
34:27will buy your product just because it's
34:28cool and amazing then you have in the
34:31next category the early adopters and
34:33whether you have this big Delta between
34:35the early adopters and the early
34:37majority of people that will buy your
34:39product and so crossing that chasm is
34:41very critical in terms of getting to
34:44stay in growth and it's especially in
34:47health care the health care is driven by
34:49regulations and guidelines and and laws
34:53and it's it's it's challenging to get to
34:56the to the tale of this graph and the
34:59late the late majority of the laggards
35:01is really half of their population it
35:03takes time to in order to get to to get
35:05there and that's why you know getting
35:08past this revenue mark is is challenging
35:10I also mentioned follow the money that's
35:14why it's really important to understand
35:15you may have a great product market fit
35:17but the idea is if you don't know where
35:19that money is coming from or how to tap
35:21into that it will not work and
35:24especially in healthcare it's very
35:25convoluted and complex you have an
35:27insurance you have brokers you have the
35:29hospital's payers and it's just you need
35:32to tap into that system and again it's
35:34it's it's challenging to understand how
35:36you how you can see how you can take a
35:38little skim off the top and everybody's
35:40trying to do that especially with the
35:43changing and regulations and now and
35:45administration things things are things
35:48that are in flux for sure
35:52the next is successes in the people not
35:55the technology so again you you kind of
35:58have to have a great technology to start
36:00and if you don't then you're probably
36:01not to get funding but it just helps
36:05having you being surrounded by great
36:06people and and I can't emphasize that
36:09that enough is just this is a picture
36:11that we took during our during our
36:14retreat and this is about maybe half of
36:17our team here but it's really important
36:19this too that you trust your team around
36:21you Matt and then you bring in a a
36:24quality people I can't can't have as
36:26emphasize that or company culture is
36:31extremely important and I call it a no
36:34ass rule basically one but one bad
36:37person can ruin your company culture and
36:41I want to emphasize that ego is it's
36:46really the the root of all evil if you
36:49have one person has a big ego and you
36:52have the rest of the company that are
36:53really incentivized or and trying to hit
36:56the company goals and versus this person
36:59has an alter ulterior set of goals
37:01it just clashes and so far we've we've
37:05knock on wood we have we'd have great
37:08people but I've just known stories you
37:10read post-mortem said if you get people
37:11in there that that have a different set
37:13of objectives it's it just ruins like
37:16company culture Apple is a very great
37:21example I mean Steve Jobs initially had
37:23a had a great vision about what he
37:25wanted to build and he had a great early
37:26team and they were all aligned but what
37:29is eventual when you get big enough you
37:30just have different business units that
37:32are all pushing for their team in a
37:33different way and it's very challenging
37:35just to just to be able to grow that
37:37culture even when the companies the
37:40thousands of employees and I'd say
37:43actions speak louder than words I I
37:47really like the people on my team that
37:50that I that I that I trust the most I
37:54guess are the people that just just do
37:56the job that just hit their targets
37:59without saying anything people that come
38:00to me and saying hey I need a raise hey
38:02I'm looking for a promotion that that
38:06it it just it's kind of forced versus
38:10the people that are in the corner that's
38:11that do their job and that do the job
38:13extremely well by instinct III look at
38:16them too and I compensate them typically
38:19higher just because I they're doing the
38:21job and they're thinking about the
38:22company first in terms of kind of
38:28learning the lessons fire fast so if
38:31even if you look everyone make makes
38:33mistakes but it's really important to to
38:37make sure that you don't cause a cancer
38:39in your company we've been we've been
38:44but it's hard it's it's hard to do to
38:47get to get rid of people but really you
38:49have to think about it as part of the
38:51business and think about how I can grow
38:53the business and and I'm not trying to
38:55create a personal relationship with
38:56everyone but it's really about how do I
38:58get the most value out of everyone here
39:00and push the company first and that's
39:02one of the most difficult things I've I
39:04have to do is just you know making these
39:07very difficult decisions motivation is
39:12is extremely important so how do you
39:15keep your team motivated we have a we
39:19have kind of a culture ambassador on our
39:21team and that's also one of my roles as
