00:00 Hardware is never a choice by preference
00:04 you don't build a hardware because you
00:05 want to build a cool Hardware most of
00:07 the time if you do things like that it
00:09 it will fa miserably a lot of good cases
00:12 I learned and I experienc is that you
00:15 have a software that's so good that you
00:18 want the dedicate Hardware to make it
00:19 better it's always around the software
00:21 it's always around you know what's
00:23 inside and then if you have a software
00:26 that's very very edgy very very new you
00:29 want to D on a hardware at least for the
00:32 first generation got it so so to me R1
00:35 is a result of choose Hardware by
00:38 necessity not by preference this week in
00:41 startups is brought to you by open phone
00:44 brings your teams business calls texts
00:47 and contacts into one delightful app
00:49 that works anywhere get 20% off your
00:52 first 6 months at open phone.com
00:56 Twist scalable path want to speed up
00:59 your product development without
01:01 breaking the bank since 2010 scalable
01:04 path has helped over 300 companies hire
01:07 deeply vetted engineers in their time
01:09 zone visit scalable path.com twist to
01:14 get 20% off your first month and wizard.
01:19 will help bring your vision to life in
01:21 minutes not days with the power of AI
01:25 visit wizard. i/ twist to get 25% off
01:29 wizard Pro for an entire year that's u z
01:36 D.I twist all right everybody last week
01:40 we discussed the rabbit R1 which was I
01:43 think the most notable new product to
01:45 come out of CES the Consumer Electronic
01:47 Show in Las Vegas you may have seen the
01:50 demo and viral on uh X Twitter really
01:54 cool looking product a handheld a pocket
01:57 AI companion it has a wheel on it for
01:59 nav navigation it's got a camera that
02:01 flips over and it's got an LED screen I
02:03 guess and uh is in this Bright Beautiful
02:05 orange my favorite color uh and it was
02:09 the bell of the ball they did a great
02:11 demo and I thought we would bring uh the
02:14 founder on today to talk about it really
02:16 feels retro in some ways uh but it's got
02:18 a custom llm a large language model and
02:21 the $200 price point made everybody lose
02:23 their mines they' sold over 60,000 of
02:27 these units already over the last 5 days
02:30 they've been selling about 10,000 units
02:32 a day so anyway the founder is here his
02:34 name is Jesse Lou and the firm is based
02:37 in C Monica California Jesse how are you
02:39 doing good how are you Jason nice to
02:41 meet you nice to meet you as well did
02:43 you expect this kind of reaction for
02:45 your device did you think it would
02:47 become the most viral product I guess
02:50 along with the see-through TV that was
02:53 the other product that a lot of people
02:54 are losing their minds about which I
02:56 don't even know what the application of
02:58 a see-through TV is if it's up against
02:59 the wall wall what am I going to see the
03:01 wall behind it makes no sense um really
03:03 to me so uh what was it like and did you
03:06 expect this uh no not at all uh I mean
03:10 to be perfectly fair with with with our
03:13 audience and to be honest with with you
03:16 we expect maybe 500 units for the first
03:20 day and maybe 3,000 unit for the early
03:25 adulter Market uh that's probably it um
03:28 but we we actually prepared to ramp up
03:32 the order if necessary so we we kind of
03:35 have a plan B but I guess I'm the most
03:38 conservative one in my entire team and I
03:42 think you know our marketing team and
03:45 design team might have a little bit more
03:47 confidence than us uh but me I'm super
03:51 conservative but one thing that I have
03:54 to mention is that I love the product
03:56 the first prototype was actually a
03:58 Rasberry p with a screen because because
04:02 we're I guess we're like super Underdog
04:04 team and we actually roll out a web
04:06 version for for for a very limited
04:09 amount of of users to test like one
04:12 feature of the Lamb the large action
04:14 model which is play Spotify uh earlier
04:17 uh last year and the result is good and
04:21 and I love the product because I had the
04:25 first you know prototype in this correct
04:30 about I guess 4 months ago and I had the
04:34 pre-built handmade prototype around
04:37 eight months ago and I've been you know
04:40 playing with them and carrying with them
04:41 I love the product but on the other hand
04:44 I'm little bit in fear that you know
04:46 maybe we're just a bunch of Geeks and
04:49 this is like just our own little thing
04:51 well making something for yourself that
04:54 gives you Joy at least you know you have
04:56 one customer yourself right and then you
04:58 just have to figure out there's other
05:00 customers so I guess let's start off
05:02 with this crazy price point it's 200
05:04 bucks for the device yeah and so I
05:06 looked at that and and there's no
05:08 subscription I mean you do need to have
05:10 a data subscription if you want to put
05:11 an LTE card in it right if you want to
05:13 put 5G in it that's on you you just buy
05:15 I guess a Google fi card for a data I
05:18 don't know what those cost probably like
05:19 20 30 bucks a month they're pretty cheap
05:21 how will you ever make money if this is
05:24 $200 and the hardware is so beautiful
05:27 how are you going to make a business out
05:30 yeah so uh I guess that's one of the
05:32 most asked questions from business
05:35 perspective uh first of all I can tell
05:38 you that Jason we worked really hard to
05:41 make the perfect balance between design
05:44 and and bomb and even though I cannot
05:47 give you a very exact number of the bomb
05:49 because I'm not allowed to share that
05:51 but we're making money out of out of the
05:54 hardware uh but but but given by what I
05:58 observed and what I learn in my past
05:59 careers the hardware margin is so low
06:02 you know if you look at a phone you're
06:03 looking at negative 25% to 7% 8% that's
06:06 it uh still most of people they're
06:09 trying to make money out of you know
06:11 Hardware uh obviously subscriptions uh
06:15 the the choice of making the same tray
06:17 in comparison of eim is both reasons is
06:21 one we want to further cut down the
06:23 budget the bomb cost of the device
06:25 because e seem require you know a more
06:28 expensive par some and same trade is
06:30 literally just a trade uh but more
06:32 importantly is that we want to sell to
06:34 multiple destinations instead of you
06:36 know we have to have a negotiation with
06:38 a a career like in us we're talking
06:41 about Verizon or T-Mobile or at and
06:43 stuff like that uh we just move too fast
06:46 and there's no way for us to sit down
06:48 you know a year to to to talk with these
06:51 guys and now they're reaching out to us
06:53 which is great uh fantastic so so so so
06:57 it's all it's all strategic decision um
07:00 I think first of all I want to correct
07:02 one thing uh I I I actually watched your
07:04 last episode talked about us yeah um
07:07 we're are not making any large language
07:09 model we we it's the large action model
07:13 is neuros symbolic it's not a large
07:15 language model U large language model
07:18 where we usually talk about his GPT 4
07:20 you know and Bard and and you know Gro
07:23 and all that uh those are based on
07:25 Transformer and they require a a
07:27 ridiculous amount of GP you on the cloud
07:30 to be able to train and and get things
07:33 right uh I don't think any startup raise
07:35 $30 million can just all of sudden make
07:38 their own large language model so for
07:40 the record we're not doing that we're
07:42 working with all the best language
07:44 models and small language models and
07:47 potentially if there's open source
07:49 language models that's in the future we
07:51 would like to basically we set up a Evo
07:55 internal evaluation to keep monitoring
07:56 the performance from all these major
07:58 vendors and we can switch on the Fly
08:00 that's how rabid o works but we are
08:02 focusing on large action model which we
08:05 we know for a fact that the language
08:07 models were the Transformer were
08:08 designed to understand language better
08:11 but it performs really really miserably
08:14 at least for now on finishing the tasks
08:17 and we don't like the way to work with
08:19 apis because API one you're betting
08:22 everyone will give you API which is not
08:24 the case it's easier for open ey to
08:26 encourage or big companies to encourage
08:28 everyone to build API for them but it's
08:30 really hard for startup to convince you
08:32 know all of a sudden you have 2,000
08:34 vendors working on your API in your
08:35 format um but even if you have all the
08:38 API uh oftentimes they don't they don't
08:41 100% replicate the full feature of the
08:44 application you know say I worked with
08:46 Uber API before back in my previous
08:48 company rentch 10 years ago their API
08:51 can do three out of 10 things that you
08:53 can do and it's really hard to convince
08:55 them say hey can you do all the things
08:57 from their app you know they have no
09:00 incentives uh so we don't like the
09:02 API that's why we were like okay hold on
09:06 let's take a step back let's think about
09:08 a universal solution let's create an AI
09:10 for a universal solution to whatever the
09:13 application is is Android app IOS app
09:15 Windows app how can we build a universal
09:18 solution uh knowing that language model
09:20 weren't designed to trigger actions so
09:22 we actually use neur symbolic to
09:24 basically start we actually start
09:26 working with data labeling companies
09:28 with started to collecting based on our
09:31 own assessment we actually assembled
09:34 test groups to have real human
09:36 interacting with different kinds of
09:37 software like uber like Spotify like all
09:39 the frequent apps so we started this
09:41 process about two and two and a half
09:42 years ago we started collecting real
09:45 human interacting with all kinds of
09:46 softwares and we we get all those things
09:49 in video recording and then we set up a
09:51 neuros symbolic algorithm which become
09:53 today's large action model that to
09:55 basically fit in all the clips into the
09:58 large action model and ask our lamb to
10:00 read the the the clips frame by frame
10:03 are you still using your personal number
10:05 for business well stop it's such a
10:07 common mistake that Founders make but
10:09 you never have to make that mistake
10:11 again because of open phone open phone
10:14 has reot every single detail of what a
10:17 modern business phone should look like
10:19 open phone makes it super easy to get a
10:21 business phone number not only for you
10:23 but for your entire team and here's the
10:24 magic it works through a gorgeous app
10:27 that works on your phone and your
10:28 desktop I can tell you open phone is
10:30 amazing because all of our sales team
10:32 and op teams use it every day why I
10:35 don't want people using their personal
10:37 number then they leave the company and
10:38 they're still getting phone calls from
10:39 our customers clients and partners no I
10:41 want all of that to be professional and
10:43 open phone is the number one rated
10:45 business phone on G2 for customer
10:47 satisfaction because it's so
10:49 professional and easy to use here's a
10:51 feature I love you can create a shared
10:52 phone number with multiple employees
10:54 fielded in calls and texts from that
10:56 number so we want to reply to our
10:59 Founders to our partners really quickly
11:01 and we don't want to miss a call we
11:02 don't want to miss a text and that's why
11:04 we use open phone and it's already
11:05 affordable starting at just $13 per user
11:08 per month but twist listeners get an
11:10 extra 20% off for the first 6 months at
11:12 open phone.com Twist if you got existing
11:15 numbers and you're paying through the
11:17 nose for some insane service you can
11:18 Port those right over to open phone at
11:21 no extra cost so here again is the offer
11:24 go to open phone.com Twist and get this
11:26 all organized get the 20% off as well
11:30 /wi so you would record me using Uber
11:34 Eats to order my Sushi or to order my
11:38 car you record enough people doing that
11:41 over time then the large action model
11:44 the lamb is understanding the pixels on
11:46 the page it knows it's an app and it
11:48 knows uh where I'm clicking in the app
