00:00:00
all right everybody welcome to your
00:00:02
favorite podcast in the world's number
00:00:04
one podcast the all-in podcast it's
00:00:06
episode
00:00:07
1,790 oh wait that's just how it feels
00:00:10
welcome to episode
00:00:12
179 with me today of course is your
00:00:15
Sultan of science I don't know if that's
00:00:17
a movie background or it's just his
00:00:19
favorite vegetables what's going on
00:00:21
there what's the crop that's AI
00:00:22
generated it's AI generated crop okay
00:00:25
great I'm trying AI backgrounds I'm
00:00:26
going to try it out for a while with
00:00:27
different props your fans are going to
00:00:29
be crushed that you're not doing deep
00:00:30
movie polls with us of course man about
00:00:34
town
00:00:35
DC new products being launched David
00:00:38
saaks the Rainman yeah how you doing
00:00:40
buddy good good yeah good week lots
00:00:43
going on yeah yeah definitely a good
00:00:45
week shth poopaa chairman dictator he
00:00:50
puts the chairman in dictator I would
00:00:52
like to take this
00:00:54
opportunity to
00:00:57
wish my child a happy
00:01:00
birthday I absolutely love you well now
00:01:04
the rest of us look like yeah great I've
00:01:07
never done that before sax in your desk
00:01:09
in your desk is a piece of paper with
00:01:12
your children's names and their
00:01:13
birthdays you want to pull it out I got
00:01:14
three birthdays a year and I've never
00:01:16
done
00:01:18
one let your winners
00:01:20
[Music]
00:01:25
ride and instead we open source it to
00:01:28
the fans and they've just gone crazy
00:01:30
love
00:01:30
[Music]
00:01:34
you no no no but I'm saying it rarely
00:01:36
lands on the same day today is the day
00:01:39
today is the day today the day today is
00:01:41
the day con congratulations child oh
00:01:44
congratulations yeah how old Jam no
00:01:47
gender name or any other specifications
00:01:48
folks we can't we can't tip anybody off
00:01:50
no pronouns no pronouns uh yes Abol so
00:01:54
how are they them experiencing their
00:01:56
birthday this child has experienced a
00:01:59
wonderful and this child is an
00:02:01
incredible
00:02:02
person for whom I have tremendous
00:02:05
admiration and love and compassion and
00:02:07
hope for the future all right and did
00:02:08
you order them some chicken
00:02:10
fingers I cannot comment on who this
00:02:12
person
00:02:13
is chicken fingers are you talking of
00:02:16
course about Phil
00:02:17
helmuth your child Phil helm can we
00:02:20
please talk about last weekend's
00:02:22
festivities in what a disaster he is oh
00:02:25
my God guys just so you guys know so we
00:02:28
missed you last weekend we missed so
00:02:30
much fun to we miss you on Saturday
00:02:32
night Saturday night was really fun I
00:02:34
had such a lovely time coming home to be
00:02:36
totally honest with you we had a cabana
00:02:38
set up on Saturday played blackjack I
00:02:41
miss you guys too I had a fomo saw the
00:02:43
videos it was so fun well you don't have
00:02:45
to have too much fomo because Phil sent
00:02:47
the entire group chat to pokernews.com
00:02:50
they did an article twice the flop. org
00:02:56
poker Dash update
00:03:00
he leaked every single person who's
00:03:02
there and the
00:03:03
TW he's like look here's me and Elon
00:03:06
Elon came by for my dinner no it was
00:03:08
worse than that no it's worse than that
00:03:09
he said I got to hang out with our guy
00:03:12
Elon for 10 minutes and 14
00:03:15
seconds wait what he di he intercepted
00:03:19
him at the
00:03:21
valet wait
00:03:23
what how many minutes God 10 minutes and
00:03:26
14 seconds he had the exact time down to
00:03:28
the second oh my God I want to wish
00:03:30
helmouth a happy birthday CU I did miss
00:03:32
his 60th party yeah it's coming up bir
00:03:36
good news is it wasn't actually his
00:03:37
birthday it was Bill Gurley so he just
00:03:39
hijacked Bill gurley's birthday I also
00:03:41
got to enjoy for my first time ever uh
00:03:44
the experience of bakarat which I've
00:03:45
decided is the most d dgen game on earth
00:03:48
it's literally the most you just flip a
00:03:51
coins coins it's flipping coins well C
00:03:54
you you make betting decisions all you
00:03:56
do in bakarat is you say bank or player
00:03:59
and then you freak yourself out about
00:04:00
how you flip the cards and the smartest
00:04:02
people I know on Earth are all sitting
00:04:03
around this table at 2 3 in the morning
00:04:06
saying turn this corner this way no no
00:04:08
no no no turn it this way turn it this
00:04:10
way there's two dots and they're
00:04:11
debating the right way to flip
00:04:14
a people on Earth no the Bak sweat is
00:04:17
the most incredible performative act in
00:04:19
the casino Weir yeah you're right
00:04:22
everyone's got their own little
00:04:23
technique about how they bend the cards
00:04:25
the cards are all destroyed by the end
00:04:27
of the deck they don't they get
00:04:30
I go like this and I try to see curling
00:04:33
your mustache like an evil
00:04:35
vill and then you call out oh my God no
00:04:37
spotter if you see a no spotter or two
00:04:40
across so great and then you get to
00:04:43
decide whether the bank turns over their
00:04:44
cards and it what when they turn over
00:04:47
then you lose a small house and then
00:04:48
you're like oh let's try again yeah
00:04:49
you're convincing yourself that you have
00:04:50
all this control and ways to change the
00:04:52
outcome literally flipping a carard well
00:04:55
it's that's all it is high card it's
00:04:57
even worse than that you're basically
00:04:59
sitting down at the casino's table and
00:05:01
then they tell you whether You've Won or
00:05:02
lost and in order to convince yourself
00:05:05
that that's not what's going on you have
00:05:07
to play with the card but really they
00:05:10
just tell you you either win or lose and
00:05:13
I'm watching the smartest guys we know
00:05:15
staring at the window at the little
00:05:16
machine that tells you whether bank or
00:05:18
player one and they're studying it
00:05:20
rubbing their chin doing an analysis
00:05:23
it's
00:05:24
been black H is like I'm calling it now
00:05:27
Bank bank player player player and all
00:05:29
the guys like let's do it then
00:05:31
everyone so helth asked us to play in
00:05:34
the high stakes poker game on poker go
00:05:37
so it was me helmuth Stanley Sammy house
00:05:41
and then Jen Tilly and Nick airball and
00:05:44
Robel so most of the guys from the hus G
00:05:46
Plus gen Tilly and Nick airball Jennifer
00:05:49
til is amazing what a great human listen
00:05:51
to this well listen to this hand
00:05:54
literally the second hand of the actual
00:05:58
poker game
00:06:02
GTI is in the big blind no sorry she's
00:06:06
under the gun she raises Howen buold
00:06:09
three bets it comes all the way around
00:06:11
to me on the button I look and I have
00:06:14
pocket kings oh I ship the whole
00:06:17
cheeseburger comes back to Tilly she
00:06:20
ships house ships listen to these hands
00:06:23
Gentilly has Aces Jeff housenbold has
00:06:26
Kings I have Kings oh my God never seen
00:06:30
a cooler hand like this in my life and
00:06:33
the second in the second hand of the
00:06:35
game anyways wow don't worry guys I came
00:06:38
back and I won
