00:08 welcome everyone to how to build and
00:12 succeed as a technical founder for the
00:15 startup School talk quick intro I'm
00:17 Diana who I'm currently a group partner
00:20 at YC and previously I was a co-founder
00:24 and CTO for Azure reality which was a
00:27 startup building augmented reality SDK
00:30 for game developers and we eventually
00:33 had an exit and sold to Niantic where I
00:37 was the director of engineering and
00:38 heading up all of the AR platform there
00:41 so I know a few things about building
00:43 something from was just an idea to then
00:47 a prototype to launching an MVP which is
00:50 like a bit duct tapey to then scaling it
00:52 and getting to product Market fit and
00:55 scaling systems to millions of users so
00:58 what are we going to cover in this talk
01:00 is three stages first is what is the
01:05 role of the technical founder and who
01:07 are they number two how do you build in
01:11 each of the different stages where all
01:13 of you are in startup school ideating
01:16 which is just an idea you're just
01:17 getting started building an MVP once you
01:20 got some validation and getting it to
01:22 launch and then launch where you want to
01:24 iterate towards product Market fit
01:26 and then I'll have a small section on
01:29 how the role of the technical founder
01:31 evolved Pro product Market fit I won't
01:33 cover it too much because a lot of you
01:35 in startup School are mostly in this
01:38 earlier stage and I'm excited to give
01:40 this talk because I compiled it from
01:43 many conversations and chats with many
01:45 YC technical Founders like from algolia
01:50 segment optimal easily way up so I'm
01:53 excited for all of their inputs and
01:55 examples in here all right the technical
01:58 founder sometimes I hear non-technical
02:01 Founders say I need somebody to build my
02:04 app so that isn't going to cut it a
02:08 technical founder is a partner in this
02:11 whole journey of a startup and it
02:14 requires really intense level of
02:16 commitment and you're in just a Dev what
02:18 does a technical founder do they lead a
02:23 building of the product of course
02:26 and also talking with users and
02:29 sometimes I get the question of who is
02:32 the CEO or CTO for a technical founder
02:34 and this is a nuanced answer it really
02:37 depends on the type of product the
02:39 industry you're in the complete scale
02:41 composition of the team to figure out
02:44 who the CEO of CTO is and I've seen
02:47 technical Founders be the CEO the CTO or
02:50 various other roles and what does the
02:52 role of the technical founder look like
02:54 in the early eight stages it looks a lot
02:57 like being a lead developer like if
03:00 you've been a lead developer a company
03:01 you were in charge of putting the
03:03 project together and building it and
03:05 getting it out to the finish line or if
03:07 you're contributing to an open source
03:08 project and you're the main developer
03:10 you make all the tech choices but
03:12 there's some key differences from being
03:15 a lead developer you got to do all the
03:18 tech things like if you're doing
03:20 software you're gonna have to do the
03:22 front and the back end devops the
03:25 website the ux even I.T to provision the
03:29 Google accounts anything if you're
03:32 building hardware and maybe you're just
03:34 familiar familiar with electrical and
03:35 working with eaglecad you'll have to get
03:37 familiar with the mechanical too and
03:40 you'll of course as part of doing all
03:42 the tech things you'll have to talk with
03:43 users to really get those insights to
03:46 iterate and you're going to have a bias
03:48 towards building a good enough versus
03:51 the perfect architecture because if you
03:54 worked at a big company
03:56 you might have been rewarded for the
03:58 perfect architecture but not for a
04:00 startup you're going to have bias
04:02 towards action and moving quickly and
04:04 actually deciding with a lot of
04:06 incomplete information you're gonna get
04:08 comfortable with technical debt
04:10 inefficient processes and a lot of ugly
04:13 code and basically lots of chaos and all
04:17 of these is to say is the technical
04:20 founder is committed to the success of
04:24 and that means doing whatever it takes
04:26 to get it to work and it's not going to
04:29 cut it if you're an employee at a