11:50 so when you have the rabbit uh and
11:53 somebody talks to it and says hey order
11:56 me an Uber black and by the time I get
11:59 home I would like to have uh some sushi
12:02 uh for a family of five ready to go give
12:05 me a range of things uh including some
12:07 rolls and some vegetarian options it
12:10 would know how to do that because you've
12:12 trained it so many times watching
12:14 hundreds I don't know thousands of
12:16 interactions in a singular app how many
12:18 interactions do you need and did I
12:20 describe correctly what you're doing
12:22 with a large action model yeah yeah
12:24 you're you're describe uh described in a
12:26 correct content uh first of all we don't
12:28 learn we don't require from you we have
12:30 a test group that we assign the tasks
12:32 you know we actually work with data
12:33 liing part parties so all those CPS were
12:36 purposely connected without violating
12:38 anyone's privacy we're not like setting
12:40 up something to record on your logo we
12:41 never do that um but yeah the idea is
12:44 correct we actually had the entire paper
12:47 published on the rabbit Tex research
12:50 with the actual you know uh behind the
12:53 scenes stuff you can go there and take a
12:54 look uh at at the at at at that paper
12:58 but yes uh we collect from real human
13:01 interacting with this apps and what
13:03 happens is that if if it's very ironic
13:06 because neuros symbolic runs better on
13:08 CPU than GPU so our Cloud host is not as
13:12 crazy merely insanely compared compared
13:15 to openi compared to any large language
13:17 model our Cloud deployment is very very
13:20 reasonable uh we're not talking about
13:22 100 of millions of cash we're not even
13:24 talking about tens of millions of
13:25 dollars um we have a good enough uh GPU
13:29 cluster and we have a good enough Cloud
13:31 computer in in CPU and what happens is
13:34 that we're not collecting per request
13:38 we're just asking people to play around
13:40 you know for example the way we collect
13:42 the data so the task been deploy be like
13:45 hey you have 10 minutes on Spotify you
13:48 know try as much as you want and do as
13:50 much as you want I'm not going to tell
13:52 you you have to play this track and
13:54 click here and do that you explore free
13:57 flow for 10 minutes um and the neuros
14:00 symbolic algorithm one of the biggest
14:02 difference between the traditional RPA
14:04 for instance if you're familiar with RPA
14:06 is like basically you're recording the
14:08 screen of course but you are then
14:10 deploying an algorithm a pre-programmed
14:13 sequence to navigate your mouse your
14:15 cursor to the XY location based on the
14:17 absolute coordination okay you're
14:19 referring to robotic process ordination
14:21 exactly yeah so I'm trying to say so
14:23 people understand when you're
14:24 programming a robot you can actually
14:26 take the robot arm yes and move it to
14:29 pick something up put it into this box
14:31 pick a different object up put it into
14:33 box B right it records that it learns
14:36 that and then it can just do that over
14:37 and over again you you've literally done
14:39 the task it's almost like showing a
14:41 monkey you know how to peel a banana and
14:43 then it peels it the way you do it
14:45 monkey see monkey do basically so that's
14:47 RPA but neuros symbolic take one step
14:49 further because we're not recognizing
14:51 all those Elements by absolute
14:54 coordination of the entire ratio of the
14:56 screen we extract and Auto label some
14:59 and reasoning about elements directly
15:01 from symbolic method so that means that
15:04 it doesn't really matter if a app
15:06 completely ramped the UI got so when
15:10 Spotify redo its app and it moves
15:12 podcasts from tabs and then it puts the
15:14 tabs at the lower part or puts it in a
15:16 hamburger drop down it still knows
15:19 that's the word podcast and that's where
15:21 you find podcast inside of Spotify
15:23 exactly because the fundamental
15:25 philosophy logic is that all these
15:26 modern software are designed for human
15:28 to proceed the information and they have
15:31 to have a setting somewhere right they
15:34 have to have some symbolic thing and and
15:36 and and and text somewhere and search
15:39 bar somewhere so that's the advantage
15:41 for us in comparison with you know let's
15:43 say just build a hardware on top of
15:46 gtp4 uh so I I first want to clarify
15:49 we're not working on any L we work with
15:52 L but we also created L which is this
15:55 neuros symbolic method and so you have
15:58 some Bank of people in Manila or
16:00 offshore somewhere with these apps
16:02 playing with them every day uh that's
16:04 that's been done that's that's that's
16:06 all been done so you trained it all yeah
16:08 I've trained we trained let me ask a
16:09 question there when you when I tell
16:11 rabbit hey order me sushi family of five
16:15 you know this amount of food Etc yeah um
16:18 it knows how to use the Uber Eats app or
16:20 the door Dash app now the rabbit device
16:24 then fires off to your service somewhere
16:26 in the cloud yes this request I have
16:28 already authenticated in a web interface
16:32 right hey I authenticated my Uber Eats
16:34 and my door Dash account it knows the
16:35 sushi restaurant I like and then it
16:37 starts this ordering process and then I
16:40 guess comes back to me with like hey
16:42 just want to confirm this is what you
16:43 want and I say yes and then what does it
16:46 do it pops up an emulator of some type
16:48 in the cloud and then you have this web
16:50 emulator that has my login authenticated
16:54 already how does that work so first of
16:56 all uh let's start from beginning auth
16:58 indication right because if you think
17:00 about a device it works quite different
17:03 than the previous generation uh because
17:05 it doesn't have any software pre baked
17:07 in it doesn't have anything
17:09 pre-installed it's it's literally just a
17:12 a Ai and it's your choice what kind of
17:16 service you want to enable and it's your
17:18 choice how complicated and how advanced
17:20 you want this device to be if you just
17:23 say hey this is a cool looking iPod I
17:25 just want to use this to listen music
17:28 then you unlock the music feature choose
17:30 whatever vendors it only will do music
17:32 but in tomorrow if you want to start
17:33 ordering food you have to unlock that
17:36 feature and the login process is where
17:38 you're correct we have a web portal
17:41 that's kind of like our own mini version
17:43 of itunesi cloud if you understand that
17:45 in that sense that helps with all the
17:48 authentication settings and feature
17:49 management so you go to the website and
17:52 you basically choose whatever service
17:55 you want to unlock because again to
17:59 model Spotify YouTube music Apple music
18:03 doesn't make a difference it it is an
18:05 interface you know in fact in fact xedia
18:08 and YouTube music doesn't even make a
18:10 difference they are all interfaces so
18:12 you we give you the freedom to choose
18:14 whatever preferred Services you want to
18:16 unlock you go there and you you you
18:18 basically click the connect with Spotify
18:21 button and what happens next is that it
18:23 will redirect you to Spotify login and
18:26 we don't save your credential we don't
18:28 touch that you go to Spotify you go to
18:31 Uber you go to Door Dash you log in
18:33 through them and then we recognize oh
18:36 this account is being connected with
18:37 rabid OS and then what happens is that
18:41 on on our Cloud we have a very very uh
18:44 creative structure and Innovative
18:46 structure that we have considered we
18:48 have a super host we have a superhost
18:51 computer that when Jason talks to his
18:53 rabbit R1 about ordering a a a hamburger
18:58 what's going on is that we first see if
19:00 Jason if Jason's Lo logging with door
19:03 Dash or loging with Uber Eats right and
19:05 we saw okay Jason choose door Dash and
19:08 on on a super host a lamb is interacting
19:12 virtually with a door Dash app or
19:15 website a interface and you don't see
19:18 all of that because all of a sudden this
19:20 is done all at once because it's it's Ai
19:22 and then we will render rabbit themed UI
19:26 on your device get your result got it so
19:30 you're not you're not directly
19:32 interacting with the with the host as
19:34 well you're just you're just talking to
19:36 it the intention goes l l understand oh
19:39 you want to do this then lamb do that in
19:41 the virtual environment and render the
19:44 result to your device that's how it
19:45 works and so uh do you need permission
19:48 from Spotify to do this or you could do
19:51 this with any app um once you've trained
19:53 it on the data I mean it's nice to get
19:55 the permission but it sounds like you
19:56 could do this without it it is it is Def
19:59 definitely nice to get these guys
20:01 permission or I wouldn't say permission
20:03 I think we should we should develop a
20:04 better business model because this is to
20:07 me is like okay you know remember the
20:09 early days Steve Jobs called Sony be
20:11 like from tomorrow because we have this
20:12 device from Tomorrow 99 cents per track
20:15 you know I I feel like this kind of like
20:17 a similar situation because we are not
20:19 first of all we are not creating new
20:21 users right we're not creating spam
20:22 users we're not creating prepaid users
20:25 you are you you're Jason on your
20:28 authentication to their interface to to
20:33 service when whenever the way that you
20:36 you as as if you're using on the phone
20:38 or on the TV so uh we take a good amount
20:43 of time to studying the terms and
20:45 agreements and trying to understand this
20:49 that unless they shut down their
20:51 interface which is not going to happen
20:54 because we're we're not violating
20:55 anything and we're not even create fake
20:58 user spam there's a lot of ways can we
20:59 can do that right you know I'm not I'm
21:02 not sure if you saw the the the uh
21:04 nothing uh Sunbird IM message bypass
21:08 through Android you know there's a lot
21:11 of weird way that that you can you can
21:13 set it up but we're not setting up any
21:15 of that you are you yeah and it makes a
21:19 lot of sense because um for them this is
21:22 just their user interfacing with their
21:24 app through essentially a voice
21:27 interface with some confirmations on the
21:30 screen is explain to the users who are
21:32 listening like as a user I order that
21:35 hamburger through door Dash it says Five
21:38 Guys or does it give me like a choice
21:40 say like hey you've ordered from Five
21:42 Guys and Shake Shack before where would
21:44 you like to order it would you like your
21:45 standard order would you like this would
21:47 you like to repeat in order how how um
21:50 at this point in the 1.0 of what you've
21:52 released right uh or the point one of
21:55 what You released how does that
21:58 you know what if the restaurant is
22:00 taking two hours to deliver or they've
22:02 stopped delivering and you get the idea
22:04 yeah that's a great question uh I think
22:06 that's something that we're really
22:08 focusing on uh because this is a new
22:09 problem to solve right so uh one part we
22:13 know for a fact that we will just
22:15 directly trigger the service if it's a
22:16 direct action you know if you want to
22:18 listen get lucky that's it you know play
22:20 Get Lucky you want to beat to to be
22:22 there right away that's that's easy um
22:24 but we also identifi there's couple of
22:26 scenarios I guess one thing the reason
22:29 that I didn't show on my keynote with
22:31 the BAU on the Travel I booked the
22:33 entire travel plan through xedia uh some
22:36 serious inary stuff or some serious
22:39 stuff relates to you know I guess a lot
22:41 of deposit we basically want to create a
22:46 way for you to check back over and over
22:48 again confirm confirm confirm exactly
22:51 yeah you're everyone's meing about that
22:53 I'm meing about that myself so so so so
22:56 that's why I think you know we need to
22:57 separate categories there are certain
22:59 categories you want to get a uber to
23:00 your home right and and and and that's
23:02 it that's simple enough we we get it but
23:04 if it's more complex te text in the
23:07 future you should be able to see a copy
23:09 of what exactly going on with the actual