00:06:40
G tripl up she triples up and then into
00:06:44
lockdown mode the first time I ever play
00:06:46
with her then I stacked her right I
00:06:48
anyways well I don't want to reveal the
00:06:50
game but it was it was wonderful it was
00:06:51
fun I show up at a mutual friend of ours
00:06:54
game and there's a like beautiful
00:06:56
Porsche or something in the driveways a
00:06:57
really notable car and the I I noticed
00:07:00
on the license plate it says Deen but
00:07:02
it's spelled with a J and I'm like oh
00:07:04
degenerate what a great license plate I
00:07:06
wonder who that is I go it's
00:07:08
Jennifer she is so cool she's very
00:07:11
Charming cool very Charming great
00:07:13
actress movie she was
00:07:16
in that's what it was yeah you don't
00:07:18
have to ask me twice yeah exactly tour
00:07:21
what a great gangster film yeah with uh
00:07:24
Gina gersin I mean Gina gchen that's who
00:07:28
that film that film oh my god well let's
00:07:30
not get canceled here okay yeah um it is
00:07:33
quite a film all right speaking of uh
00:07:37
action big week the AI industrial
00:07:40
complex is dominating our docket here
00:07:43
apologies to Biden Ukraine and Nikki Hy
00:07:45
but we got to go AI right now open AI
00:07:47
launch chat GPT 40
00:07:51
4.0 Monday three days after Sam wise
00:07:54
came on Allin as a programming note and
00:07:57
we'll go to freeberg about this we
00:07:59
probably made a bit of a strategical or
00:08:02
tactical error in not postponing his app
00:08:05
appearance In fairness uh freeberg
00:08:07
Samwise did tell us originally he was
00:08:09
coming on to talk about those things but
00:08:11
then it got pushed back anything you
00:08:12
want to add to that as a programming
00:08:14
note because people are wondering what
00:08:17
happened I've been talking with Sam for
00:08:19
a while a year about coming on the show
00:08:22
and every time I see him I'm we're like
00:08:23
hey you should come on the show he's
00:08:24
like I want to come on the show okay
00:08:25
let's find a date we never got a date
00:08:27
that worked I saw him in March and he
00:08:30
said hey I want to come on the show I
00:08:31
said okay well come on let me know when
00:08:33
works and a couple weeks later he's like
00:08:35
what about this date in May and I'm like
00:08:37
yeah that's that's fine we can make that
00:08:39
work he's like well I've got a big
00:08:41
announcement we're going to be doing and
00:08:43
I was like Perfect come on the show that
00:08:44
that that sounds great and then the um
00:08:47
the night before he asked me he told me
00:08:50
he texted me he's like hey we're
00:08:51
actually not gonna have this
00:08:52
announcement happen tomorrow it's going
00:08:54
to be delayed he didn't tell me how long
00:08:56
and I'm like well is it CH is it GPT 5
00:08:58
he's like no it's not gp5
00:09:00
and I was like okay well you know come
00:09:01
on the show anyway because he didn't
00:09:03
tell me when he's doing the announcement
00:09:04
or when it's being pushed to so it
00:09:05
didn't seem like that big a deal and I
00:09:07
thought we were just going to be able to
00:09:08
have a good chat anyway so it's really
00:09:09
unfortunate I think the fact that the
00:09:11
announcement happened two days after and
00:09:12
he had to stay quiet about it during our
00:09:14
interview but um that's the story I
00:09:16
think in the future if someone says
00:09:18
they've got a big announcement to do we
00:09:19
should probably push them uh if they if
00:09:21
they have to something don't be but I
00:09:24
don't think we're going to be doing a
00:09:25
lot of these interviews anyway I think
00:09:26
people clearly don't love them and it's
00:09:28
better for us to just kind of out and
00:09:30
talk I think I think if we had just
00:09:32
gotten Sam on the day after the launch
00:09:36
of GT4 Omni as opposed to what was it
00:09:39
three days before yeah he could have
00:09:41
talked much more freely about it and it
00:09:42
would have been interesting yeah it was
00:09:44
supposed to happen same day so it's
00:09:45
unfortunate this all worked out this way
00:09:47
the other little trick is to say you can
00:09:48
tell us under embargo but my
00:09:51
understanding is they were still doing
00:09:52
the videos o over the weekend so I think
00:09:55
those videos and stuff they were still
00:09:57
figuring them out and so yeah's learned
00:10:00
in terms of the interviews on the show
00:10:02
just a recap for people we've done a
00:10:04
dozen half of them have been
00:10:06
presidential candidates sometimes they
00:10:07
break out sometimes they don't we follow
00:10:10
our interest and our passion here on the
00:10:13
Pod it's got to be interesting for us
00:10:15
too so if we think this person's going
00:10:16
to be interesting we do it and yeah we
00:10:18
understand you miss a news subject but
00:10:20
yeah it is what it is and to your point
00:10:23
a lot of the people that come on and
00:10:25
increasingly a lot of people ask to come
00:10:27
on because they know we're not
00:10:28
journalists and so for all of those
00:10:30
folks that expect us to
00:10:33
be journalists that's not what we are
00:10:36
we're for entrepreneurs we're four
00:10:38
business people we're four friends we're
00:10:39
four technologists we're four curious
00:10:41
people we're four poker players but
00:10:44
we're not four journalists and so we're
00:10:46
going to ask whatever we feel like
00:10:48
asking sometimes those things will touch
00:10:50
a cord because it's what you wanted to
00:10:52
have asked and sometimes we won't go to
00:10:55
a place whether we didn't have time to
00:10:57
or whether we forgot or whether we chose
00:10:59
not to and I think it's important to
00:11:01
have that disclaimer like we have day
00:11:03
jobs and this is what we
00:11:05
do to Coal us a bunch of information in
00:11:08
the way that we're thinking about the
00:11:09
world so we are not journalists so
00:11:12
please don't have that I think what that
00:11:14
means is that if the guest doesn't want
00:11:16
to talk about something we're not going
00:11:18
to start peppering him with gotcha
00:11:20
questions and things like that I
00:11:22
appeared at a conference a couple of
00:11:24
days ago uh to promote glue which we'll
00:11:26
get to and the first half of the
00:11:28
conversation was like a normal
00:11:30
conversation about what we were
00:11:31
launching and then the second half was
00:11:32
basically the reporter peppering me with
00:11:35
fastball questions which is fine I knew
00:11:37
what I was signing up for it's a totally
00:11:38
different style it's a totally different
00:11:40
style than coming on the Pod just having
00:11:41
a normal conversation but it's not
00:11:43
really our job to make somebody open up
00:11:46
if they don't want to talk what was the
00:11:48
spiciest question sax that what was the
00:11:50
fast ball anything come close to your
00:11:51
head she no I mean it's not worth really
00:11:54
getting into you can watch it on yeah I
00:11:56
was just curious like look I I kind of
00:11:58
like sometimes when reporters pitch me
00:12:00
fast balls because yeah you can strike
00:12:02
out or you can hit out of the park when
00:12:03
they do that that's an important part
00:12:05
here I I think you know as a former