04:30 company I sometimes hear oh this task or
04:33 this thing is not in my pay grade no
04:35 that's not going to cut it here you got
04:37 to do you gotta do it this next session
04:39 on how to build the first stage is the
04:43 ideating stage where you just have an
04:46 idea of what you want to build and the
04:48 goal here is to build a prototype as
04:52 soon as possible with the singular Focus
04:55 to build something to show and demo to
04:58 users and it doesn't even have to work
05:01 fully in parallel your CEO co-founder
05:05 will be finding a list of users in these
05:08 next couple days to TF meetings to show
05:10 the Prototype when it's ready so the
05:12 principle here is to build very quickly
05:14 in a matter of days and sometimes I hear
05:17 it's like oh Diana a day prototype that
05:20 seems impossible how do you do it and
05:23 one way of doing it is building on top
05:25 of a lot of prototyping software and you
05:28 keep it super super simple so for
05:30 example if you're a software company you
05:33 will build a clickable prototype perhap
05:35 using something like figma or Envision
05:38 if you're a devtools company you may
05:40 just have a script that you wrote in an
05:42 afternoon and just launch it on the
05:45 if you're a hardware company or heart
05:47 attack it is possible to build a
05:49 prototype maybe it takes you a little
05:50 bit longer but the key here is 3D
05:52 renderings to really show you the
05:54 promise of what the product is and the
05:56 example I have here is a company called
05:58 Remora that is helping trucks capture
06:02 carbon with this attachment and that
06:05 example of that rendering was enough to
06:08 get the users excited about their
06:11 product even though it's hard tech so
06:13 give you a couple examples of prototypes
06:15 in the early days this company
06:16 optimizely went through YC on winter 10
06:19 and they put this prototype literally in
06:22 a couple of days and the reason why is
06:24 that they had applied with YC with a
06:26 very different idea they started with a
06:28 Twitter referral widget and that idea
06:30 didn't work and they quickly found out
06:33 so they strapped together very quickly
06:35 this prototype and it was because the
06:37 founders uh Pete and Dan and Dan was
06:40 actually heading analytics for the Obama
06:42 campaign and he recalled that he was
06:44 called to optimize one of the funding
06:47 pages and thought huh this could be a
06:49 startup so they put a very together very
06:51 quickly together and it was the first
06:53 visual editor by creating a a b test
06:55 that was just a Javascript file that
06:58 lived on S3 I literally just opened
07:00 option command J if you're in Chrome and
07:03 they literally run manually the A B test
07:05 there and it would work of course nobody
07:07 could use it except the founders but it
07:09 was enough to show it to marketers who
07:11 were the target users to optimize sites
07:13 to get the user excited so this was
07:16 built in just few days other example is
07:20 my startup Azure reality since we're
07:22 building more harder Tech we had to get
07:25 computer vision algorithms running on
07:27 phones and we got that done in a few
07:29 weeks that was a lot easier to show a
07:32 demo of what AR is as you saw on the
07:34 video than just explaining and hand
07:37 waving and made selling and explaining
07:39 so much easier now what are some common
07:42 mistakes on prototypes you don't want to
07:45 overbuild at this stage I've seen people
07:48 have this bias and they tell me hey
07:50 Diana but users don't see it or it's not
07:54 good enough this prototype doesn't show
07:55 the whole Vision this is the mistake
07:57 when founder things you need a full MVP
07:59 and the stage and not really the other
08:02 mistake is obviously not talking or
08:05 listening to users soon enough that
08:07 you're gonna get uncomfortable and show
08:08 this kind of prototyping duct type thing
08:11 that you just slap together and that's
08:14 okay you're gonna get feedback the other
08:16 one at the stage as an example for
08:18 optimizely when founders get too
08:21 attached to idea I went up the feedback
08:23 from users is something obvious that is
08:25 not quite there not something that users
08:27 want and it's not letting go of bad
08:30 ideas okay so now into the next section