23:11 paperwork on your rabbit hole web portal
23:14 that's how we want to design it uh so
23:15 you always get a get a portal to check
23:18 you know more legitimate stuff like for
23:19 for for backup and stuff and you're no
23:21 taking you know your your summary of the
23:23 meetings all those stuff will be synced
23:25 to to the white portal and that's why I
23:27 call it more like our mini version of
23:29 iCloud but on the device because it's so
23:32 tiny um I I think a lot of people
23:34 because I have ridiculous small hand I
23:37 think a lot of people get wrong about
23:38 the size but so it's half an iPhone it's
23:41 precisely half an iPhone oh it's
23:44 precisely half an iPhone you're right
23:45 it's same thickness of an iPhone 15
23:47 looks like it's a ma it's Max model it's
23:50 bigger it's a bigger model but you know
23:52 I'm putting this next to it yeah uh
23:55 because it's such a small device uh we
23:57 actually had a hard debate whether we
23:59 should just cancel the screen at all
24:01 right because oh right uh that's that's
24:03 a make it like the iPod Shuffle where
24:06 just you know like really really tiny so
24:08 it just becomes like a tiny recorder or
24:11 whatever you can make it into a watch
24:12 you can make it into any number of
24:13 things exactly take me around the form
24:15 factor you used a really great design
24:18 firm to design this for you maybe you
24:19 could talk a little bit about who you
24:22 use to design it and the inspiration
24:24 behind the device because it it looks
24:26 modern and retro at the same time
24:28 yeah so it's iconic yeah uh I I I I I
24:32 don't think I'm confident to say it's
24:34 already iconic but I saw a lot of people
24:36 are making free works for us for the
24:37 case and for everything I think that's a
24:39 good start you know at least um it's a
24:42 great story between me and teenage
24:44 engineering I respect them they're my
24:46 hero company I started collec vintage
24:48 synthesizer about 15 years ago uh so
24:51 when they when I saw the op1 the
24:53 portable synthesizer they introduce I
24:55 immediately bought it uh but it's it's
24:58 very very expensive to me couple of
25:00 couple months to get the fings together
25:02 in in the beginning but uh first of all
25:04 teenage engineering is not a design
25:06 design form I have to make clear of that
25:09 they're not they're not a firm that's
25:10 actively looking for collaboration and
25:12 taking design fees and do that uh they
25:15 are a very focused you know music
25:17 technology consumer firm that is been
25:20 over it's been there for almost like
25:23 what 16 almost 20 years 15 16 20 years
25:27 um it's a such amazing talents of groups
25:31 of people that I I told my team when I
25:33 was in Raven I'm like hey if these guys
25:35 gave me a call and asked me to go work
25:37 for them I I'd go you know and then fast
25:40 forward uh when I was working on Raven
25:42 Hardware project which Raven H and Raven
25:46 2017 um I'm like okay hold on a second
25:50 maybe I can convince them to work with
25:52 me instead of I work for them because I
25:56 want to pick the best and they're in my
25:58 mind the best so I literally just called
26:00 email reach out and 3 days later I I sit
26:04 in Stockholm in at their office and then
26:08 yasper Works in yasper is a CEO and Co
26:10 co-founder of teenage and and we both
26:14 had an iPad and pencil and we started to
26:18 you know start to draw things and during
26:20 that process he asked me about oh what's
26:22 your what's your favorite artist what's
26:24 your favorite car design you know and
26:26 couple questions WIP chat questions
26:28 right and it's quite surprising to me
26:31 that we have the exact same taste on
26:35 almost everything like all the question
26:37 he asked he showed me his vinyl
26:39 collection I showed him my vinyl
26:41 collection it was exactly same band in
26:44 the same sequence you know and and he
26:47 said oh I I I love this and I love that
26:50 and we talk about a couple of the things
26:52 uh so so they're like okay let's do this
26:55 um the whole process is is like I imic
26:58 you know I I don't think I ever talked
27:00 about this on any Media or social
27:03 network but we actually set it up a
27:05 secret Instagram account we have no
27:08 email communication no phone calls no
27:10 nothing we just start we just post
27:13 sketches on the visuals on Instagram and
27:15 then we I we just like he likes we leave
27:18 comments that's that's it so we did that
27:20 whoa we did that we did that Raven h u
27:23 project no slack connect room no
27:25 iMessage group a secret Instagram
27:28 account the teenage group post to it you
27:32 post yeah comments Etc is it a public
27:35 account or is it a private account it's
27:37 a private private account what an
27:38 interesting way to do sketches yeah um I
27:42 think I think because it's so long maybe
27:43 later later down the road I I might be
27:45 able to share some of the early early
27:47 works but but you know all I want to say
27:49 is it's it's two group of intuitive
27:51 people that we meet each other and then
27:54 there's a strong Synergy and then uh
27:56 fast forward 2018 I officially join as a
28:00 board of director to their board and
28:02 then I I you know I have a better
28:05 exposure of of the company and uh you
28:08 know I understand they need to be
28:10 extremely focused on their current road
28:11 map they have a lot of things to deliver
28:14 uh but I'm start seeing that a a wider
28:17 spectrum of recognization towards their
28:19 industrial design in the past you know
28:20 three four years which I'm super happy
28:23 um and then uh Teenage also as a C
28:27 funding design partner towards another
28:30 company called pays nothing I'm I'm not
28:33 sure if you heard of company called
28:35 nothing they make phones and oh yes the
28:37 nothing phone yes yeah yeah we were
28:39 behind nothing ah so uh we also we also
28:44 are co-founders of nothing as teenage as
28:46 a whole company so we help nothing set
28:48 up the initial whole design language and
28:51 everything uh and then uh when we
28:58 it's the opposite way we want to create
29:01 cool which is quite funny because of
29:04 course the first thing we're we're going
29:05 to look at is who's out there right we
29:08 saw some strong competitors Humane AIP X
29:11 Apple folks huge respect to the team um
29:14 but then I talk I I kind of like
29:17 persuaded myself I'm like hey you're
29:19 offering this completely new generation
29:22 of interaction software wise and it's
29:25 just to me it's just too risky offer a
29:28 Sci-Fi Gadget which which no one knows
29:31 how to use it so in my theory Hardware
29:35 is never a choice by preference you
29:38 don't build a hardware because you want
29:40 to build a cool Hardware most of the
29:42 time if you do things like that it it
29:44 will fa miserably a lot of good cases I
29:47 learned and I experienced is that you
29:49 have a software that's so good that you
29:52 want the dedicated Hardware to make it
29:53 better it's always around the software
29:55 it's always around the you know what's
29:57 inside and then if you have a software
30:00 that's very very edgy very very new you
30:03 want to drisk on a hardware at least for
30:06 the first generation got it so so to me
30:09 R1 is a result of choose Hardware by
30:12 necessity not by preference if I want to
30:15 do Hardware I probably want to do a lot
30:16 of other phone Factor like you said
30:18 right how about a fancy glass how about
30:21 whatever that is so it so the first
30:24 thing we we realize is that we want to
30:26 set up a good position so that we can
30:29 compete with all these big companies and
30:31 upcoming competitors um but as well as
30:34 offering something that can resonates in
30:37 your culture and in your memories and in
30:40 your existing workflows that you don't
30:43 necessarily needs a menu to understand
30:45 how to use it it's hard to balance
30:48 hiring top tier developers and keeping
30:51 your burn rate under control but these
30:54 days I see a ton of Founders
30:55 successfully doing this by hiring remote
30:58 Talent so let me tell you about scalable
31:01 path it's a software staffing company
31:03 that can help you build an awesome
31:05 remote developer team and the right
31:07 developer isn't just a list of technical
31:09 skills we all know that it's about their
31:11 personality it's about their work ethic
31:13 their motivation and their fit within
31:15 your team and scalable path knows this
31:18 so here's what they do their team will
31:20 get to know your vision they're going to
31:22 get to know your needs and then they're
31:23 going to develop technical challenges
31:25 tailored to the roles your hiring for
31:28 and these challenges are conducted live
31:30 and on video so there's no gaming of the
31:32 system you're going to get great people
31:35 they also evaluate each candidate's soft
31:37 skills like communication attitude and
31:39 work style scalable path has completed
31:42 more than 300 projects for their clients
31:44 and they have a network of 30,000
31:46 developers they've been doing this for
31:48 over a decade they know what they're
31:50 doing so you're going to be in great
31:51 hands here's the best part twist
31:53 listeners get 20% off their first month
31:56 if you're ready to scale your Dev team
31:58 and your business check out scalable
32:00 path.com twist once again that domain
32:03 name scalable path.com twist for 20% off
32:07 you've been using this product for a
32:08 little while you said or at least Pi
32:11 version when do you feel yourself
32:13 reaching for your iPhone when do you
32:15 feel yourself reaching for the rabbit
32:17 when do you hit frustration points with
32:19 either device and and how does that go
32:21 into your thinking about does this at
32:24 some point replace me bringing my phone
32:26 with me yeah uh and I have to maybe
32:29 think about the hardware capacity of
32:31 this device or maybe there's you know
32:33 this this rabbit but there's two other
32:36 versions and do do did you ever leave
32:38 home without your iPhone and just take
32:39 your rabbit talk to me about that yeah
32:41 yeah so so first of all any single
32:44 direct actions I I'm I'm leaning towards
32:46 rabbit now why why because when I start
32:51 trying to accomplish that task it's a
32:53 speed of thoughts and I have that
32:55 already and I feel like almost quicker
32:59 and faster more intuitive than find that
33:01 analog button without even looking at it
33:04 and just talk and the AI is accurate and
33:06 good enough lamb is fast enough to get
33:08 things done so I'll give you some quick
33:10 quick reference I'm always on multiple
33:13 screens I'm always on a lot of things
33:16 daily when I'm working and maybe talking
33:18 to other guys if I have something that I
33:20 don't know this is definitely faster
33:24 just forget about a lamb just searchwise
33:27 this is way faster then I have to set up
33:30 a new tab or go to the Chrome and start
33:33 typing because you have the action
33:35 button it's kind of like a CB radio you
33:38 press the button and you say Define the
33:41 word iconic clastic or Define the word
33:45 uh Donny Brook you don't know the word
33:47 Donnie Brook you say Define the word
33:48 Donnie Brook I don't have to think
33:50 command tab go to Google define Donnie
33:53 Brooke Etc six or seven steps yeah well
33:55 well not even that some very Advanced
33:57 you know uh numbers this gives you
34:00 faster because I I give you an actual
34:02 case I was having a meeting with our one
34:04 of our existing investors they were ask
34:06 about you know performance in comparison
34:08 you know on sales and and some numbers
34:11 uh from other company I generally don't
34:12 know I don't know what the what the
34:14 revenue from from you know a company
34:17 last year and if you think about I
34:20 Google there's there's 200 tabs which
34:24 one is accurate which one is correct so
34:26 that's we set up the Strategic
34:28 partnership with practicly just to
34:29 enhance that part so that's search alone
34:32 another thing is music I can tell you
34:35 you will love this just because it can
34:37 play music it's a second it's it's a
34:39 second iteration of the classic iPod
34:42 with with with the same level or even
34:45 simpler level of of control of this uh I