00:12:07
editor and chief journalist myself I
00:12:10
sometimes like to ask I would say a
00:12:12
challenging question in a respectful way
00:12:14
I did that for example vake you know
00:12:17
just clarifying his thoughts on trans
00:12:19
and gay rights wasn't disrespectful was
00:12:23
thoughtful would you consider it spicy
00:12:25
or hardcore I don't think it was
00:12:26
Hardcore he likes to talk about because
00:12:28
no but that's because you asked it from
00:12:30
a position of curiosity you weren't
00:12:31
trying to catch the guy no see the
00:12:33
difference I'm actually interested in
00:12:34
his opinion that's this is my point
00:12:36
that's why it comes out differently and
00:12:37
that's why I think people enjoy these
00:12:39
conversations and sometimes we don't get
00:12:41
to the other kind of answer because I'm
00:12:43
not interested in trying to got you
00:12:45
somebody that's working hard I always
00:12:47
have the same conditions when I do
00:12:48
interviews which is I don't clear
00:12:50
questions and I don't let people edit it
00:12:51
but you know everybody's got a different
00:12:53
view on how to do interviews and feel
00:12:56
differ if you like it you like it if you
00:12:57
like Lex Freedman version or Tim Ferris
00:13:00
version or you prefer you know fox or
00:13:03
CNN go go watch those interviews there
00:13:05
you can have a whole range of different
00:13:07
interviews and interview Styles
00:13:08
available to you in the media landscape
00:13:10
we are but one Sam Weiss mentioned on
00:13:12
the Pod last week that the next big
00:13:13
model might not be called GPT 5 so on
00:13:17
Monday they launched GPT 4 o uh the O
00:13:20
stands for Omni it's everything you love
00:13:23
about tech it's faster it's cheaper it's
00:13:25
better but uh from my perspective the
00:13:28
the the real show was the massive amount
00:13:30
of progress they made on the the
00:13:32
uiux the O stands for Omni as in
00:13:34
omnivore it takes in audio text images
00:13:37
even your desktop and video from your
00:13:39
camera to inform what it's doing you can
00:13:42
consider it like 360 degree AI producer
00:13:45
Nick will show a couple of videos while
00:13:47
I describe them here before we go to the
00:13:49
besties for their reaction to the
00:13:50
announcement first they made great
00:13:53
progress in solving the CB problem we
00:13:55
mentioned last week that's where like
00:13:56
when you use Siri or any of these tools
00:13:59
you say you know hey J GPT what's 2 plus
00:14:01
two over and you have to wait and then
00:14:02
if you talk over each other it breaks
00:14:05
they now have that working much smoother
00:14:08
they did an example of counting where
00:14:09
they said speed up slow down they did a
00:14:11
translator that worked really well I
00:14:14
would like you to function as a
00:14:15
translator I have a friend here who only
00:14:16
speaks Italian and I only speak English
00:14:19
and uh every time you hear English I
00:14:21
want you to translate it to Italian and
00:14:22
if you hear Italian I want you to
00:14:24
translate it back to English is that
00:14:26
good perfecto
00:14:37
Mike she wonders if whales could talk
00:14:40
what would they tell us um they might
00:14:42
ask uh how do we solve linear
00:14:55
equations certainly yes great looks like
00:14:59
works I think dingo stock took a hit
00:15:01
during that most impressive to me at
00:15:04
least and also I had Su MRA on my other
00:15:07
podcast and we talked about it was their
00:15:10
desktop and IOS app what this app does
00:15:12
is fascinating it watches your desktop
00:15:15
on your Macintosh or on your iPad or
00:15:18
iPhone and it
00:15:20
does things like adaptive learning
00:15:23
here's a clip of s KH from KH Academy
00:15:25
and his son basically uh using a drawing
00:15:29
app to do some arithmetic or or geometry
00:15:32
and it does adaptive learning and
00:15:34
basically makes a personal coach because
00:15:36
the app is watching his son try to solve
00:15:38
a problem I'm here with my son and I'd
00:15:41
love you to tutor him on this math
00:15:44
problem but don't give him the answer
00:15:46
you can ask questions and nudge him in
00:15:48
the right direction but I really want to
00:15:49
make sure he understands it himself and
00:15:52
he's here in the room so you can talk to
00:15:53
him directly can you first
00:15:56
identify which sides of the triangle are
00:15:59
the opposite adjacent and hypotenuse
00:16:03
relative to angle Alpha all right so I'm
00:16:07
pretty sure this is the angle Alpha
00:16:08
right here
00:16:10
right correct now looking at the
00:16:13
triangle which side do you think is the
00:16:16
hypotenuse
00:16:18
um remember the hypoten I'm not totally
00:16:22
sure I think I think it might be this
00:16:25
one but I really am not sure this side
00:16:27
AC
00:16:29
you're close actually side AC is called
00:16:32
the adjacent side to the angle Alpha the
00:16:36
hypotenuse is the longest side of a
00:16:38
right triangle and is directly opposite
00:16:40
the right angle can you find which one
00:16:43
is the hypotenuse oh okay I see so um I
00:16:47
think the hypotenuse is this really long
00:16:50
side from A to B would that be
00:16:52
correct exactly well done it can also
00:16:56
participate in Zoom calls explain charts
00:16:58
all that great stuff and so it's going
00:17:00
to be your guide on the side it's it's
00:17:02
going to be a present you know
00:17:04
personality while you're using your apps
00:17:06
it's really impressive I have to say so
00:17:08
I guess let's start freeberg with your
00:17:11
takeaways on all of these innovations
00:17:13
that we saw I think it's become quite
00:17:16
apparent that there's an evolution
00:17:19
underway in model
00:17:22
architecture we've and I think you may
00:17:24
remember we talked about this briefly
00:17:26
with Sam last week but we're moving away
00:17:27
from these very
00:17:29
big bulky models that are released every
00:17:33
couple of months or quarters and cost a
00:17:35
lot of money to rebuild every time they
00:17:37
get
00:17:38
re-released towards uh a system of
00:17:42
models so this multimodal system
00:17:44
basically leverages several models at
00:17:46
once that work together or that are
00:17:49
linked together uh to respond to the
00:17:52
inputs and to provide some generative
00:17:54
output and that those individual models
00:17:57
themselves can be continuously tuned and
00:18:00
or continuously updated so rather than
00:18:02
have you know hey there's this big new
00:18:04
release that just happened this new
00:18:05
model just got trained cost $10 million
00:18:07
to train it it's been pushed these
00:18:09
models can be upgraded um with tuning
00:18:12
with upgrade features and then link
00:18:14
together with other new smaller models
00:18:15
that are perhaps specialized for
00:18:17
specific tasks like doing mathematics or
00:18:19
rendering an image or rendering a movie
00:18:22
and so I think what we're going to see
00:18:23
is soon more of an
00:18:26
obfuscation of the individual
00:18:29
models and more of this general service
00:18:32
type approach where the updates are
00:18:34
happening in in in a more continuous
00:18:35
fashion I think this is the first step
00:18:37
of open AI taking that architectural