39:25chief operating officers it's just
39:28making people feel engaged and motivated
39:30and wanting to work there every day
39:32that's what I want to spend my mo the
39:35most most amount of time on right now
39:37and it's not on algorithms
39:38I mean it's when you when you just start
39:41off you're gonna be working on the
39:44technology but if you want to be a
39:45founder you're gonna realize you're
39:46gonna move very quickly away from that
39:48just because you're wearing so many
39:50different types of hats and you're gonna
39:53be like again further and further away
39:55and yes you can attend those meetings
39:56but it's gonna be a meeting it's not
39:59going to be an in-depth analysis so if
40:00you want to be a coder and also be a
40:02founder of a large company it's just not
40:04gonna work and so like I said I spend
40:07most of my days thinking about how I'm
40:10gonna motivate my everyone just to
40:11continue to work there and that's
40:13through the values and through the
40:14through the culture that that that that
40:16were that we're building I also
40:20used tools to our advantage so we use a
40:22lot of different metrics independent
40:25analytics internally so like a tool
40:28called moon map which is a great kind of
40:30survey that goes out to all the
40:32employees of the company and that tells
40:34you whether people are happy whether
40:36people are frustrated and the product
40:38and their progress for the week and you
40:40can track that and you can track it by
40:41team and it's again tools that just to
40:44help you do your job and we use tools
40:52I think feedback is really really
40:54critical it's trying to understand and
40:57communicate with with everyone in the
40:59team and get feedback 360 feedback so
41:01your direct reports should give you
41:04feedback and and and and you should get
41:06feedback too you know and and vice-versa
41:10I think what's what really retains good
41:13employees is the positive relationship
41:15with their manager and and then being
41:18challenged I think that that's there's
41:20been studies on this and it's really
41:21really important how many minutes did do
41:24I do I have like okay great I just have
41:30a few more slides the the next thing is
41:36lessons learned so if any of you are
41:39interested in health care what I'll say
41:41is healthcare IT software is very
41:44different from medical device software
41:47so so you'll hear about a lot of apps
41:49that are in the health care space that
41:51collect data and that push data around
41:54it's very different compared to medical
41:57device and well what I'll emphasize
42:00about that is if you're creating a
42:02medical device you're a regulated you're
42:04in a regulated environment and what his
42:06regulations mean it means the whole slew
42:09of infrastructure and documentation
42:12before you can sell your product and one
42:15example is what's called a quality
42:17management management system this is or
42:19QMS this is a whole set of processes and
42:22procedures on how you run your business
42:24so it's think of them as SOPs or
42:27standard operating procedures each
42:30country requires a certain unique set of
42:33of SOPs and you always going to need a
42:35full-time team just to make sure that
42:37that's your procedures are in line with
42:40the regulations that change on a yearly
42:41basis that's a very core component of
42:44having of selling medical devices and
42:45think of it every device that you own
42:49we'll probably have a C II label on it
42:50now if you look at your iPhone it says
42:53see that having a C mark means that you
42:55need to have a quality management system
42:57in place and that requires a lot of
42:58infrastructure and support and
43:00documentation the next thing is design
43:03controls so one thing that's yeah that
43:08you can do as if you're not selling a
43:09medical device is you can hack something
43:11up very quickly you can productize it so
43:13you can push it to the Apple store and
43:15and I start making money very quickly
43:17that's not the case in medical and for
43:20medical software you need to have very
43:21specific design controls which means you
43:25know having a design inputs to providing
43:28certain set of requirements upfront you
43:31need to be able to develop based on
43:33those requirements you need to be able
43:34to test do the verification validation
43:36and there's a whole set of very very
43:39specific set of milestones before you
43:42get your product out and it's it's it's
43:45time-consuming and how does that fit or
43:48that waterfall type of software
43:49development fit into agile software
43:51developments of being able to just hack
43:53something together iterate very quickly
43:54that's uh that's that's an issue and
43:57that's something that we're still trying
43:58to solve today is how do we quickly
43:59iterate on the product get feedback but
44:02also push it out as quickly as possible
44:04that's like I said very tough problem