34:48 had very my own own privilege to have a
34:51 phone call with Tony
34:53 Fadel actually yesterday so we we were
34:56 on the call uh for for more than three
34:59 hours and and I think you know just
35:02 music play alone I that's definitely my
35:05 number one in the past eight months
35:08 other than search search is probably 70%
35:10 of the music uh but I think a lot of
35:12 people still don't get it like I I start
35:15 posting stuff on my Twitter there's one
35:17 scenario I use the vision so basically I
35:19 double tap the camera rotates and points
35:21 to whatever you want to point I I use
35:24 the vision to look at a Discord cuz grab
35:27 it now in the past three days we have
35:28 5,000 members and I'm I'm the customer
35:31 service guy there I want to listen and
35:33 learn from from from my as a Founder to
35:36 listen to the customer support line
35:37 that's where the truth is I I reply
35:40 almost almost as much as I can but I
35:42 start getting lost right because the
35:44 message is too much and I was actually
35:46 on a zoom call with some other guy and
35:50 then I literally just pointed the camera
35:52 and and and and said what are people
35:55 talking here oh a lot of people don't
35:58 get it thato a lot of people are
35:59 thinking oh you have eyes you are not
36:01 blind and this is stupid and why are you
36:03 doing this but I'm actually on something
36:05 else and and I don't have time to scroll
36:08 my mouse for 50 pages to and even though
36:12 the current generation of gp4 vision and
36:15 our own Vision model needs to be faster
36:17 actually it needs to be 10 times faster
36:19 by shipping which I'm pushing it but
36:21 after four or five seconds it says oh
36:24 it's here's a conclusion people are
36:25 talking about how rabid R1 can
36:27 potentially waste work it gives me the
36:30 report and which you would normally you
36:32 might have a college educated person who
36:35 you say summarize what's happening in
36:37 the customer support line exactly for me
36:39 every day or you might point an llm at
36:41 it and have it do a report here you just
36:43 take a quick picture yeah I I too was
36:45 having this problem I was skiing this
36:47 weekend and I kept trying to change my
36:49 music using Siri right and it is so
36:52 painful to just uh lower the volume and
36:56 or change the track or play a different
36:58 playlist you're talking about like five
37:01 to 10 seconds of Siri getting it wrong
37:04 and then you got to do it again and then
37:06 your music cuts out the idea that you
37:08 could press this button and then very
37:09 quickly at the response time it I think
37:12 all of those smart speaker systems which
37:15 The Raven you mentioned a couple of
37:17 times that you sold to BU do that was
37:19 your first company I think or second
37:20 company yeah first company you must have
37:22 learned a lot about just response time
37:25 to voice command it's just so terrible
37:29 to wait that four or five seconds to for
37:31 it to wake up and to understand what
37:33 you're doing right and that's something
37:35 you're trying to solve is that first
37:37 five what do you call that the 5c delay
37:39 yeah so so so right now if you break it
37:42 down uh we the delay came from two parts
37:46 the large action model again is fast if
37:49 you if you go check the the demo I
37:51 released on the Twitter it's fast play a
37:53 song is instant uh if you ask some
37:56 random generic questions we we have a
37:58 technology just basically make the
38:00 streaming tokens with with L make it
38:02 very fast if you ask what's the nature
38:04 of reality what's the difference between
38:06 oranges and M oranges anything that does
38:09 not require up to-date information where
38:10 are 500 milliseconds anything search up
38:15 information that starts to slow down if
38:17 you if you try open AI in The First
38:19 beginning I haven't tried uh recently I
38:21 think recently improved again but you
38:24 we're talking about two to three seconds
38:25 instead in of 500 milliseconds um but I
38:29 F I figure the visual part is where the
38:31 most delay came from uh we're talking
38:34 about summer around you know 8 to 10
38:36 seconds um but that's not us that's the
38:39 fastest in the industry right now and
38:41 we're working restlessly trying to
38:44 figure out a way to to further cut it
38:46 down but but really right now the delay
38:48 is really just search up toate
38:50 information and vision that's it um if
38:53 you just trigger lamb stuff that's fast
38:55 that's that's no burn that's very very
38:57 fast um but frustrations let me tell you
39:00 the frustrations so yes number one I
39:03 don't want to carry two devices no one
39:04 wants to carry two devices uh I'm not
39:07 trying to convince people that this is
39:08 just sexy so that you know maybe you
39:10 want to carry no uh what we can do for
39:13 now we know it cannot be an app so same
39:17 Footprints prints same footprint as your
39:20 iPhone it's actually very very light
39:23 it's only 110 gram so what's 10 gram go
39:27 to your fridge grab two eggs uncooked
39:29 eggs that's it and it it's light enough
39:32 so that every single time I I put in my
39:34 pocket forgot about it and then I
39:36 appreciate myself for the design of the
39:38 analog push to talk button because I
39:40 don't really need to take it out and and
39:43 and look at the screen I reach out to my
39:44 pocket and I feel that button that's
39:48 that's that's actually how I use it most
39:49 of the time you know I just connect to
39:51 my airpods or whatever Bluetooth devices
39:55 or car systems and I don't really need
39:57 to look at it I I reach and and feel but
40:00 still there's a lot of things I would
40:02 rather go to the phone at least for now
40:04 first of all Heavy social features you
40:07 know this was not designed at least for
40:09 this generation this was not designed
40:10 for connecting all your friends and
40:13 chitchat and figure out what's going on
40:15 uh this is more focused on task solving
40:17 um so that part unfortunately I have to
40:20 go back to the phone uh another part is
40:23 for your group chats if you want to get
40:24 your group Chats on Signal WhatsApp
40:26 whatever your jam is iMessage it's not
40:28 going to replace that just yet
40:30 but we could I mean it's it got it got a
40:34 tray it got a same card yeah it is it is
40:37 a phone I mean come on it is it is a
40:39 phone we we're not trying to be a phone
40:41 but it is capable of doing all the
40:44 features of your phone uh another thing
40:46 is which is I'm not sure if it's just
40:49 the the fact or if it's just my angle
40:51 but phone is actually a Content
40:54 consuming device right like if you think
40:56 about about the highest gross apps on on
40:57 the app store right now it's it's
40:59 Netflix and stuff it's it's Tik Tok
41:01 Netflix Instagram and all that stuff um
41:04 that you know I mean come on iPhone has
41:07 a better screen yeah big screen you big
41:09 screen better much better screen I I
41:11 have to say that yeah I that's why you
41:13 know when I start thinking about laser
41:15 projectors I'm like oh I'm not sure you
41:17 know like CU I showed the I I saw I saw
41:20 the Humane demo that they they they do
41:22 the the gestures to send a message I'm
41:24 like no I probably wouldn't that here
41:27 you know no it seems crazy like forcing
41:29 it right now startups have to do more
41:32 with less we all know that and that
41:34 means increasing your product velocity
41:36 while maintaining or even lowering your
41:38 cost now don't forget product velocity
41:40 is how startups beat incumbit so here's
41:43 the great news AI is going to help you
41:46 do that so let me tell you about wizard
41:48 it's spelled UI z a Rd it's an AI
41:51 powered Suite of UI and ux design tools
41:55 with wizard you can generate your app or
41:57 web designs from Simple Text prompts you
41:59 can then iterate on these designs with
42:01 an AI assistant and then you hand off
42:03 your completed designs as react or css
42:06 code Wizard's text to UI mockup tool is
42:08 called autod designer and it's really
42:10 cool if you're watching you can see it
42:12 on the screen right now here's the brass
42:14 tax wizard is going to help you go from
42:16 idea to mockup in minutes so if you're
42:19 creating a product from scratch this is
42:21 going to save you so much time start
42:23 building products today faster with 25%
42:26 off wizard pro at wizard. twist that's
42:29 UI D.I twist for 25% off stop wasting
42:35 time and start shipping faster the voice
42:39 recognition is getting better and better
42:41 and these llms seem to correct errors in
42:45 real time when it's transcribing so talk
42:48 a little bit about that the ability to
42:50 transcribe uh and translate voice to
42:53 text yeah really accurately and how
42:56 that's changed things because when you
42:57 were running Raven the speaker right it
43:00 was kind of janky right wasn't it
43:02 terrible voice to text yeah when I
43:05 started Raven that was 2013 uh I was
43:08 actually working with SRI a couple of
43:10 folks in Sri Stanford institution that
43:13 that institution later uh started new on
43:16 and new on incubated the so it's it's a
43:19 long history um but I remember very
43:21 clearly the first time I paid a visit to
43:23 Sri the dictation was like roughly at 73
43:27 74% native English which is very very
43:30 bad uh but it it went up very very
43:33 quickly another problem is not the
43:35 dictation the problem is intend
43:37 understanding and then again we're
43:38 talking about pre- Transformer era and
43:41 then back at the Raven we work really
43:43 hard on natural language processing or
43:44 NP uh that was the best technology
43:47 before uh GPT can understand you um I
43:52 guess the idea is not because we didn't
43:53 thought of we could design an algorithm
43:56 such way it's just back then we don't
43:57 have strong enough computition power
43:59 like gpus like Nvidia gpus to be able to
44:02 run that um so if you talk about intent
44:05 recognization it's it's really really
44:07 painful it's heavily human consuming
44:11 because essentially all these smart
44:12 speakers I created and Alexa and every
44:15 previous generation when we first
44:17 started create that it's like you're
44:20 assembling a menu to this quote unquote
44:23 Ai and you're basically hard that be
44:26 like hey here's 70 ways that you talk to
44:29 the speaker that means you want to
44:32 track Like I want to listen to play to
44:35 uh get it on you know like there's
44:37 there's a cluster of of sentences you
44:40 can you can describe to what the same
44:42 intention and NLP is for us to figure
44:45 out that to assemble that menu so a lot
44:48 of things were really really not ideal
44:50 and I totally get a skeptism towards R1
44:54 be like hey it's a voice first device as
44:57 as soon as I saw that that's no go I get
45:00 it because we are all the same
45:02 generation of consumer that still share
45:05 this PTSD from early days of the very
45:07 poly exactly Siri has been so
45:10 frustrating for people and Alexa was
45:13 better but not available on your phone
45:15 to the same extent Google was yeah
45:18 assistant never really caught on yeah
45:20 and so we just assume it sucks and
45:22 exactly the truth is it's much better
45:25 now even the chat gp4 app yeah um I took
45:29 my um you know the iPhone 15 now has the
45:31 action button which is similar to what
45:33 you're doing on rabbit I connected that
45:35 to chat GPT 4's voice interface right so
45:38 I actually when I press that it goes to
45:40 that uh what are they conversational I
45:42 guess mode yeah in the app and it's
45:45 quite nice it's a little bit slow um
45:47 it's like the hardware seems to be
45:49 working against it yeah whereas in yours
45:51 it's going to work quickly when do you
45:53 know you'll have made a product that is
45:57 ready to challenge the phone in your
46:00 pocket and you will win the which device
46:04 I go back home for if you forget your
46:06 phone R you go back for it if you forget
46:08 your wallet you don't go back for it
46:10 you're like I forgot my wallet I got
46:11 some payment options on my phone when
46:13 will your device you know if you leave
46:15 home without it you got to go back and
46:17 get it when will you hit that Milestone
46:18 of I gotta turn the car around and go
46:20 back and get my rabbit yeah um uh first