00:18:39
approach with uh GPT 40 and what's
00:18:43
behind the curtains we don't know we
00:18:45
don't know how many models are there we
00:18:46
don't know how frequently they're being
00:18:47
changed whether they're being changed
00:18:49
through actually upgrading the
00:18:50
parameters or whether they're being
00:18:52
fine-tuned and so this seems to be
00:18:54
pretty obvious if you look at this link
00:18:56
one of the criticisms that initially
00:18:59
came out when they released GPT
00:19:01
40 was that there was some performance
00:19:05
degradation and Stanford actually runs
00:19:08
this massive multitask language
00:19:10
understanding
00:19:12
assessment and they publish it I think
00:19:14
daily or pretty frequently on how all
00:19:16
the models perform and you can see the
00:19:17
scorecard here that GPT 40 actually
00:19:20
outperforms GPT 4 and so this goes
00:19:23
counter to some of the narrative that in
00:19:24
order to get some of the performance
00:19:26
improvements and speed improvements they
00:19:27
got in 40 that they actually made the
00:19:29
model worse and it seems actually the
00:19:31
opposite is true that the model's gotten
00:19:32
slightly better it still underperforms
00:19:34
Cloud 3 Opus which uh you can see here
00:19:37
ranks top of these charts but there's
00:19:38
lots of different charts all the
00:19:39
companies publish their own charts they
00:19:41
all claim that they're better than
00:19:42
everyone else but I like Stanford
00:19:43
because it's independent sh any thoughts
00:19:45
uh after seeing it and in combination
00:19:47
with our interview do you think chat GPT
00:19:50
is running away with the consumer
00:19:52
experience or do you think this is like
00:19:55
neck and neck with some of the other
00:19:56
players not to tell Tales out of school
00:19:58
but somebody that we all know in our
00:20:00
group chat hosted something about the
00:20:03
fact that the consumer growth had
00:20:05
stalled I don't know how they knew that
00:20:07
that they maybe they got some data or
00:20:10
maybe they're an investor you you guys
00:20:12
know what I'm talking about and and they
00:20:14
said that they're trying to reinvigorate
00:20:15
growth into the consumer app into at
00:20:18
open AI I mean any insights as to why it
00:20:21
might be plateauing in your
00:20:22
perspective I wrote this in my annual
00:20:25
letter but there are these phases
00:20:29
growth and when you look at like social
00:20:32
networks as a perfect example frster was
00:20:34
magical when it was first created right
00:20:37
and then you had Myspace that just ran
00:20:39
circles around them because frer didn't
00:20:41
really invest the money and the quality
00:20:44
that it took to to create a
00:20:47
Moe and then Myspace really wasn't able
00:20:49
to compete so we were you know Facebook
00:20:51
we were the eighth or ninth when we
00:20:52
showed up on the scene and we ran
00:20:54
circles around
00:20:55
everybody I think what it means is that
00:20:58
there are these phases of product
00:20:59
development
00:21:01
which exist in many markets this Market
00:21:04
I think is going through the same thing
00:21:06
and right now we're in the first what I
00:21:07
would call primordial ooze phase which
00:21:10
is everybody's kind of like running
00:21:11
around like a chicken with their heads
00:21:13
cut off there's all these core basic
00:21:16
capabilities that are still so magical
00:21:17
when you see them but we all know that
00:21:19
five and 10 years from now these things
00:21:21
will be table sticks right and what
00:21:23
freeberg just showed is a
00:21:25
table of many companies and many
00:21:28
trillions of market cap all effectively
00:21:31
running to the same destination so I
00:21:33
think where we are is probably within
00:21:35
two years of where the basic building
00:21:38
blocks are
00:21:39
standardized and then I think the real
00:21:41
businesses get built so I will maintain
00:21:44
my perspective here which is the quote
00:21:46
unquote Facebook of AI has yet to be
00:21:49
created okay and here it is chat BT web
00:21:53
visits as you can see have plateaued
00:21:55
this data is similar web I would agree
00:21:58
with you jamath it seems like the use
00:22:00
cases and the lookie L who were just
00:22:04
trying this software because they heard
00:22:05
about it they've gone away and then we
00:22:08
have to find actual use cases Sachs I'm
00:22:10
wondering but our friend Jason just to
00:22:12
kind of complete the thought said
00:22:13
something about the premium conversion
00:22:15
right that's what he said I don't know
00:22:16
how he knew that paid paid version to be
00:22:18
clear paid versus free and then what Sam
00:22:21
said on the podcast last week was it
00:22:23
seems like whenever they come out with
00:22:25
something new the old stuff becomes free
00:22:27
in my talk with Sunny this week he
00:22:29
mentioned that these new models are so
00:22:32
much more efficient that you actually
00:22:34
can throw the old model in the garbage
00:22:36
garbage because it's so inefficient and
00:22:39
these are now becoming about 90% cheaper
00:22:42
every year which means every two years
00:22:44
these things are going be 99% cheaper
00:22:45
and better y yep and it might open AI
00:22:50
sacks on a strategic level is going to
00:22:53
make all this free or close to free and
00:22:56
maybe just charge for multiplayer
00:22:58
version
00:22:59
that seems to be where it's heading you
00:23:00
don't have to log in to use 3.5 you
00:23:02
don't have to log in to use uh Google
00:23:05
Ser no you do have to log in still on
00:23:07
Google services but I think these are
00:23:08
going to just be free so on a product
00:23:10
basis what are your thoughts and then
00:23:11
maybe could talk about free to pay do
00:23:13
you think everybody in the world is
00:23:14
going to pay 20 30 40 bucks 500 a year
00:23:17
200 a year to have one of these or are
00:23:19
they just going to all be free well I
00:23:21
think you're assuming there that the
00:23:23
long-term business model of open AI is
00:23:25
in you to see subscriptions and I think
00:23:27
that's probably the least attractive
00:23:30
business model they have available to
00:23:31
them it's sort of the first one and the
00:23:33
most obvious one because they put out
00:23:35
chat GPT and then it's pretty easy just
00:23:37
to roll out a premium version but in my
00:23:40
experience BTC subscriptions it's just
00:23:42
not a very attractive business model
00:23:44
because consumers just aren't willing to
00:23:45
pay a lot and they have high churn rates
00:23:48
and there's no possibility of expansion
00:23:50
really so I suspect they're going to
00:23:52
move in more of a B2B Direction over
00:23:54
time because that's where the real money
00:23:55
is and probably the way they do that is
00:23:58
is by uh monetizing all the apps that
00:24:01
are built on top of it and I think that
00:24:03
in that sense GPT
00:24:05
40 is a really important
00:24:09
innovation by the way the the O stands
00:24:11
for Omni which I think stands for Omni
00:24:13
Channel I think you may have said
00:24:15
omnivore kind yes it's Omni yeah which
00:24:18
means all the different media types
00:24:20
current currently coming in right like
00:24:23