08:33 so imagine you have this prototype you
08:36 talk to people and there's enough
08:38 interest then you move on to the next
08:39 stage of actually building an MVP that
08:42 works to get it to launch and the goal
08:44 is basically build it to launch and it
08:47 should be done also very quickly ideally
08:49 in a matter of can be done a few days
08:51 two weeks or sometimes months but
08:54 ideally more on the weeks range for most
08:56 software companies again exceptions to
08:58 hardware and deep tech companies so the
09:00 goal here at this stage is to build
09:03 something that you will get commitment
09:06 from users to use your product and
09:08 ideally what that commitment looks like
09:10 is getting them to pay and the reason
09:12 why you have a prototype is while you're
09:14 building this your co-founder or CEO
09:17 could be talking to users and showing
09:19 the Prototype and even getting
09:20 commitments to use it once is ready to
09:23 launch so I'm gonna do a bit of a bit of
09:25 a diversion here because sometimes
09:28 Founders get excited it's like oh I show
09:29 this prototype people are excited and
09:31 there's so much to build is hiring a
09:35 first is thing is like okay I got this
09:37 prototype got people excited I'm gonna
09:39 hire people to help me to build it as a
09:41 first-time founder he's like oh my God
09:42 oh my God there's a fit people want it
09:43 is it a good idea it really depends it's
09:46 gonna actually slow you down in terms of
09:50 because if you're hiring from a pool of
09:53 people and Engineers that you don't know
09:54 it takes over a month or more to find
09:57 someone good and it's hard to find
09:59 people at this stage with very nebulous
10:01 and chaotic so it's going to make you
10:02 move slowly and the other more Insidious
10:05 thing is going to make you
10:08 not develop some of the insights about
10:10 your product because your product will
10:12 evolved if someone else in your team is
10:14 building that and not the founders
10:15 you're gonna miss that key learning
10:17 about your tag that could have a gold
10:19 nugget but it was not built by you I
10:21 mean there's exceptions to this I think
10:23 you can hire a bit later when you have
10:26 things more built out but at this stage
10:28 it's still difficult so I'll give you a
10:31 example here uh Justin TV and twitch it
10:34 was just the four Founders and three
10:36 very good technical Founders at the
10:39 beginning for the MVP it was just the
10:42 founders building software as software
10:45 engineers and the magic was Justin
10:47 Emmett and Kyle Building different parts
10:50 of the system you had Kyle who become an
10:52 awesome Fearless engineer tackling the
10:54 hard problems of video streaming and
10:57 then Emma doing all the database work
10:59 Justin with the web and that was enough
11:00 to get it to launch I mean I'll give you
11:03 an exception after they launched they
11:05 did hire good Engineers but the key
11:07 thing about this they were very good at
11:10 not caring about the resume they try to
11:12 really find The Misfits and engineers at
11:15 Google overlooked and those turned out
11:16 to be amazing so Amon and Golem were
11:20 very comfortable and awesome engineers
11:22 and they took on a lot of the video
11:24 weapon just three months since joining
11:25 you want people like that that can just
11:27 take off and run all right so now
11:30 going back into the principles for for
11:33 building towards your MVP principle one
11:35 is the classic hologram essay on do
11:39 things that don't scale
11:41 basically find clever hacks to launch
11:43 quickly in the spirit of doing things at
11:46 those scale and the Drake posting
11:49 edition of this avoid things like
11:51 automatic self onboarding because that
11:53 adds a lot of engineering building a
11:55 scalable back-end automated scripts
11:57 those sounds great at some point but not
11:59 the stage and the hack perhaps could be
12:01 manually onboarding you're literally
12:03 editing the database and adding the
12:05 users or the entries and the data on the
12:07 other counterter thing is insane custom
12:10 support it's just you the founders at
12:12 the front line doing the work doing
12:14 things that don't scale a classic sample
12:16 is with stripe this is the site when
12:17 they launch very simple they had the API