46:24 of all it's it's such a
46:26 ambitious task to challenge I say uh we
46:30 never set ourself at least mentality
46:32 wise we never set it up ourselves be
46:34 like oh the whole purpose of rabbit is
46:35 to is to do this is to kill iPhone
46:38 whatever that is um but on the other
46:42 hand we understand deeply especially
46:44 given by my experience you know my
46:45 company so to BU do and we were at a
46:47 very exclusive working relationship that
46:50 Raven still keeps Ren but I have seen
46:52 enough uh not only bu do by Microsoft a
46:55 the other companies uh how big company
46:58 works you know I I I think we're at the
47:00 point that we don't think we're
47:02 confident to challenge right now but we
47:05 don't want to wait either like like yes
47:09 you can be the user right I I post this
47:11 on Twitter I said okay you would you
47:14 want rather be the wave itself or the
47:16 best server in town or that floating
47:18 body in the middle of the oce right
47:20 because you can only be three of those
47:22 things so I think what we are is that
47:25 we're not saying oh we are so ambitious
47:27 and so dilutional at the same time that
47:29 I have a very clear master plan this is
47:32 going to work that's just not the nature
47:34 of startup I mean you've seen enough of
47:36 startup things you know you never think
47:38 like that way you're like okay I want to
47:39 do it I want to do now that's how it
47:41 works but I don't think that one thing I
47:44 have strong confidence is that for for
47:46 all these current generation of
47:48 application uh application based
47:50 operating system it's impossible for
47:53 them to improve because technology is
47:56 not is always not I mean fundamental
47:59 operating uh OS shifts it's not it's
48:02 never be an improvement it's always
48:03 reamp it's always Revolution and the
48:05 problem is not about technology I think
48:07 a lot of Engineers having a different
48:10 opinion towards engineer perspective be
48:11 like oh you know what they just do this
48:13 and that and Sir become R1 and Sir
48:16 become rabid o it's not like that the
48:19 problem is their incentives behind how
48:21 they make money you know they set it up
48:24 this entire app ation store with
48:27 billions of developers billions of apps
48:30 and all of a sudden they said you know
48:31 what guys no apps how does that even
48:34 work like I don't see a very smooth
48:36 transition how that can happen so
48:38 quickly to the point that we think we
48:41 should wait a bit you know and lower
48:42 further lower our risk we cannot do that
48:45 we just start right away um but I think
48:48 consider R1 is iPod you know it's it's
48:52 it's a good phone factor for a certain
48:55 Niche amount of requests we're talking
48:57 about slightly wider request than iPod
49:00 because iPod really just to replace
49:01 Walkman right but but with in the iPod
49:04 era you still have your Blackberry phone
49:07 I do that yes I I have my Blackberry 800
49:10 8900 I have my iPod um but two back
49:13 pockets you put one in each back pocket
49:15 right exactly and and it kind of like
49:17 fix in the shape you know especially if
49:19 you're wearing Jam you don't even want
49:20 to swat you just you just you know keep
49:22 it there but but you know we have R1
49:25 this is our first attempt or approach
49:28 towards a future structure you know the
49:31 future how software will work with human
49:34 how you interact with software um but I
49:37 think maybe in next a year and a half
49:41 and two years we'll have a better answer
49:44 yeah uh and for now you're going to go
49:46 after you know people who are early
49:49 adopters the on guard people want to try
49:52 it people who want to invest in the time
49:55 and give you really good ideas have any
49:57 of the app developers come to you
49:58 directly and said can I just you know
50:02 make stuff directly for your lamb yeah
50:04 your large action model can we have an
50:06 interface where you know you have your
50:08 language Mo action model you expose us
50:11 to it and Uber engineers and your
50:14 engineers can be in the same yeah space
50:18 you know you know collaboration space
50:20 making it uh operate flawlessly it's
50:24 actually happening from both sides Jason
50:26 it's it's one side yes you know you saw
50:28 the praity deal that's that's a new way
50:31 we literally handshake on Twitter we we
50:33 have no no pre-communication you know we
50:35 just decid to do it and now we offer uh
50:38 the first 100,000 units with $200 credit
50:42 of praxity if you think about the device
50:44 itself which is $99 so it's happening
50:46 from from this side uh company to
50:48 company but more importantly you know I
50:51 think the most underrated feature of
50:53 rabid OS which is kind of EXP
50:55 experimental right now but we're
50:57 confident to roll it out at least for
50:59 the beta version very soon after people
51:01 start receiving their device is a teach
51:03 mode and how the teach mode is works is
51:05 that you actually explained that uh
51:07 before is that as if monkey monkey like
51:11 you go to this web portal right the
51:14 rabbit hole you turn on teach mode you
51:16 decide today I'm going to teach you
51:17 something and it's like if you're
51:19 teaching your 8-year-old children you'll
51:22 be like hey come to Daddy sit right here
51:23 and watch exactly what I'm doing and
51:26 from it so so if you think about neuros
51:28 symbolic same metaphor and there's a
51:30 psychological Factor behind it is that
51:33 Jason if you if I ask you to teach me
51:35 all on skateboard would you rather do
51:37 that or not yeah there's two decisions
51:41 is one you want to do it or two even you
51:45 want to do it you probably wouldn't
51:46 willing to but more importantly if you
51:48 don't know how to do Alli at all you
51:50 would just say no right cor correct so
51:53 yeah you go to the teach mode as user
51:56 you know exactly you're gonna teach
51:58 right you're not going to teach how to
52:00 use a skateboard if you've never been on
52:01 it or skiing but if you said Hey how do
52:03 I carve skiing and if I opt into that
52:06 it's going to be good training data is
52:07 your point exactly so if there's an app
52:10 that people aren't using like let's say
52:12 you haven't trained Twitter yet I could
52:15 say hey let me train Twitter of I like
52:17 to go to my replies I like to go to my
52:19 mentions and I like to read these out
52:21 loud please read me five of them very
52:24 quickly but I need you to read me the
52:26 date I just want you to read the content
52:28 of it and just read me the first 200
52:30 characters not the whole thing exactly
52:32 you already using the teach mode that's
52:34 a powerfulness of it so the
52:36 psychological factor is you go there
52:38 teach under the impression of you know
52:40 exactly what you're doing so the data
52:42 you connect the the the you uh of course
52:45 we don't record on your local device we
52:47 go we bootstrap the virtual machine and
52:48 virtual machine is like old technology
52:50 so you go to Virtual Machine choose
52:52 whatever platform you want to teach on
52:53 and you go teach and that recording is
52:56 very very clean so that the Rabbid o can
52:59 get one shot in just one shot so you're
53:02 letting the entire Community teach the
53:04 rabbit OS to build your own rabbits
53:07 amazing that's how we make money by the
53:09 way ah I'm s your question got it so I
53:13 could make a rabbit which would be my
53:15 way of interpreting say Tik Tok or
53:18 social media app like we talked about
53:20 and I could sell those I could sell my
53:22 own rabbits to other users yeah so so
53:25 Reinventing any business model it's
53:27 classic app Apple App Store or Google
53:29 Play Store um logic but but we're not uh
53:33 we're not worry about making money for
53:34 now because making Mone rais 10 million
53:36 from venod congrats on that he believes
53:39 in it and how big's the team like how
53:41 big of a team do you need to make a
53:42 hardware device like this and right you
53:44 know build this stuff out because we see
53:46 a lot of people doing more with less I'm
53:48 curious yeah so uh we had uh full head
53:51 count roughly around 17 people to date
53:54 but we hiring um but I'm also working
53:57 with good odm OEM I had resources over
54:00 the past Raven career so it's the same
54:03 team plus you know design Arma Team Ag
54:05 helps us us really a lot um but we're
54:09 hiring yeah we we we're working on the
54:12 future version of lamb so consider this
54:14 is lamb one we have a lamb two uh in
54:17 design uh which that's going to be much
54:19 more interesting next year uh with with
54:22 different uh phone factors but uh the
54:25 general idea is that I have worked from
54:28 my my my past companies that I know that
54:31 if you aiming to make money on day one
54:35 you're probably not going to be a
54:36 consumer hardware company anyway you you
54:39 become like a more like a how do I put
54:42 it like Oracle well you seem to have
54:45 been very inspired by the iPod yes and
54:49 if you think about the iPod my favorite
54:51 iPod was the clip on I don't know if you
54:53 remember of the second generation of the
54:56 yes oh I mean that was I believe yeah
54:59 like Peak Apple Insanity it was a clip
55:02 yes that you could it was almost like a
55:04 tie clip and you could use it as a tie
55:05 clip with a shorter head Airbus with a
55:08 shorter headphone cable yes have it have
55:10 change that too it's a loop yeah I
55:12 remember that I have that it was a short
55:14 headphone cable because they knew you
55:15 were going to put it above your waist
55:17 you didn't need to like run it to your
55:19 pocket but it also had physical controls
55:21 on it yes so it was very elegant in that
55:24 way but it you know for other people who
55:27 love to have a large collection you know
55:30 the the classic iPod the fat one you
55:32 know the original iPods you know those
55:35 were amazing just in terms of having a
55:37 flywheel on it with that physicalness to
55:39 it right and you have that you have that
55:41 scroll wheel what does the scroll wheel
55:42 do on this exactly uh we're we're
55:45 working on different ideas uh that I
55:48 couldn't share much but I can share one
55:50 idea we're working on yeah this is how
55:52 you're gonna this is how you're going to
55:55 ah so your passcode will be like a a
55:59 lock when you're in high school two
56:00 turns to the left two turns to the right
56:03 Ah that's super clever with with haptic
56:06 CPS on it with haptic ah so I go one two
56:08 three push in down one two push in up
56:11 push in boom 3 two one whatever it is
56:14 that's super clever and no one's ever
56:16 gonna know that right because you don't
56:17 it's it's not even it's not even a text
56:21 or anything it's just that's sick yeah
56:23 and so there could be different sizes of
56:25 this you said like you don't actually
56:26 need the screen um do you need to make
56:28 your own responsive headphones or you
56:31 use Bluetooth because one of the things
56:33 I have a problem with is you know I have
56:36 my airpod third generation it work so
56:38 elegant but I prefer uh the pixel buds I
56:42 don't know if you use pixel buds but
56:43 they're flush flat they're much more
56:44 elegant device they work better with a
56:47 ski helmet work better if you're
56:48 sleeping and you want to listen to a
56:49 podcast before you go to bed or an
56:50 audiobook like I do yeah how do you
56:52 think about that because if it was a
56:55 unique headphone man would there be
56:58 Hardware gains or user interface gains
57:01 that you could achieve so definitely if
57:04 you if you customize from kernel
57:05 perspective you can further lower down
57:07 the latency the latency is a issue
57:09 especially if you if you try in car
57:11 system you know that's that's why all
57:13 the car series doesn't work because car
57:16 system latency is very very outdated so
57:18 it has a very long latency but right now
57:21 it's Bluetooth 5.0 which is which is by
57:24 default work with any different default
57:27 airp and third party manufacturer I'm
57:31 more interested in maybe we can do a
57:33 rabbit airpods but it's fully
57:38 Standalone that's fascinating so but you
57:41 put them in and it's your version of the
57:42 rabbit I just talk to it and I
57:44 authenticate it so I just leave home