there that's the difference it's not
00:24:24
like you just give it an image or give
00:24:25
it a video it's absorbing all those at
00:24:27
the same time in par I believe that's
00:24:29
right so there's three big Innovations
00:24:31
with this model right so one is Omni
00:24:34
Channel which means text audio video and
00:24:36
images second it's more conversational
00:24:40
like it understands the tone of people
00:24:43
talking it understands sort of sentiment
00:24:45
in a way it didn't before and then the
00:24:47
third thing which is really important is
00:24:48
that it's just much faster more
00:24:50
performant than the previous version gp4
00:24:53
turbo in the speed test they say it's
00:24:55
twice as fast we've played with it at
00:24:57
glue we can talk about that in a minute
00:24:58
and it feels 10 times as fast it is much
00:25:00
faster but it's the combination of all
00:25:03
three of these things that really makes
00:25:05
some magical experience is possible
00:25:07
because when you increase the speed of
00:25:10
processing you can now actually have
00:25:12
conversations with it in a much more
00:25:13
natural way before it was the the models
00:25:16
were just too slow so there'd be a long
00:25:18
delay after every prompt yeah so now
00:25:22
like you showed it can do things like
00:25:24
you point the camera at a a Blackboard
00:25:26
or something with math equations on it
00:25:28
it and it can walk you through how to
00:25:30
solve that problem or two people can be
00:25:33
talking and it does real-time
00:25:35
translation you know there's that old
00:25:37
saying that every Star Trek technology
00:25:39
eventually becomes true they've just
00:25:40
basically invented the whole natural
00:25:42
language real time yes Universal
00:25:44
translator yeah it's so anyway so those
00:25:47
are some interesting use cases but I
00:25:49
just think they're going to be able to
00:25:50
unleash a whole lot of new applications
00:25:54
and if they're metering the usage of the
00:25:57
models and providing the best Dev tools
00:25:58
I think there is a business model there
00:26:00
this thing is moving so FAS they in like
00:26:02
Leonardo DiCaprio mode every two years
00:26:04
they throw the old model
00:26:06
away okay let's uh keep thank you sax a
00:26:10
b is this thing did you write that ahead
00:26:13
of time or in the
00:26:16
moment that good just just one one point
00:26:18
on that is there are a whole bunch of of
00:26:22
startups out there that were creating
00:26:24
virtual customer support agents and they
00:26:28
been spending the last couple of years
00:26:29
working on trying to make those agents
00:26:32
more conversational quicker more
00:26:35
responsive I think their product road
00:26:37
maps just became obsolete now that's not
00:26:39
to say there isn't more work for them to
00:26:41
do in workflow in terms of integrating
00:26:44
the AI with customer support tools and
00:26:47
doing that last mile of customizing the
00:26:51
model for the vertical specific problems
00:26:53
of customer support but my guess is that
00:26:57
hundreds of Millions dollars of R&D just
00:27:00
went out the window and probably this is
00:27:03
the best time to be creating a customer
00:27:04
support agent company if you're doing it
00:27:06
two years ago five years ago your work
00:27:08
has just like been well that is the
00:27:10
thing of this pace like you know you
00:27:12
used to have to throw away client server
00:27:14
stuff or you know whatever you had a web
00:27:16
based thing you you get an app out you
00:27:17
throw away some of the old code but this
00:27:19
is like every 18 months your work has
00:27:21
been replaced if you're an app developer
00:27:23
the key thing to understand is where
00:27:26
does model Innovation end and your
00:27:28
Innovation begin because if you get that
00:27:31
wrong you'll end up doing a bunch of
00:27:33
stuff that the model will just obsolete
00:27:35
in a few mons think you're totally right
00:27:37
I think that's such a really important
00:27:38
observation that's why I think the
00:27:40
incentive for these folks is going to be
00:27:42
to push this stuff into the open source
00:27:44
because if you if you solve a problem
00:27:47
that's operationally necessary for your
00:27:49
business but it isn't the core part of
00:27:51
your
00:27:52
business what incentive do you have to
00:27:54
really keep investing in this for the
00:27:55
next 5 and 10 years to improve it you're
00:27:57
much better off like Clara for example
00:27:59
right we talked about the in The Amazing
00:28:02
Improvement and savings that Clara had
00:28:04
by improving customer support release it
00:28:07
in the open source guys let the rest of
00:28:09
the community take it over so that it's
00:28:10
available to everybody else otherwise
00:28:13
you're going to be stuck supporting it
00:28:14
and then if and when you ever wanted to
00:28:16
switch out a model you know GPT 40 for
00:28:20
to 40 to Claude to llama it's going to
00:28:23
be near impossible and it's going to be
00:28:24
costly so I I also think SX the
00:28:27
incentive to just push towards open
00:28:29
source in this market if you will is so
00:28:33
much more meaningful than any other
00:28:34
Market yeah I mean listen you you were
00:28:37
there when I I think you were there at
00:28:38
Facebook when they did the open compute
00:28:40
project and they just were
00:28:42
like talk about talk about torching an
00:28:44
entire Market explain what it is so
00:28:47
there was this moment where when you
00:28:50
were trying to build data centers you'd
00:28:52
have these like oneu rack mounted kind
00:28:54
of like machines that that you used and
00:28:57
what Facebook obs OB ered was there was
00:28:58
only a handful of companies that
00:29:00
provided it and so it was unnecessarily
00:29:02
expensive and so Facebook just designed
00:29:04
their own and then release the specs
00:29:06
online just kind of said here it is and
00:29:08
they went to these Taiwanese
00:29:10
manufacturers and other folks and said
00:29:12
please make these for your Cost Plus a
00:29:14
few bucks and it was revolutionary in
00:29:17
that market because it allowed this open
00:29:19
platform to sort of embrace this very
00:29:22
critical element that everybody needs
00:29:24
and and I think there's going to be a
00:29:25
lot of these examples
00:29:28
inside of AI because the costs are so
00:29:31
extreme so much more than just building
00:29:33
a data center for a traditional web app
00:29:35
that the incentives to do it are just so
00:29:37
so
00:29:38
meaningful yeah and I just showed it on
00:29:40
the screen sax you've actually been
00:29:42
dancing along this line last night I was
00:29:43
using your new slack killer or coexist
00:29:47
I'm not sure it feels like a slack
00:29:48
killer to me because I'm moving my
00:29:49
company to it on over the weekend we're
00:29:51
moving to glue and you and I were doing
00:29:54
some very I think I may need to wet my
00:29:56
beak on this one we want you to W your
00:29:59
beak it feels like aund bagger to me uh
00:30:02
I'm in killer that's the way we're
00:30:04
thinking about it killer ask because Jal
00:30:07
jcal can you do that again in
00:30:08
Christopher Walkin voice please I get to
00:30:10
wet my beak it's like a 100x sliding 500