12:19 for developers to send payments but on
12:22 the back end the thing that did not
12:24 scale it was literally the founders
12:26 processing every manual request and
12:29 filling Bank forms to process the
12:31 payments at the beginning and that was
12:33 good enough to get them to launch sooner
12:34 now principle number two this is famous
12:37 create 9010 solution that was coined by
12:41 Paul bukite who was one of the group
12:44 Partners here at YC and original
12:45 inventor of Gmail the first version is
12:48 not going to be the final remember and
12:49 they will very likely a lot of the code
12:51 be Rewritten and that's okay push off as
12:54 many features to post launch and by
12:57 launching quickly I created a 9010
12:58 solution I don't mean creating bugs I
13:01 still want it good enough but you want
13:03 to restrict the product to work on
13:06 limited Dimensions which could be like
13:08 situations type of data you handle
13:10 functionality type of users you support
13:13 could be the type of data the type
13:15 number of devices or it could be Geo
13:17 find a way to slice the problem to
13:19 simplify it and this can be your secret
13:21 superpowers that startup at the
13:22 beginning because you can move a Lot
13:24 quickly and large companies can't afford
13:26 to do this or even if your startup gets
13:28 big you have like lawyers and finance
13:31 teams and sales team that make you kind
13:34 of just move slow so give you a couple
13:36 examples here doordash at the beginning
13:38 they slapped it in one afternoon soon
13:41 and they were actually called Palo Alto
13:45 delivery and they took PDS for menus and
13:48 literally put their phone number that
13:50 phone number there is actually from one
13:52 of the founders and there's the site is
13:54 not Dynamic static it's literally just
13:57 plain HTML and CSS and PDF that was our
14:00 front end they didn't bother with
14:02 building a back end the back end quote
14:03 unquote was literally just Google forms
14:06 and Google Docs where they coordinated
14:07 all the orders and they didn't even
14:10 build anything to track all the drivers
14:12 or ETA they did that with using fancy on
14:16 your iPhone find my friends to track
14:18 where each of the deliveries were that
14:20 was enough so this was put together
14:21 literally in one afternoon and they were
14:24 able to launch the very genius thing
14:26 they did is that because they were
14:27 Stanford student they constrained it to
14:29 work only on Palo Alto and
14:31 counterintuitively by focusing on Palo
14:33 Alto and getting that right as they grew
14:35 it got them to focus and get delivery
14:38 and unit economics right in the suburbs
14:41 right at the beginning so that they
14:43 could scale that and get that right
14:45 versus the competition which was
14:47 focusing on Metro cities like GrubHub
14:49 which make them now you saw how the
14:52 story played out the unit economics and
14:54 the Ops was much harder and didn't get
14:56 it right so funny thing about focusing
14:58 at the beginning and getting those right
15:00 can get you to focus and do things right
15:02 that later on can serve you well so now
15:04 at this stage how do you choose a tech
15:08 stack so what one thing is to balance
15:10 what makes sense for your product and
15:12 your personal expertise to ship as
15:15 quickly as you can keep it simple don't
15:18 just choose a cool new programming
15:20 language just to learn it for your
15:21 startup choose what you're dangerous
15:23 enough and comfortable to launch quickly
15:25 which brings me to the next principle
15:28 choose the tag for iteration speed I
15:31 mean now and the other thing is also
15:34 it's very easy to build MVPs very
15:37 quickly by using third-party Frameworks
15:39 on API tools and you don't need to do a
15:42 lot of those work for example
15:43 authentication you have things like auth
15:45 zero payments you have stripe
15:46 cross-platform support and rendering you
15:48 have things like react native Cloud
15:50 infrastructure you have AWS gcp landing
15:53 pages you have webflow back-end back-end
15:56 serverless you have lambdas or Firebase
15:59 or hosted database in the past startups
16:03 would run out of money before even
16:04 launching because they had to build
16:06 everything from scratch and shift from
16:08 metal don't try to be the kind of like