57:46 with my air my rabbit airpods yeah and
57:48 say hey ordered me an Uber it says okay
57:50 Uber black or Uber X right you say Uber
57:52 X I'm alone or Uber you know SUV and
57:55 it's okay yeah boom we have oh wow
57:57 that's sick and that that challenges
57:59 came from one you you're waiting for a
58:02 smaller chip with more power because
58:05 it's little room to wiggle with you know
58:07 if I do a airpod uh Max I can I can do
58:11 it now but if you want to do it right uh
58:14 there's some challenges the Bas station
58:16 your your case would have to have a SIM
58:19 card or eim or whatever battery life
58:22 starts to come into play here yeah a lot
58:24 of different issues I mean that's a sick
58:26 idea though if you had over-the-ear
58:28 headphones and the rabbit was built into
58:30 it yeah then you just have those and oh
58:33 my Lord that'd be incredible what has
58:35 anybody ever put uh a GP a CPU inside of
58:39 headphones I've never seen that that
58:41 would be amazing there must be some in
58:42 Shenzhen or akih Habra we could go find
58:45 stic I mean shenan already checked you
58:47 know I I called my bu I I checked
58:50 there's there's there's yeah there's
58:51 there's qualcom there's there's Apple
58:55 different parties are working on that uh
58:58 we're not only talking so they might
58:59 come out with it you think they might
59:00 come out with I think I think that's
59:02 where I was working towards too but so I
59:05 would put on my just the AI I put on my
59:07 Apple Air Max whatever they call those
59:09 over the- year ones yeah and then I say
59:12 hey Siri play me The Clash and I don't
59:16 need to have my phone with me and it's
59:17 got Wi-Fi and LTE belt into it that
59:19 would be amazing and and Jason I'm going
59:21 to show you something this is the actual
59:24 Board of R1 it's just that big oh okay
59:27 so it's the size of we we we didn't even
59:30 try to make it smaller because because
59:32 look at the phone factory it fits right
59:34 FAS we we didn't even try to make it
59:37 smaller if I tried really hard to make
59:38 it smaller uh I mean definitely I me
59:40 half of a playing card if you cut the
59:42 Ace of Spades in half that's about how
59:44 big the the car the the the motherboard
59:47 is as it were yeah so so so over the air
59:51 I think we can probably do now but I
59:53 want to do the airpod version yeah which
59:56 that's going to be a little bit
59:57 challenging and also of because you know
01:00:00 you want to set it up to always
01:00:01 listening and then the battery become
01:00:03 the issue you know so always listening
01:00:05 takes uh yeah a little bit more battery
01:00:07 yeah yeah so so but I think you know I'm
01:00:09 pretty confident in the next you know 48
01:00:13 months that's something that we can
01:00:15 start figuring out um but another thing
01:00:18 I think is very very good to think about
01:00:21 and don't quote me on this I'm not I'm
01:00:22 not leaking whatever this is
01:00:26 I'm not trying to say oh this is what
01:00:28 we're going do next this isn't the
01:00:30 product Road we're just riing here this
01:00:32 isn't yeah this isn't um I think a watch
01:00:36 makes a lot of sense too uh a lot of
01:00:38 community members already designed
01:00:40 rabbit watch on Twitter and tagged me uh
01:00:42 amazing work um but if you think about
01:00:45 it Apple watch is probably the weakest
01:00:47 product in their entire protu
01:00:50 L I have to say Apple watch sucked until
01:00:53 the ultra yes Ultra battery is bettery
01:00:57 the battery is better the screen size is
01:01:00 better yeah and um I I like when I jump
01:01:03 in the water yeah uh that it tells me
01:01:05 how many feet I'm underwater right right
01:01:08 it's just having the battery life
01:01:10 actually last two days yeah is such a
01:01:13 GameChanger compared to the other one
01:01:15 and um the screen being I don't know
01:01:17 what is the screen 50% bigger than the
01:01:19 other ones it's it's bright it's very
01:01:21 bright 100 it's very bright and that's a
01:01:24 big Victor too I I mean the the way I
01:01:27 put it this from my old my own
01:01:29 experience that I said maybe Apple watch
01:01:32 is the weakest in their entire product
01:01:34 life is that it's accessory it's not a
01:01:36 standalone device they really haven't
01:01:39 focused on it as a standalone device I
01:01:41 agree they said it's a standalone device
01:01:43 I understand you can make phone calls I
01:01:45 understand someone it sucks for phone
01:01:47 calls some some YouTubers carry that for
01:01:49 a week you know I saw the YouTuber
01:01:51 challenge that yes I think someone like
01:01:53 really like have their Apple watch for
01:01:55 like a week it works but it's not
01:01:58 designed for it um I think there's a
01:02:01 good opportunity or ideas in general
01:02:04 that because I'm start pay attention to
01:02:07 kiss and what is that I I start to pay
01:02:10 attention to how how future generation
01:02:12 like kids how kids are kids kids how
01:02:14 kids are using their devices oh yeah
01:02:17 this is one of the community I just
01:02:18 wanted to show this while we're on air
01:02:20 you're this is a fanade this so for
01:02:22 people who AR watching it's the IC
01:02:25 orange and it is gorgeous and then it's
01:02:29 got like a you know the the crown but
01:02:32 the the best part about it is it has a
01:02:34 popup camera and the screen is designed
01:02:37 with a little carve out of a popup
01:02:38 camera completely impractical but
01:02:40 awesome yeah it h it has preserved all
01:02:43 the key elements right like the V the
01:02:45 venting grills and all that it's super
01:02:48 impressive how did this Community start
01:02:50 by the way I just have to stop there for
01:02:51 a second because did you Goose this
01:02:55 lunatics in industrial design or is it
01:02:57 they follow teenage what is it called
01:02:59 teenage um Teenage engineering
01:03:01 engineering is it that teenage
01:03:03 engineering inspires PE other
01:03:05 engineering students or people it's nuts
01:03:08 to watch how many people have started to
01:03:10 make derivative products out of your
01:03:12 iconic yeah and I it is Iconic I know
01:03:14 you don't want to say that but it's such
01:03:15 an iconic uh device that it's inspiring
01:03:19 people to make very refined product
01:03:21 extensions for you for free yeah exactly
01:03:23 so so I think first of all a huge credit
01:03:26 to teenage engineering if you ever ever
01:03:29 worked in design or industry design or
01:03:33 day they're god tier they're like the
01:03:36 ultimate outter uh at least that's my
01:03:39 opinion I don't care if anyone agrees or
01:03:41 disagrees with me they're in my heart
01:03:43 the god tier um so that's one part
01:03:48 another part is Jason phones are just
01:03:50 too boring now like industrial design is
01:03:52 pretty much dead
01:03:55 like explain that yeah because I can't
01:03:57 tell the difference between my last five
01:03:59 iPhones it's all the same to exact
01:04:01 exactly like like phone companies they
01:04:03 set on their current phone Factor
01:04:06 usually let's think about the fun time
01:04:08 right let's think about Sidekicks engage
01:04:12 think about those things you know like
01:04:13 flipping nokas like spinning nokas um I
01:04:18 think I think it's been so long uh if
01:04:21 you look at the past especially past
01:04:22 five five years I think a very good
01:04:25 example from Samsung and Apple you know
01:04:28 now it is Iconic I saw iPhone I know
01:04:31 it's iPhone I saw an a galaxy I know
01:04:34 it's a Galaxy phone yes but I also can
01:04:37 tell you 100% the next generation will
01:04:38 be look like the same AB with a with a
01:04:41 new different col and this is because of
01:04:44 manufacturing these things are so
01:04:46 expensive to start a new version it's
01:04:49 exactly like cars you know once you make
01:04:52 the for the Toyota Prius every time you
01:04:54 redesign it yeah you're basically
01:04:57 restarting like a whole production line
01:04:59 and you don't get any of the scale so
01:05:02 they're it's a scale business they're
01:05:03 trying to get massive margin in a
01:05:04 competitive space so they don't change
01:05:06 it yeah but I guess Steve Jobs doesn't
01:05:10 a he does not give a you can be sure he
01:05:13 did not give a it just it just that that
01:05:15 spirit is generated debt and I think
01:05:18 we're not helping we're not we're not
01:05:20 we're not basically doing this it's just
01:05:24 feeling towards all of us as
01:05:27 consumers that that that if you bring
01:05:30 something new and fresh yes it is it is
01:05:33 relatively cheap yes we're a new company
01:05:35 but people see the joy of it by looking
01:05:38 at it and then then that inspired them
01:05:41 to create their own iteration of it
01:05:43 which is another round of Ripple and
01:05:46 this Ripple just start getting bigger
01:05:47 and bigger and I really really
01:05:51 understand why cpay started nothing in
01:05:53 the beginning we have couple chats you
01:05:55 know he his whole company's position is
01:05:59 phones are boring that's it that's why
01:06:01 there's and I mean the nothing phone is
01:06:04 wild what what do you love about the
01:06:06 nothing phone what is it that uh
01:06:09 inspires you about it I I think I think
01:06:12 they are probably the most
01:06:14 engineering uh and Industrial design uh
01:06:20 company come you know giv by the fact
01:06:22 they are still they are still startup
01:06:25 most of the startup wouldn't invest this
01:06:27 much into both of these fundamental
01:06:30 areas to make a product I think I gave
01:06:32 you I gave them huge credit for that
01:06:35 yeah um and and you know I I haven't
01:06:39 developed directly you know uh involved
01:06:42 in in their management or anything like
01:06:45 that but you know being with teenage
01:06:48 seeing this company from beginning uh
01:06:50 it's quite a journey it's quite a
01:06:52 journey it's very challenging you know
01:06:53 there is isn't cheap you can tell is
01:06:56 isn't cheap so it's not cheap it's a
01:06:58 luxury item and they're trying to make
01:07:00 it's it is a luxury item or not uh well
01:07:03 it's still a little bit cheaper $100
01:07:06 cheaper than the flagship Android phones
01:07:08 you know say okay but but it's not as as
01:07:11 cheap as like shenen stuff yeah so so I
01:07:14 I guess you know the younger
01:07:17 generation if you think about younger
01:07:19 generation like all the all the kids
01:07:21 that born after 20 2000
01:07:25 they when they grow up have the first
01:07:29 access to the phone and then for the
01:07:31 past five years it's the same boring
01:07:33 stuff how crazy is that and think about
01:07:35 when we started to get have our phones
01:07:38 we have we have 200 different things to
01:07:40 choose from right and and we're getting
01:07:42 a phone just because it's cool that's it
01:07:45 I think I think that part is is is
01:07:47 mostly gone and you're right is but it's
01:07:49 because of the what's behind the the the
01:07:51 the manufacturer logistic stuff um but I
01:07:56 think it's good for us because we're so
01:07:58 small and so in the beginning that we
01:08:01 can take this you this massive risk yeah
01:08:04 you can take the risk we can take this
01:08:06 uh but but also you have nothing to lose
01:08:08 you nothing to lose yeah like if you
01:08:11 make this form factor and it sucks you
01:08:13 change it and you try the next one and
01:08:15 you try the next one until the audience
01:08:17 responds if Apple screwed up an iPhone
01:08:19 15 the stock would lose $500 billion
01:08:23 yeah and when they've had problems
01:08:25 shipping problems Etc something gets
01:08:27 screwed up they lose a hundred billion
01:08:29 dollars in market cap and then everybody
01:08:31 loses their mind I mean this is one of
01:08:32 the great things about being
01:08:34 small yeah and that's exactly the reason
01:08:37 why I don't think Syrian and Google uh
01:08:39 assistant or or Alexa can catch up so
01:08:41 quickly I think they have to throw it
01:08:43 away I mean honestly I I think the
01:08:45 technology has moved so fast that Siri
01:08:48 and Alexa need to be thrown in the
01:08:50 garbage and they should just start a new
01:08:52 project it actually go faster yes yes
01:08:55 yes I agree with that I age the
01:08:57 technological debt of Siri must be