00:30:16
wow saxs tell me about product decisions
00:30:21
where does the AI Ed and your product
00:30:24
begin yeah well it's a good point I mean
00:30:27
I I think where the AI ends we want to
00:30:30
use the most powerful AI models possible
00:30:32
and we want to focus on Enterprise chat
00:30:35
so you could think of us as for sure a
00:30:37
slack killer or slack competitor it say
00:30:40
that slack wasn't built for the AI era
00:30:42
glue is AI native what does that mean no
00:30:45
channels you know I showed this to jth
00:30:47
the first thing you said is you add me
00:30:48
at no channels right people are so sick
00:30:50
of channels you have to keep up with all
00:30:52
these hundreds and hundreds of channels
00:30:53
and the real problem with channels is
00:30:55
there's one thread in a channel that you
00:30:57
want to see in order to see it you have
00:30:59
to join the whole channnel and now
00:31:00
you're getting all this noise people
00:31:02
just want the threads so if you look at
00:31:04
what's the chat model inside of chat GPT
00:31:06
it's just threads right you create a
00:31:09
topic based thread in chat GPT the AI
00:31:12
comes up with a name for it puts it in
00:31:14
the sidebar and then if you want to talk
00:31:16
about something else you create a new
00:31:17
chat that's exactly the way that glue
00:31:19
works it's just multiplayer you just put
00:31:21
the groups and individuals you want on
00:31:23
the thread let me just show you real
00:31:25
quick here's my uh glue here and you can
00:31:27
see that in the sidebar I've got all the
00:31:29
threads that I've been involved in and
00:31:31
like I said you can address them to
00:31:32
multiple people or groups and then
00:31:34
you've got the chat here now we've also
00:31:37
fully integrated Ai and so Nick who's
00:31:40
our producer just in this thread said at
00:31:42
glue AI what countries to saaks talk
00:31:44
about most in episodes episodes is a
00:31:47
group we created to be the repository of
00:31:49
all of the transcripts of our
00:31:51
episodes and so glue did a search and it
00:31:55
said David Sachs frequently discusses
00:31:57
Ukraine why the most yeah really so then
00:32:00
so then Nick said be more specific about
00:32:02
Sach stance on Ukraine Russia War oh boy
00:32:05
and it's gonna overload the server well
00:32:08
here it said here David sax has
00:32:10
articulated a nuance and critical
00:32:11
perspective on the Ukraine Russia War
00:32:12
across various episodes of the all and
00:32:14
pod here are some key points
00:32:15
encapsulating his stance and it like
00:32:18
nailed it it's talked about prevention
00:32:20
through diplomacy opposition to Nato
00:32:22
expansion humanitarian concerns
00:32:24
skepticism military intervention peace
00:32:26
de proposal you know I'll I'll copy and
00:32:28
paste this onto Twitter X later today
00:32:31
but the point is it like nailed it
00:32:33
across all these different episodes and
00:32:35
then this a feature of glue it provided
00:32:38
sources so it CES where it got all the
00:32:40
information from so imagine you know
00:32:43
we're we're doing this for the all-in
00:32:45
Pod but you could imagine that instead
00:32:46
of it being transcripts of a podcast it
00:32:49
could be your work documents you now
00:32:52
have in your main chat the ability just
00:32:53
to ask hey act glue AI remind me where
00:32:58
we left that project or tell me who the
00:33:00
expert is on this subject matter or
00:33:02
who's contributed the most to this
00:33:03
project I've actually figured out using
00:33:05
glue AI who's contributed the most deal
00:33:07
flow at craft is pretty amazing now let
00:33:10
me show you like some the answer what's
00:33:12
the answer I'm not gonna say here I
00:33:15
think it's Brian Brian's my guy those
00:33:17
two my guy I want if I got go Daniel and
00:33:22
Brian boom bo that's how I get sacks on
00:33:24
the line okay you take Daniel I'll hire
00:33:26
Brian
00:33:28
reconstitute craft we'll call
00:33:33
itft all right look let me let me show
00:33:35
you a
00:33:35
cple SX I just looked at the um chat jp4
00:33:39
Omni server and when you did that query
00:33:42
it actually rolled its eyes keep going
00:33:44
okay hell so so we talked about how chbt
00:33:48
40 understands sentiment in
00:33:50
conversations and I think this is a
00:33:52
really good example so Evan who's my
00:33:54
co-founder at glue was just testing this
00:33:56
for us and he asked what is each besties
00:33:59
personality type based on conversations
00:34:01
and episodes and first of all it figured
00:34:03
out who the besties are like we didn't
00:34:05
program that it just figured it out on
00:34:07
its own which is pretty impressive and
00:34:09
here's what I said about each bestie's
00:34:11
personality so it says Jason calanis
00:34:13
role the moderator and Entertainer
00:34:16
personality traits charismatic humorous
00:34:18
Lively and the glue of the group often
00:34:20
praise often pra for his ability to make
00:34:22
the podcast entertaining and keep the
00:34:24
conversation flowing Jason's energy and
00:34:26
ability to bring humor and disc discusss
00:34:27
or key elements that Define his role in
00:34:29
the podcast comments he's acknowledged
00:34:30
for his role in bringing the group
00:34:31
together jamaath ptia the Visionary and
00:34:34
Hustler personality TR traits insightful
00:34:36
driven persistent Chamas Relentless
00:34:38
drive and ability to hustle admire by
00:34:40
your peers Mak him a core part of the
00:34:41
podcast intellectual discussions I'll
00:34:43
skip me but does freeberg the pragmatic
00:34:45
scientist and realist pragmatic
00:34:47
methodical and a bit reserved oh
00:34:49
freeberg is often the voice of reason
00:34:51
bringing a scientific and realistic
00:34:52
perspective the discussion he focuses on
00:34:55
analysis logical reasoning
00:34:58
and then it CES where it got this from
00:34:59
and it says here overall the dynamic
00:35:00
between the four besties creates a
00:35:02
well-rounded and engaging podcast with
00:35:04
each member bringing their unique
00:35:05
strengths and personality traits at the
00:35:06
table I think that's pretty incredible
00:35:08
how woke is this have you uh have you
00:35:11
put any rails on or is this just pure
00:35:14
chat GPT 40 combined with the data yeah
00:35:18
yeah so what we're doing here is we're
00:35:19
wrapping chat GPT 40 with glue features
00:35:24
that we've implemented to get the most
00:35:26
out of the conversation there's things
00:35:27
we have to do to scope the The Prompt
00:35:31
and then we're using a retrieval
00:35:33
augmented generation service called
00:35:36
raggi which does rag as a service that
00:35:38
basically slurps in our transcripts and
00:35:41
makes them accessible to the AI so um
00:35:44
that's basically the stack that we're
00:35:45
using but as the models get better and
00:35:47
better glue just gets better and better
00:35:48
again can I can I just make a comment on
00:35:50
this it's just so clean
00:35:52
jcal was the key for me in abandoning
00:35:56
slack he told me
00:35:58
two or three years ago he called me and
00:36:01
he said I have you can tell me the exact
00:36:04
channels I eliminated some channels at
00:36:05
work
00:36:06