16:10 cool engineer just build things from
16:12 scratch no just use all these Frameworks
16:14 but I know ctOS tell me oh it's too
16:16 expensive to use this third-party apis
16:19 or it's too slow it doesn't skill to use
16:22 XYZ so what I'm going to say to this I
16:25 mean there's there's two sides of the
16:27 story with using third party
16:29 I mean to move quickly but it doesn't
16:31 mean this this is a great meme that Sean
16:35 Wang who's the head of developer
16:38 experience that everybody posted the
16:40 funny thing about it is you have at the
16:42 beginning quartile kind of the noob that
16:45 just learned PHP or just JavaScript and
16:48 just kind of use it to build the toy car
16:50 serious engineers make fun of the new
16:52 because oh PHP language doesn't scale or
16:55 JavaScript and all these things it's
16:57 like oh our PHP is not a good language
16:59 blah blah and then the middle or average
17:02 or mid-wit Engineers like okay I'm gonna
17:04 put my big engineer pants and do what
17:07 Google would do and build something
17:09 optimal and scalable and use something
17:11 for the back end like Kafka Linker Ros
17:13 AMA Prometheus kubernetes Envoy big red
17:17 or hundreds of microservices okay that's
17:20 the average technical founder the
17:22 average startup dies so that's not a
17:23 good outcome another funny thing you got
17:25 the Jedi Master and when you squint
17:28 their Solutions look the same like the
17:30 new one they chose also PHP and
17:33 JavaScript but they choose it for
17:35 different reasons not because they just
17:36 learned it but they wreck recognizes
17:39 this is because they can move a lot
17:41 quicker and what I'm going to emphasize
17:43 here is that if you build a company and
17:46 it works and you get users good enough
17:48 the tech choices don't matter as much
17:50 you can solve your way out of it like
17:52 Facebook famously was built on PHP
17:55 because Mark was very familiar with that
17:57 and of course PHP doesn't quite scale or
17:59 is very performant but if you're
18:02 Facebook and you get to that scale of
18:04 the number of users they got you can
18:05 solve your way out and that's when they
18:08 built a custom transpiler called hip hop
18:11 to make PHP compound C plus plus so that
18:14 it would optimize see so that was the
18:16 Jedi move and even for JavaScript
18:18 there's a V8 engine which makes it
18:21 pretty performant so I think it's fine
18:22 way up was a 2015 company at YC that
18:27 helps company hire diverse companies and
18:29 is a job board for college students so
18:32 JJ the CTO although he didn't formally
18:35 study computer science or engineering at
18:37 UPenn he that taught himself how to
18:39 program on freelance for a couple years
18:42 before he started way up and JJ chose
18:45 again as the Jedi Master chose
18:47 technology for iteration speed he chose
18:50 Django and python although a lot of
18:52 other peers were telling him to go and
18:55 use Ruby and rails and I think in 2015
18:58 Ruby and rails were 10 times more
19:01 popular by Google Trends and that was
19:03 fine that that didn't kill the company
19:05 at all I mean that was the right choice
19:06 for them because he could move and get
19:08 this move quickly and get this out of
19:10 the door very quickly I kept it simple
19:12 in the back end postgres python
19:14 Heroku and that worked out well for them
19:17 now I'm going to summarize here the only
19:20 Tech choices that matter are the ones
19:23 tied to your customer promises for
19:25 example at Azure we in fact rewrote and
19:28 threw away a lot of the code multiple
19:30 times as we scale in different stages of
19:32 our Tech but the promise that we
19:34 maintain to our customers was at the API
19:36 level in unity and game engines and
19:39 that's the thing that we cannot throw
19:40 away but everything else we rewrote and
19:42 that's fine all right now we're gonna go
19:44 part three so you have the MVP you built
19:48 it and launched it now you launched it
19:50 so what happens on this stage your goal
19:53 here in the launch stage is to iterate
19:56 to get towards product Market fit so
19:59 principle number one is to quickly
20:02 iterate with hard and soft data use hard
20:06 data as a tech founder to make sure you
20:09 have set up a dashboard with analytics