01:09:00 insane the entire Paradigm has shifted
01:09:02 right to lambs and llms and they're not
01:09:05 built on that yeah I mean it's a 10 year
01:09:07 old product you know so toss it in the
01:09:09 garbage I mean that's I mean this what
01:09:11 you know if you look at what elon's
01:09:12 doing with self-driving yeah I think he
01:09:15 doesn't have that fear of saying hey we
01:09:17 built this software to stay in the lane
01:09:19 but it's not you know there's new
01:09:22 technology that's going to disrupt it we
01:09:23 need to to that now the hardware stack
01:09:25 may not change right right the hardware
01:09:27 stack could be the exact same for the
01:09:29 iPhone but Siri has to die you know one
01:09:32 thing I I want to give huge credit to
01:09:34 Elon is that you know there's certain
01:09:36 ways he thinks that I think really
01:09:38 really resonates with me uh one example
01:09:41 is that if you look at the all the
01:09:43 cables in the Cyber Tru I'm not sure if
01:09:45 you ordered one I have one coming in
01:09:47 yeah if if you if you look at how they
01:09:49 redesign the cables it's crazy because
01:09:52 all the cars have like what what like
01:09:54 now like 10 miles long of cables all
01:09:56 around all wrapping in different p and
01:09:59 he just doesn't like that he said okay
01:10:01 that's too complicated let's make one
01:10:05 big fat cable and with two plugs and
01:10:09 that's it um I I think this is this is
01:10:12 this is some thoughts that really pushes
01:10:14 a lot of Innovations and and you know
01:10:16 from where we are benefiting from it is
01:10:19 that like you said we're small we take
01:10:21 the risk but this is how we want to
01:10:22 think about things you know
01:10:24 like like we're not saying oh maybe we
01:10:26 wait for two more years there's more API
01:10:28 we can work with maybe we can you know
01:10:31 suit up and and walk into four apps
01:10:33 offices talking to their Executives and
01:10:35 maybe we can get a deal we're like okay
01:10:38 how about we get rid of all that we
01:10:39 don't work with apis we don't work with
01:10:41 with apps we just do something from
01:10:44 grown up I think we hugely benefit from
01:10:46 thinking like that yeah yeah the Cyber
01:10:48 truck went to this 48 volt system um
01:10:53 which reded red the W the total wiring
01:10:56 was reduced like I think it was 80% in
01:10:58 the car it's Bonkers and you know now I
01:11:03 guess the only downside to that is if
01:11:05 you lose that cable yeah the car is done
01:11:08 right you lose so much but then the
01:11:11 upside to that is if there is a problem
01:11:13 it's easier to diagnose you swap the
01:11:15 cable out yeah I mean it's it's hard to
01:11:17 lose the cable because you got
01:11:18 bulletproof panels right so yeah exactly
01:11:21 you can't sh I mean Joe Rogan tried to
01:11:22 hit it with a with an arrow and it
01:11:24 didn't work and he's a pretty good shot
01:11:26 so you're good you're good listen Jesse
01:11:28 you're like an amazing guest I want you
01:11:30 to come back when it ships when are
01:11:31 people going to get these in their hands
01:11:32 do you think what's the best estimate
01:11:34 okay so uh we will have uh our our
01:11:38 promise date is Easter which is March
01:11:40 31st that's where where we start
01:11:43 shipping we we're working with local
01:11:46 compliance and legal teams to get
01:11:47 International ORD ship as fast as we can
01:11:50 Easter seems reasonable Easter seems
01:11:52 reasonable knowing that we said okay the
01:11:54 international batches should be you know
01:11:56 could be June July but we're trying to
01:11:58 move everything faster than than it
01:11:59 should be um but uh we can
01:12:03 allocate um maybe up to five 50 50 to
01:12:07 100 somewhere in that range of media
01:12:10 test devices great thank you for
01:12:12 including me I'm I'm super excited yeah
01:12:14 you already signed yourself up I know
01:12:16 that so so that uh expecting somewhere
01:12:20 mid October so in the in next two three
01:12:24 and I will start rolling out all the
01:12:25 demos as I showed on a keynote uh
01:12:28 starting from next week I already showed
01:12:30 a couple Bunch uh here and there but the
01:12:32 you know like I said latency is
01:12:34 something that we want to really really
01:12:35 focus on and uh how many years have you
01:12:38 been working on this by the way I think
01:12:39 this is something people don't
01:12:40 appreciate that this is not something
01:12:41 you've been working on for six months no
01:12:44 uh the original idea like I said rabbit
01:12:46 is basically Riven 2.0 it carries a lot
01:12:49 of the it's the same original passion
01:12:51 and carries a lot of the Legacy but if
01:12:53 you want to me won't have me today you
01:12:56 know back track to day one I actually
01:12:58 had my YC uh application video that
01:13:01 community members put up from from my my
01:13:04 like uh YouTube channel and that was
01:13:06 dated at nine years ago nine years ago
01:13:09 when you started this journey two
01:13:10 companies ago and then this company you
01:13:12 formed a couple of years ago right 2020
01:13:14 or something 2020 yeah it's just it just
01:13:16 you know the idea was always right I
01:13:18 guess the direction was always correct
01:13:20 but unfortunately like I said 10 years
01:13:22 ago we don't have all these cards on the
01:13:24 table like the intention wasn't solved
01:13:27 there isn't there isn't GPU for us to or
01:13:29 CPU for us to utilize at this current
01:13:31 rate um so I feel like I just waited
01:13:35 till till you know the hardware the
01:13:38 infrastructure had to catch up to the
01:13:39 vision that's basically it exactly and
01:13:42 and and yeah I will say that you know
01:13:45 it's it's been sticking with my hand for
01:13:47 so so long and that that also explains
01:13:49 Jason that we're moving so fast we're
01:13:52 we're not we're not not starting from
01:13:54 nothing and in two months and and build
01:13:57 a hardware that's impossible right yeah
01:13:59 yeah and the the the ability to create a
01:14:02 hardware startup today is much different
01:14:04 than when you started 10 years ago
01:14:06 exactly yeah I mean exactly the the
01:14:08 sophistication level in Shenzhen as just
01:14:10 one example when I went to shangzhen for
01:14:12 the first time I think it was
01:14:14 2006 or seven okay it
01:14:18 was I it was so rare to see a an
01:14:22 American a white person when I went
01:14:24 there yeah I I kid you not 20 school
01:14:27 students surrounded me to touch me and
01:14:30 take a picture with me because they'
01:14:31 never seen because they had come down
01:14:32 from the north there's a there's a
01:14:34 famous um um amusement park in shenen
01:14:38 and it's not for American Tourist it's
01:14:40 for Chinese you know native uh Nationals
01:14:43 and I went to it just to check it out
01:14:45 and I literally got stopped you know 10
01:14:47 times just people wanted to touch me and
01:14:48 take a picture now there's a lot of
01:14:50 Americans there there's a lot of
01:14:51 startups there and literally when I was
01:14:53 there in 2007 15 years ago I guess there
01:14:57 more there were there'd be like a tall
01:15:00 building 30 story building y y and for
01:15:02 four blocks around it were all parking
01:15:04 lots yeah and you would drive from one
01:15:07 building to another you could see the
01:15:08 building you were driving to but it was
01:15:09 like driving through the desert and you
01:15:11 know what you see that in um uh the
01:15:13 Middle East now some of the regions
01:15:15 there where like boom a tall building
01:15:16 comes out boom Another TR building comes
01:15:17 up and in between it there's no
01:15:19 infrastructure just a road but nothing
01:15:21 else it it was great now you go over
01:15:23 there my Lord I I left my I was
01:15:26 originally born in China but I was
01:15:28 raised on in know where Western sty
01:15:30 family uh you know like my parents be
01:15:32 like once you're 15 you're on your own I
01:15:34 started learning English right after I I
01:15:37 I where did you grow up I grew up in
01:15:39 Shian where the terracota warrior is but
01:15:42 I oh yeah but my my parents start
01:15:44 teaching me English on day one and uh
01:15:47 and then they're like once you're 15
01:15:48 you're on your own I I literally uh left
01:15:52 my left my house um at the age of 15 I
01:15:56 got a transfer opportunity briefly
01:15:57 started at a a high school um in
01:16:01 Singapore and then I went to UK so I I
01:16:05 graduate from University of Liverpool uh
01:16:07 mathematics and then uh spent
01:16:10 another year and a half in London and uh
01:16:13 just to hang out with whole bunch of
01:16:14 artists U uh so I I don't really
01:16:18 understand China now you know I've been
01:16:20 I've been leaving the country so long
01:16:22 but I I believe you like you know it's
01:16:25 crazy and the the heavy odm OEM
01:16:29 infrastructure uh that they set it up
01:16:31 over the past decades is still you know
01:16:34 at least serving Apple you know all the
01:16:36 iPhones were made there you know yeah
01:16:38 and and and uh but but you know I think
01:16:42 a lot of things are start shifting a bit
01:16:45 because I think um what I can see from
01:16:49 Hardware perspective is that a lot of
01:16:51 things are become modular but on the
01:16:53 other hand uh core chips are become
01:16:57 centralized so the core chips are being
01:16:59 centralized t uh tswc uh tsmc
01:17:04 semiconductor yes and Nvidia are locking
01:17:08 that down or you know arm and other
01:17:11 folks are locking it down yeah but the
01:17:13 components anybody can make them and
01:17:15 just plug them into those CPUs G anybody
01:17:17 can anybody can make it but I think no
01:17:20 one still beats the shinen speed
01:17:23 that's a cool part that's a cool part
01:17:25 yeah they can they can fabricate
01:17:28 something for you that's brand new in
01:17:30 days and have it back to your office in
01:17:33 America exactly at iterate I mean the
01:17:36 iteration cycle used to be months yeah
01:17:38 now it's weeks or days no it's
01:17:41 definitely weeks and if you have some
01:17:43 special resources you can make it happen
01:17:45 in days for sure which is wild when you
01:17:47 think about it the it used to be like
01:17:49 quarterly you could yes get your product
01:17:52 you know iterated on I know because I
01:17:54 have a bunch of Hardware startups I
01:17:55 tried to invest in 10 years ago and it
01:17:57 was just the iteration process was too
01:17:59 slow um and you you couldn't be nimble
01:18:01 now you can get it done in weeks maybe
01:18:02 even under a week that's extraordinary
01:18:05 right yeah but but I I I I see the the
01:18:08 chips are getting centralized which is
01:18:10 not very good for startup to be honest
01:18:13 ah why oh because you know you're going
01:18:16 to work with only selected partners and
01:18:18 then they have so many big clients to
01:18:21 with and then there like okay if you
01:18:23 don't sell more than this number we
01:18:25 don't even talk to you got it uh so yeah
01:18:29 if you want to get the attention of
01:18:30 Invidia in the age where Facebook's
01:18:33 buying 10,000 yeah you know h100s is
01:18:36 going to be hard what do you think about
01:18:37 risk and the open source chip movement
01:18:39 is that going to have an impact at some
01:18:41 point in the future risk for and stuff I
01:18:43 I think I think it's more from how you
01:18:45 want to design a system a lot of people
01:18:47 are debating you know I've been debating
01:18:48 myself and been debating with my own
01:18:50 community on Discord a lot of people
01:18:52 like like hey can we make a local L
01:18:55 right can we can we run x amount of
01:18:57 features on device instead of in the
01:18:59 cloud yeah um for me I get it from the
01:19:03 Privacy reasons I get it I think that's
01:19:05 that's probably the only valid reasons
01:19:08 like why why are you even willing to run
01:19:10 it on local because yeah and how many
01:19:14 people care about that very small number
01:19:15 of people if you encrypt my instructions
01:19:18 do them in the cloud and then delete
01:19:19 them yeah what's the I mean or you
01:19:22 delete them every whatever 60 days or
01:19:24 something you may need to have them for
01:19:26 some historical purposes but it it just