44:07also I mean test-driven development we
44:09what we try to do is really write the
44:11tests upfront and make sure that that's
44:13that that we that we use those tests and
44:16and the integration testing frameworks
44:17and all of that to make sure that our
44:20medical device is safe and effective so
44:23like I said all those things you
44:26healthcare IT companies may do but it's
44:29just an order of magnitude more work to
44:31try to get something regulated because
44:33you need to prove that works you need to
44:34have the data to support that it works
44:36and either clinical trials or or having
44:41data with it or have users use the data
44:43so it's it's very very important the
44:46picture to the left is actually
44:48our latest submission it was I was 13
44:51inches tall several thousand pages and
44:55it's a lot of work so that's again you
44:58would never have to do that if you're if
44:59you're publishing an app I'll also
45:04mention this particular slide of how
45:06important it is to communicate these are
45:09also kind of lessons that again you
45:11don't learn in school of how to work
45:13with the team I mean yeah they'll maybe
45:15put you in a team but how do you really
45:17communicate effectively and and pass on
45:19information from one team did to the
45:22next so this is an example of a customer
45:25asking for a rope swing and here you'll
45:29see that how the customer explained it
45:30you know maybe I'm a rope swing with
45:33with three steps how the project leader
45:36understood it so this is really really
45:38relevant to us like we we have customers
45:41talking to us every day we have product
45:44managers in that second category and
45:47then those are requirements get sent to
45:48the engineer which then writes these
45:51requirements outs in this case came up
45:53with that design then you pass them to
45:55the dev team the dev team takes that and
45:58they have no idea what the customer
46:00wants and they build something like that
46:02the sales team will undoubtedly try to
46:05sell something magical even though it's
46:07half-baked and doesn't work and there's
46:11no documentation typically you have you
46:14have the what we have the operations or
46:17service team that goes installs it again
46:19there's no communication how the
46:22customer was billed how the how the
46:24helpdesk supported it and what the
46:26customer really wanted so that is a
46:29funny cartoon but it's truly what what
46:31what's very typical and in what we see
46:34and every team is really important for
46:37the clinical further for the end product
46:38so what I have there on the right is how
46:40important is each team to to making the
46:43products be successful and again these
46:46are some of the things you just don't
46:47don't learn in school what I'll mention
46:51here is kind of transitioning from
46:54academic to entrepreneur really focuses
46:57really requires focusing on commercial
47:01don't be a technology looking for a
47:03problem I think that's very important
47:06and it's hard to it's hard to detach
47:09from the technology focus on the people
47:13management on the hiring on the
47:14operations or the processes and really
47:16the the go-to-market so how are you
47:18gonna sell this product how are you to
47:20market that those are really types of
47:21things that are difficult to just
47:23extract from with that I I will I think
47:29what we are is it's a time almost I'll
47:32just mention a start-up is a roller
47:33coaster promotions you're gonna be happy
47:36one day and and angry another day I
47:38would say be even keeled and because
47:41it's extremely stressful I'll say the
47:45second thing is delegate it's really
47:47important to know when just to step out
47:50of a problem trust the people around you
47:51empower them to make very important
47:54decisions next thing for us is hire
47:57great software developers and software
48:00development teams specifically for us
48:02and we talked about this 10x product
48:04well there's the 10x uniform unicorn
48:06about but also just the 10x team I'll
48:10say entrepreneurship isn't for everyone
48:12it's stressful it's it's you know
48:15there's it's also you have a lot
48:17responsibilities you're anxious you're
48:19anxiety and maybe it's not for you and
48:21if you if you want to try things out
48:23maybe go work for start-up and maybe
48:26being an early employee at a startup
48:27just to see if that's right for you
48:30and lastly applying for the job so let's
48:34say you don't want to be a founder of a
48:35company and you want to apply for a
48:36certain job I have just a quick set of
48:39of little rule rules that I'll say is be
48:44very specific as to which companies you
48:46want to work for target them on
48:48CrunchBase go to CrunchBase comm and see
48:50the starts that have recently gotten
48:52funding and then focus on those
48:53companies and focus on who's hiring
48:55who's the hiring managers and don't just
48:57send a resume and cover letter send them