20:10 that tracks your main kpi and again here
20:13 choose technology for your analytics
20:15 stack for Speed keep some keep it super
20:18 simple something like Google analytics
20:19 amplitude mix panel and don't go
20:22 overboard with something super complex
20:24 like lock stash Prometheus these are
20:27 great for large companies but not at
20:28 your stage you don't have that load
20:30 again use Soft Data if I keep talking to
20:33 users after you launch and marry these
20:36 two to know why users stay or churn
20:39 and ask to figure out what new problems
20:41 your users have to iterate and build we
20:44 pay another YC company when they launch
20:46 they were at b2c payments product kind
20:49 of a little bit like venmo-ish but the
20:51 thing is that it never really took off
20:53 they iterated so in terms of analytics
20:56 they saw some of the features that we're
20:58 launching like messaging nobody cared
21:00 nobody used and they found out in terms
21:03 of a lot of the payments their biggest
21:04 user was GoFundMe back then they also
21:07 talked to users they talk to GoFundMe
21:08 who didn't care for any of this b2c UI
21:12 stuff they just care to get the payments
21:15 and then they discover a better
21:17 opportunity to be an API and basically
21:19 pivoted it into it and they got the
21:22 first version and again applying the
21:24 principles that did a scale they didn't
21:26 even have technical docs and they worked
21:28 with GoFundMe to get this version and
21:31 this API version was the one that
21:33 actually took off and got them to
21:34 product Market fit principle number two
21:36 in this launch stage is to continuously
21:39 launch perfect example of this is a
21:41 segment who started as a very different
21:42 product they were classroom analytics
21:45 similar stories they struggled with this
21:48 first idea it didn't really work out
21:49 until they launched a stripped out
21:52 version of just their back end which was
21:53 actually segment and see the impressive
21:56 number of launches they did their very
21:59 first launch was back in December
22:02 2012. that was their very first post and
22:05 you saw the engagement in Hacker News
22:07 very high that was a bit of a hint of a
22:09 product Market fit and they got excited
22:11 and they pivoted into this and kept
22:13 launching every week they had a total of
22:15 five launches in a span of a month or so
22:19 and they kept adding features and
22:20 iterating they added support for more
22:23 things when they launched it only
22:24 supported Google analytics mixpanel and
22:26 intercom and by listening to the users
22:28 they added node PHP support and
22:31 WordPress and it kept on going
22:34 and it took them to be then a unicorn
22:37 that eventually had an exit to Twilight
22:39 for over three billion dollars pretty
22:44 the last principle here what I want to
22:46 say for when you're launch there's this
22:48 funny state where you have Tech builds
22:50 you want to balance building versus
22:52 fixing you want to make thoughtful
22:54 choices between fixing bugs or adding
22:57 new features or addressing technical
22:58 debt and one I want to say Tech debt is
23:01 totally fine you gotta get comfortable a
23:03 little bit with the heat of your Tech
23:05 burning totally okay you're gonna fear
23:07 the right things and that is towards
23:09 getting you product Market fit sometimes
23:11 that tiny bug and rendering maybe is not
23:14 critical for you at this point to fix
23:16 like in fact a lot of early products are
23:19 very broken you're probably very
23:21 familiar with Pokemon go when it
23:23 launched in 2016 nobody could log into
23:26 the game and guess what that did not
23:28 kill the company at all in fact to this
23:31 day Pokemon I think last year made over
23:34 a billion dollars in Revenue that did
23:36 not kill them and I'll give a little
23:38 background what was happening on the
23:39 tech it was very uh very straightforward
23:42 they had a load balancer that was on
23:44 Google cloud and they had a back-end and
23:47 they had a TCP termination and HTTP
23:49 requests that were done with their nginx
23:52 to route to the different servers that
23:54 were the AFE the application front end
23:56 to manage all the requests and the issue
23:59 with there it was that as users were
24:02 connected they didn't get terminated