01:19:28 it just there's a huge dispute between
01:19:31 academic perspective and consumer
01:19:32 perspective right like consum have voted
01:19:35 they don't care they don't care exactly
01:19:37 but but I think you know the way I bring
01:19:38 this case is that there is one way that
01:19:41 we can work further towards like
01:19:43 Chromebook ways right like even the
01:19:45 operating system is in the
01:19:46 cloud if we go that route then we don't
01:19:50 worry about trips because you don't need
01:19:51 that trip to run loc Al to do the tasks
01:19:55 it's a terminal right you turn the
01:19:56 device into a terminal it's a terminal
01:19:59 uh I I'm actually Pro that route you
01:20:02 know uh just because from consumer
01:20:04 perspective why the heck do I need a CPU
01:20:08 in my car leader
01:20:09 box yeah why the heck do I need that and
01:20:13 and I think you know everything is smart
01:20:16 the TV is smart and everything's smart
01:20:18 uh I I actually I was setting up a a
01:20:21 little apartment because getting too
01:20:23 busy I I stayed out office I was setting
01:20:25 up the TV and then I need to do the the
01:20:29 air airpl thing from my my content of
01:20:32 computer so I went Best Buy I bought a
01:20:35 Apple TV and then I bought this smart TV
01:20:37 from Samsung and they both have Netflix
01:20:42 yes and so now have two remotes and one
01:20:44 of the the Samsung remote has the
01:20:45 Netflix logo on it's like why did I buy
01:20:47 the Apple TV exactly and and then
01:20:49 Samsung had their own way of chomecast
01:20:51 or whatever yeah
01:20:53 yes I I think you know then if you think
01:20:56 about what's happened behind is that do
01:20:58 we need all that and Trace further down
01:21:01 do we need all that chips you know do we
01:21:03 really need a qualcom in your TV if your
01:21:06 TV can just project things and that's
01:21:08 the TV should do right yes so I I think
01:21:11 you know it's it's a funny uh uh it's an
01:21:14 interesting uh Trend to watch in know
01:21:16 especially in the next couple five years
01:21:18 I always see who can get upper hand and
01:21:20 you know the local guy versus CL guy and
01:21:23 how it works um but yeah I do agree with
01:21:25 elom musk and a couple other folks that
01:21:28 we will have a power shortage if the AI
01:21:31 runs like that oh we'll have a shortage
01:21:35 of actual power to homes to Data Centers
01:21:38 we're going to need much more power
01:21:39 because these things run hot yes and
01:21:41 also we don't really have like
01:21:43 transistors and Transformers to be able
01:21:46 to handle the cluster of the GPU if we
01:21:50 keep the current Pace m uh so so a lot
01:21:54 of basic infrastructure to to look at
01:21:56 but for us you know for at least now you
01:21:59 now this will get people interested in
01:22:00 nuclear again if we start having uh an
01:22:03 energy crisis because of so many h100s
01:22:07 and then eventually 200s whatever being
01:22:09 deployed so people are going to say you
01:22:10 know what we just need to start building
01:22:12 some nuclear reactors next to these data
01:22:14 centers ex and putting in massive solar
01:22:16 farms and really having much more much
01:22:18 cheaper energy right um which you know
01:22:21 people haven't had to think about it all
01:22:22 that much because we we've been able to
01:22:25 keep up but if all of a sudden we can't
01:22:27 solve cancer because we don't have
01:22:28 enough energy we have enough chips to
01:22:30 cure cancer yeah we have the software we
01:22:33 have the chips we don't have enough
01:22:35 energy to put at this because a custom
01:22:37 medicine or something you know for me is
01:22:39 how far away if you're a consumer how
01:22:41 far away from the actual problem right
01:22:43 if if you're very very far away you
01:22:45 don't you don't care but but if you're
01:22:47 if you're next next to it and then and
01:22:50 then all of a sudden become a problem so
01:22:53 so I think you know like I said we are
01:22:55 we don't know whether this is the the
01:22:58 correct phone Factor you know we don't
01:23:00 know but we know this is the lowest risk
01:23:04 we can pull off and and we know the lamb
01:23:07 is most important thing yeah I mean the
01:23:09 lamb is like a sniper shot you can just
01:23:11 very easily get accomplish the task
01:23:14 and you know when we started in the
01:23:16 industry I remember in the 90s there was
01:23:18 a company called General magic oh yeah
01:23:19 and they were making a device that had
01:23:21 agents and you would tell agent I want
01:23:22 to fly to Paris and I'm in it would know
01:23:25 you're in New York and it would go out
01:23:27 try to find tickets for you go to uh
01:23:29 literally go to a human yeah uh um
01:23:33 travel agent and then get try and get
01:23:34 the results back and then put them into
01:23:36 some format that you could say okay I
01:23:38 like that itinerary um and it now here
01:23:40 we are uh 2005 2015 2025 yeah it took 30
01:23:45 years basically funny work funny enough
01:23:49 General magic was a internal Apple
01:23:51 project that Apple don't really
01:23:53 understand and spin it off and their
01:23:55 logo is a rabbit in a hat ah right I
01:23:58 remember that and the rabbit would pop
01:24:00 its head out in the little animation
01:24:02 this was I remember Sony made the first
01:24:05 actual hardware for General magic and it
01:24:07 was some early iteration of the pdas and
01:24:10 stuff yeah yeah personal digital umist
01:24:13 assistance they were I I remember when
01:24:16 the Palm came out in New York right
01:24:18 right around 97 98 99 people had a
01:24:20 starac phone and they would a pal you
01:24:23 take these two devices out and you
01:24:24 learned Palm script you learned how to
01:24:26 do script on the touchcreen and the
01:24:28 number one application wasn't sending
01:24:30 money that was like what Elon and Peter
01:24:32 chill were going to do um or Peter chill
01:24:34 was going to do with PayPal it was to
01:24:36 send a business card and then what's the
01:24:38 latest feature of iPhone 15 is tapping
01:24:40 the two phones together to send a
01:24:41 business card yeah it's hilarious yeah
01:24:44 so that's that's actually really really
01:24:46 good intake because I deeply believe
01:24:48 that you know human are never inventing
01:24:52 new demands we have certain category of
01:24:55 Demands right we want to want to rest
01:24:57 well we want to eat well and drink well
01:24:59 and be entertained and you know be
01:25:00 motivated and start working but
01:25:03 technology is not Reinventing new Dems
01:25:06 technology is solving the same problem
01:25:08 over and over again but each generation
01:25:11 in a much more intuitive way ah yes and
01:25:15 that's exactly what you said with pal
01:25:17 sending that scripts to name card versus
01:25:19 now tap to air drop it's same thing it's
01:25:22 just much more intuitive and and
01:25:24 Hardware is very interesting because if
01:25:26 if if the hardware is intuitive enough
01:25:30 then the phone Factor stays for a very
01:25:32 very long time you never I never heard
01:25:35 another 200,000 companies every year
01:25:37 claiming to make a better Hammer no why
01:25:41 because it's intuitive enough it it only
01:25:44 take it only takes your force and obey
01:25:46 the physic CLW that's it intutive like
01:25:49 that but then if you think about
01:25:51 computers and mobile devices there's
01:25:54 long there's long long way to go long
01:25:56 way to go yeah long way to go and and
01:25:58 that's why we decided to take a little
01:26:00 step forward we're like okay I
01:26:03 understand you guys are skeptical about
01:26:05 this and that but we have to try we have
01:26:08 to do this at least you know like
01:26:10 there's something about the instant
01:26:12 nature of this that I think you know if
01:26:15 you save people I've I've always felt
01:26:17 when I look at startups to invest in
01:26:19 yeah is this startup saving people time
01:26:21 yeah is it saving people money or is it
01:26:24 entertaining them in some way right
01:26:26 right and so if you look at your device
01:26:28 it's going to save people time and time
01:26:31 is the equivalent of money for some
01:26:32 people some people have more money than
01:26:34 time some people have more time than
01:26:35 money yeah but generally speaking I
01:26:37 think yours is a time-saving device and
01:26:39 then eventually will it be entertaining
01:26:41 um it could be sure um there could be
01:26:43 entertainment services on it but just
01:26:45 saving people time yeah and letting them
01:26:47 get back to their kids uh or to dinner
01:26:50 or whatever and get off their device
01:26:51 yeah that is extraordinary and so I
01:26:53 think your your recognition that the
01:26:56 latency has to change yeah massively was
01:26:59 what Sergey Marissa uh and Larry were
01:27:03 obsessed with that Google is how to get
01:27:04 it faster faster faster lower the
01:27:06 milliseconds people will search more if
01:27:09 they made the search results faster and
01:27:10 and I think people will use AI more
01:27:13 right if you you know can get that uh
01:27:16 you know yeah um I can probably get
01:27:19 interaction lower I can probably get it
01:27:21 um by the time you have your test device
01:27:24 yeah it's gonna be amazing I mean just
01:27:26 the I can tell you right now the ability
01:27:28 to order food faster I I do the same
01:27:30 order for my family for the sushi rolls
01:27:32 that we get those like hand rolls um if
01:27:34 I could just reorder that roll boom and
01:27:37 the the faster I can do it man it would
01:27:39 be amazing all right listen Jesse you're
01:27:41 awesome when this thing launches like in
01:27:43 three months will you come on again um
01:27:45 of course yeah all right here we go and
01:27:46 maybe we'll do it in person maybe we'll
01:27:47 do some oh you know what when you when
01:27:49 you're ready to launch it save a couple
01:27:50 of demos like maybe three and we'll do
01:27:52 live yeah live do live dios why not yeah
01:27:55 no no no but we'll do it live in person
01:27:57 I'll get a theater and we can just sit
01:27:58 there and talk and you show a couple of
01:27:59 live things demo or die style all right
01:28:01 listen uh if you want to learn more or
01:28:03 if you want to uh go work with Jesse and
01:28:05 do crazy awesome rad stuff and cool like
01:28:09 Steve Jobs did how can people find out
01:28:11 about your job openings oh uh simple we
01:28:14 have a career page uh we're continuously
01:28:16 updating in the roles but uh just feel
01:28:19 free to reach out to me via Twitter or
01:28:21 feel free to send a contact at rabbit.
01:28:24 te email just just go to our website
01:28:27 drop up the email we are and website is
01:28:29 rabbit. yeah rabbit. Tech and uh make
01:28:32 sure to check on the research page for
01:28:34 the detailed breakdown of lamb how it
01:28:36 works yes you have a really cool page
01:28:37 about how lamb works yeah so uh yeah
01:28:40 just send us an email we we we'll get it
01:28:42 there uh I mean we what are you hiring
01:28:44 for what what positions are like the
01:28:46 hardest yeah so uh it's it's it's
01:28:50 basically we need to expand uh
01:28:52 most of the product related roles in
01:28:55 every aspect because we're trying to
01:28:57 catch up with the demand and there's a
01:28:58 lot of things that we're trying to uh
01:29:01 you know uh build for the future and
01:29:03 more importantly we we want to set up a
01:29:05 lamb research like small lamb research
01:29:07 team uh just working for the future
01:29:09 iterations of lamb um are you in person
01:29:12 now you're in person in Santa Monica
01:29:14 trying to get people in an office uh
01:29:16 yeah but uh you know if you're good it
01:29:19 can be flexible but F yeah but you these
01:29:22 will'll be really good if you're really
01:29:24 good flexible but if you're very good
01:29:26 let me tell you something living in
01:29:27 Santa Monica there's a lot of great
01:29:29 restaurants it's incredible lifestyle
01:29:31 man just move and and hang out with
01:29:34 Jesse and Santa Monica it's one great
01:29:36 plac my goal is to to to raise more
01:29:38 money so that we can afford you guys but
01:29:41 yeah that's my you'll be fun I think
01:29:43 you'll be fun all right everybody we'll
01:29:44 see you next time when this week starts