49:00a message and and be very specific as to
49:02how your skillset can help them you have
49:05no idea how many resumes I receive and I
49:07just never look at but the ones that
49:09just semi direct business plan to say
49:13hey I know I can solve this particular
49:15for you I actually look at so that's
49:18just one one lesson also lastly this is
49:21my five month old newborn and I'll say
49:25it's it's challenging to be a family
49:29person but also be an entrepreneur so
49:31take risks when you're young it's it's
49:33much harder when you have a bigger
49:34family and with that thank you yeah
49:53that's a great question I went when
49:58you're small I think you have to focus
50:00less on it and it's really organic
50:02really that the people that you hire the
50:04two or three people that are there to
50:05start off you really don't think about
50:08it but yet you it's it's a one-degree
50:13connection between what you want and
50:15still and the values that you have with
50:17with with the people that are just core
50:19around you so it's so it's just organic
50:21that way as we grow it's much more
50:23challenging to have this this kind of
50:25self-replicating effect and that's why
50:27we put in so much more effort to try to
50:29keep that same company culture so if I
50:32were to say in an ideal world it'll
50:34always be the same and it has been
50:35that's the best answer but it's probably
50:38not the same it was what it was when
50:40there's five people but we're working
50:42it's constant evolution where we're
50:44constantly working towards and and it's
50:46not just I want to build a culture
50:48that's not just a nine-to-five job it's
50:50it's all about making people motivated
50:52and wanting to work there yeah
50:58I like to ask how did you get the same
51:02sure you come work with a just idea and
51:05these expand and go to the NGOs and get
51:07I mean before you make up for that or
51:09you make so great question so question
51:14was around how we got our seed funding
51:16and a couple options was you know you
51:19you make the product and you try to
51:21raise capital or you just have an idea
51:23or back in the envelope idea and then
51:25trying to raise capital off that for us
51:27it was really about before our seed
51:30rounds is just getting enough friends
51:32and family cash to try to develop some
51:35proof of concept that we have something
51:37there and maybe that some something
51:39could be something that you license from
51:40the from other companies or that you
51:42license from Stanford which was in our
51:44case what we did so so we just graze a
51:46little bit of money just to have some
51:48core piece of technology and then from
51:50there the way we raised our series seed
51:54round so it's it's it's unique it
51:56depends on the it depends on the on the
51:59on the technology but I think you'll see
52:02less and less of paper napkin kind of
52:05checks I think it's so competitive right
52:08now that investors are looking for
52:10something real something that shows
52:12traction and that's why you almost need
52:14either users you need revenue or you
52:17need something attractive in order for
52:25that's the new mission and market
52:28message for them I am wondering how did
52:32you how do you make sure that the market
52:34you thought is really important
52:37yeah so the question was around the
52:39market analysis and and how did we do it
52:41and what and how did we assess that it
52:43really wasn't a market need there and I
52:46would say we didn't really do a good job
52:48and actually with this
52:50we said oh there's a great technology
52:52let's let's try to sell this and I think
52:56that the real weight is to start doing
52:57is talking to customers talking to users
53:00there's a lot if you go to the B school
53:02or the business school there's a lot of
53:04resources there in terms of looking at
53:06the market and there's there's market
53:11reports so it going going there and and
53:14reading those reports aggregating data
53:15paying for data paying for surveys those
53:19are all things that that we do now and
53:21we spend tens of thousands of dollars on
53:23surveys market analysis
53:25customer interviews just and and you
53:28know the blinded studies so we do a lot
53:31of different things to see if there's
53:33really a market there and like I said in
53:37the beginning if you don't have money
53:37you just talk to a few people but I'd
53:40say talk to people outside of Silicon
53:41Valley especially if you're selling a
53:43product that's global because you can
53:46get so skewed here in the valley
53:48especially if you're trying to sell song
53:50in Stanford everyone at Stanford is an
53:52expert go talk to some hospital in rural
53:54Ohio and just see if they're willing to
53:56pay for something that's where you
![Welp, Here's what the Fed *Just* Said [Fed FOMC Minutes]](/_next/image?url=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fl8u23oq1E9E%2Fhqdefault.jpg&w=3840&q=75)