24:04 until they got to the nginx
24:06 and then as a result client also had
24:09 retries and that what happened when you
24:12 had such a huge load that in fact I
24:14 think Pokemon go by the first month
24:17 after launching they had the same number
24:19 of uh active as as Twitter which took
24:22 them 10 years to get there and they got
24:23 there in one month of course things
24:25 would break it was basically a lot of
24:26 users trying to log in was kind of
24:28 creating a bit of a dito's attack now
24:30 December is a bit on when you launch
24:32 some of the common mistakes after
24:34 launching and I myself has made CTO Doge
24:38 sad it is tempting to to build and say
24:41 what would Google do that's almost
24:43 certainly a trap would try to build like
24:47 a big company or hiring to try to move
24:51 quickly sometimes I think this is more
24:53 of a nuanced question can be a mistake
24:55 or the other thing is focusing too much
24:57 on fixing refactoring and not building
24:59 features towards iterating to product
25:01 Market fit not discovering insights from
25:03 users sometimes I see ctOS like okay we
25:06 launched I get to conquer down and just
25:08 get into building totally no again your
25:10 role as a technical founder very
25:12 different you got to be involved in the
25:14 journey and really understand the
25:15 insights of why users Stay or Leave Your
25:18 products you have to keep talking to
25:19 them and the other mistake I see is like
25:21 oh we're just building features for
25:23 but you also need to build Tech to grow
25:27 in fact some of the best growth hacks
25:30 where Engineers pair it up with sales
25:34 and growth folks who are non-technical
25:35 so now the last section on how the role
25:38 evolves so assuming you got product
25:40 Market fit what happens this is this
25:43 point where you can actually then put on
25:46 your big engineering pants
25:48 and figure out pieces of the tech that
25:52 need to be built to scale you need to
25:55 and the attack will break which is
25:57 actually a good thing breaking because
25:58 of too much demand and that's totally
26:00 okay that's my example from Pokemon go
26:03 you'll find the pieces that need to be
26:05 reworked refactor this is when you do it
26:07 not before now not before product Market
26:09 fit and you'll decide also what the
26:12 engineering culture will look like and
26:14 this is a stage where you actually do
26:15 more of the hiring and here you're
26:17 probably going to evolve from leading a
26:20 small team of Engineers to hiring your
26:22 first hires who are going to be people
26:23 that you know and at this point Your
26:25 Role really changes because you'll start
26:27 having communication overhead and this
26:30 is when you realize your role morphs
26:32 like between two to five you still get
26:34 time to code about 70 when you get to
26:36 five to ten you only have less than 50
26:38 percent and Beyond 10 you probably won't
26:41 really have time to code and have to
26:43 decide how to structure things and
26:45 whether you're going to remain as a
26:47 architect type or role or you want to be
26:49 more of a people role and be more of a
26:50 BP rich now to summarize uh hear the
26:53 talk first stage ideating Bill the goal
26:57 is to build a prototype as soon as
26:59 possible and the principle is built very
27:02 quickly in a matter of days stage two
27:04 you're in the process of building an MVP
27:07 which I think a lot of you are in this
27:09 or the previous one the goal is to build
27:11 as quickly to launch in a matter of few
27:15 and the principles are do things that
27:16 don't scale create a 90 10 solution
27:18 choose the tech for iteration speed and
27:22 the last one is once you launch all of
27:25 the previous ideas on 9010 solution do
27:28 things that don't scale still apply and
27:30 add these onto it and the goal is to get
27:33 an iteration towards product Market fit
27:35 so you're going to also quickly iterate
27:37 with hard and soft data with analytics
27:39 and user interviews
27:41 you're going to continuously launch and
27:44 you're going to find the fine balance
27:46 between building and fixing and where
27:49 techdat is totally fine
27:51 feel the heat for that Tech that is
27:53 totally fine and if there's only one
27:55 take away from this whole talk is that
27:57 startups move quickly so thank you