random there was like two or three
00:36:08
channels at your your slack instance
00:36:11
wasn't allowed to have and I was like
00:36:13
this is genius and I went in and I was
00:36:15
like all of our companies should just
00:36:16
eliminate these channels and we could
00:36:18
only get like 20% or 30% compliance but
00:36:22
it really started to turn me off slack
00:36:23
because I would get caught in these
00:36:24
threads that were just so totally
00:36:28
useless and I thought why aren't people
00:36:30
working and this is really great because
00:36:32
you cannot BL on about nonsense in glue
00:36:35
which I find really useful well this is
00:36:36
what happens when slack we used it at
00:36:38
8090 just so you know so we were the we
00:36:41
got into the early get into slack too
00:36:43
much people start to think slack is the
00:36:45
job and replying to slacks and having
00:36:47
conversations is the job when there's
00:36:49
actually a job to be done there's a job
00:36:50
to be done yeah and so it's important
00:36:53
and what I liked about this
00:36:53
implementation saxs was it's like the
00:36:56
ability to make a feed or a data source
00:37:00
inside of your communication platform so
00:37:03
the fact that you imported all of the
00:37:05
episodes and their transcripts is great
00:37:07
but what I want is like our HubSpot or
00:37:09
our cell CRM I want our zenes I want our
00:37:14
LinkedIn jobs and our LinkedIn job
00:37:16
applications I want our notion I want
00:37:18
our Koda to each have the ability and
00:37:20
when I was using it last night what you
00:37:22
do is you use the at symbol to evoke and
00:37:26
to summon in a way it's like summoning
00:37:28
Beetle Juice So you summon your AI but
00:37:31
then you tell it what data set you want
00:37:33
to go after so you say you know at AI
00:37:37
let's talk about I don't know how do you
00:37:40
manage your deal flow at craft do you
00:37:42
use software like CRM software to manage
00:37:44
deals Brian we just do it all glue but
00:37:47
we do it all in glue so it's already
00:37:48
right there but you're right so so the
00:37:51
the first thing that glue AI has access
00:37:52
to is all of your chat history which is
00:37:54
amazing because you get like you know
00:37:57
then can look at all your attachments
00:37:59
and we've got I think six Integrations
00:38:01
at launch and there'll be more so yeah
00:38:02
like all of your Enterprise data will be
00:38:04
there in the short term you're right you
00:38:05
have to summon the repository by app
00:38:07
mentioning because the AI needs a little
00:38:09
bit of help of where to look but in the
00:38:11
future it's going to figure it out on
00:38:12
its own so it's just GNA become more and
00:38:14
more seamless but it'll insert itself so
00:38:16
we have a discussion about sales and
00:38:18
then you might have a sales bot that
00:38:20
says hey by the way nobody's called this
00:38:22
client in three months well that's where
00:38:24
I want to go with it is I call that
00:38:25
prompt lless which is want the AI just
00:38:28
to chime in when it determines that it
00:38:30
has relevant information and can help
00:38:32
the team even if it hasn't been summoned
00:38:34
yet but we need some model Improvement
00:38:36
for that frankly I mean we'll be able to
00:38:38
get there by GPT 5 but that's totally
00:38:40
where this is headed I'll show you just
00:38:42
one more fun example if I could let me
00:38:44
just show you this so I asked it to to
00:38:47
write a letter to Lena KH to be a guest
00:38:51
at the all-in summit and I told it
00:38:54
mention positive things we've said about
00:38:56
Lena KH
00:38:58
in episodes of the all-in Pod and so it
00:39:01
wrote this letter dear chair con we hope
00:39:04
this message finds you well on behalf
00:39:05
the host the all in pod we excited an
00:39:07
invitation for you to speak at the
00:39:08
upcoming all in Summit and then it says
00:39:11
in our conversations we have frequently
00:39:13
highlighted your impressive credentials
00:39:14
and the impactful work you've undertaken
00:39:16
for example in episode 36 we acknowledge
00:39:19
your trailblazing
00:39:20
role and so the letter was able to quote
00:39:23
episodes of the all-in Pod just without
00:39:26
anyone having to go do that research and
00:39:28
figure out like what would be the best
00:39:29
because I told it only say positive
00:39:31
things don't say anything negative and
00:39:33
then and then it said warmer guards and
00:39:35
it said who the four besties were again
00:39:36
we never told it who the besties are we
00:39:38
just said write us a letter so it's
00:39:41
pretty incredible now this is just an
00:39:43
example with the all-in Pod think about
00:39:45
any work context where the AI has access
00:39:48
to your previous work documents it's
00:39:50
pretty amazing what it can do well I
00:39:53
mean it is kind of in the name like this
00:39:55
is glue put you together and slack is
00:39:57
where you slack up makes total sense the
00:40:00
brands give you a little bit of a tip we
00:40:01
should have seen it coming with slack
00:40:06
totally we have a breaking news story
00:40:09
it's a breaking news story it's an Allin
00:40:11
exclusive today on the program I got
00:40:13
breaking news coming in fredberg your
00:40:16
life's work saxs did his uh product
00:40:19
review now it's your turn freeberg we
00:40:22
got breaking news coming in I did
00:40:23
promise you that when ohal decides to
00:40:26
come out of St he and explains what
00:40:29
we've done and what we're doing I would
00:40:31
do it here on the all-in Pod first
00:40:34
before the in exclusive all in exclusive
00:40:38
so basically by the time this pod airs
00:40:40
we're gonna be
00:40:43
announcing what ohal has been developing
00:40:46
for the past five years and has had an
00:40:48
incredible breakthrough in which is
00:40:49
basically a new technology in
00:40:52
agriculture and we call it boosted
00:40:54
breeding I'm going to take a couple
00:40:55
minutes just to talk through what we
00:40:58
discovered or invented at ohal and why
00:41:01
it's
00:41:02
important and the kind of significant
00:41:04
implications for it but basically five
00:41:07
years ago we had this theory that we
00:41:10
could change how plants reproduce and in
00:41:14
doing so we would be able to allow
00:41:17
plants to pass 100% of their genes to
00:41:21
their offspring rather than just half
00:41:23
their genes to their offspring and if we
00:41:25
could do that then all the genes from
00:41:27
the mother and all the genes from the
00:41:28
father would combine in The Offspring
00:41:31
rather than just half the genes from the
00:41:32
mother and half the genes from the
00:41:34
father and this would radically
00:41:36
transform crop yield and improve the
00:41:39
health and the size of the plants which
00:41:41
could have a huge impact on agriculture
00:41:44
because yield the size of the plants
00:41:45
ultimately drives productivity per acre
00:41:48
revenue for Farmers cost of Food calorie
00:41:50
production sustainability Etc so this
00:41:53
image just shows generally how
00:41:54
reproduction Works you've got two
00:41:56
parents
00:41:57
you get a random selection of half of
00:42:00
the DNA from the mother and a random
00:42:03
selection of half the DNA from the
00:42:04
father so you never know which half
00:42:05
you're going to get from the mother or
00:42:06
which half you're going to get from the
00:42:07
father that's why when people have kids
00:42:10
every kid looks different and then those
00:42:12
two halves come together and they form
00:42:13
The Offspring so every time a new child
00:42:16
is born every time a plant has Offspring
00:42:19
you end up with different genetics and
00:42:22
this is the problem with plant breeding
00:42:23
let's say that you have a bunch of gen
00:42:26
in one plant that are disease resistant
00:42:27
a bunch of genes in the other plant that
00:42:29
are drought resistant and you want to
00:42:31
try and get them together today the way
00:42:33
we do that in agriculture is we spend
00:42:35
decades trying to do plant breeding
00:42:37
where we try and rind all these
00:42:38
different crosses find the ones that
00:42:40
have the good genes find the other ones
00:42:41
that have the good genes and try and
00:42:42
keep combining them and it can take
00:42:44
forever and it may never happen that you
00:42:46
can get all the good genes together in
00:42:48
one plant to make it both disease
00:42:50
resistant and Dr resistant so what we
00:42:53
did is we came up with this theory that
00:42:55
we could actually change the genetics of
00:42:57
the parent plants we would apply some
00:43:00
proteins to the plants and those
00:43:02
proteins would switch off the
00:43:04
reproductive circuits that cause the
00:43:07
plants to split its genes and as a
00:43:09
result the parent plants give 100% of
00:43:12
their DNA to their offspring so The
00:43:15
Offspring Have double the DNA of either
00:43:17
parent you get all the genes from the
00:43:19
mother all the genes from the father and
00:43:21
finally after years of toiling away at
00:43:24
trying to get this thing to work and all
00:43:25
these experiments and all these
00:43:26
approaches is we finally got it to work
00:43:29
and we started collecting data on it and
00:43:31
the data is ridiculous like the Y on
00:43:34
some of these plants goes up by 50 to
00:43:36
100% or more just to give you a sense
00:43:39
like in the the corn seed
00:43:41
industry breeders that are breeding corn
00:43:43
or spending $3 billion doar a year on
00:43:45
breeding and they're getting maybe one
00:43:47
and a half% yield gain per year with our
00:43:49
system we we are seeing 50 to 100% jump
00:43:52
in the size of these plants it's pretty
00:43:53
incredible here's an example this is a
00:43:55
little weed that we that you do exper
00:43:57
with in agriculture called a rabid dosis
00:44:00
so it's really easy to work with and you
00:44:01
can see that what we have on the top are
00:44:03
those two parents A and B and then we
00:44:05
applied our boosted technology to them
00:44:08
and combined them and we ended up with
00:44:09
that Offspring called boosted ABS you
00:44:11
can see that that plant on the right is
00:44:12
much bigger it's got bigger leaves it's
00:44:13
healthier looking Etc fre ask you a
00:44:15
question does that mean that the boosted
00:44:17
one has twice the number of chromosomes
00:44:19
As A and B exactly right so is that like
00:44:23
a new species then yeah so um it's with
00:44:27
twice the chromosomes yeah it's it's
00:44:29
called poly so we actually see this
00:44:31
happen from time to time in nature for
00:44:34
example humans have two sets of
00:44:36
chromosomes right so does corn so do
00:44:38
many other species somewhere along the
00:44:41
evolutionary history
00:44:43
wheat doubled and then doubled again and
00:44:46
you end up actually in wheat having six
00:44:48
sets of chromosomes wheat is what's
00:44:51
called a hexaploid potatoes are a tetrol
00:44:53
they have four sets of chromosomes and
00:44:55
strawberries are an octoploid they have
00:44:57
eight and some plants have as many as 24
00:44:59
sets of chromosomes so certain plant
00:45:01
species have this really weird thing
00:45:03
that might happen from time to time in
00:45:05
evolution where they double their their
00:45:06
DNA naturally and so what we've
00:45:08
effectively done is just kind of Applied
00:45:11
a protein to to make it happen and bring
00:45:13
the correct two plants together when we
00:45:15
make it happen and so this this could
00:45:17
only happen for a plant right this could
00:45:18
never happen with an animal it wouldn't
00:45:20
it wouldn't work in animals it works in
00:45:22
Plants okay and one way you can think
00:45:23
about plant genetics is all the genes
00:45:27
are sort of like tools in a toolbox the
00:45:29
more tools you give the plant the more
00:45:31
it is it has available to it to survive
00:45:33
in any given second to deal with drought
00:45:36
or hot weather or cold weather Etc and
00:45:39
so every given second the more tools or
00:45:41
the more genes the plant has that are
00:45:43
beneficial the more likely it is to keep
00:45:44
growing and keep growing and that plays
00:45:46
out over the lifetime of the plant with
00:45:48
bigger bigger leaves and bigger you know
00:45:50
grows taller but more importantly if you
00:45:52
look at the bottom the seeds get bigger
00:45:53
and in most crops what we're harvesting
00:45:55
is the seed that's true and you know
00:45:57
corn and many other crops and so seeing
00:46:00
over a 40% increase in seed in this
00:46:01
little weed was a really big deal but
00:46:04
then we did it in potato and potato is a
00:46:06
crazy result potato is the third largest
00:46:08
source of calories on Earth and so we
00:46:10
took two potatoes that you see here in
00:46:12
the middle a and CD we applied our
00:46:14
boosted technology to it to each of them
00:46:17
and put them together and you end up
00:46:18
with this potato ABCD that's the boosted
00:46:20
potato and as you can see these were all
00:46:22
planted on the same date and the boosted
00:46:25
potato is much bigger than all the other
00:46:27
potatoes here including a market variety
00:46:29
that we show on the far right that's
00:46:31
what's typically grown in the field now
00:46:32
here's what's most important when you
00:46:34
look under the ground and you harvest
00:46:35
the potatoes you can see that that AB
00:46:37
potato only had 33 grams CD had n grams
00:46:41
so each parent had 33 and9 grams potato
00:46:45
but the boosted Offspring had 682 grams
00:46:48
of potato the yield gain was insane and
00:46:51
so you could see this being obviously
00:46:53
hugely beneficial for Humanity potatoes
00:46:57
being the third largest source of
00:46:58
calories Indian potato farmers are
00:47:01
growing one acre of potato in India they
00:47:03
eat potato two meals a day in Africa
00:47:06
potato is a food staple so around the
00:47:09
world we've had a really tough time
00:47:10
breeding potatoes and improving the
00:47:11
yield with our system we've seen
00:47:13
incredible yield gains in potato almost
00:47:15
overnight and the other big are those
00:47:17
potatoes those are normaliz potatoes
00:47:19
that you see there those are like you
00:47:21
know table potatoes basically that looks
00:47:23
like a rusted potato right there that's
00:47:25
like a normal siiz rust
00:47:28
it started like a little creamer potato
00:47:29
basically and you blew it up into a
00:47:31
rusted potato is that yeah so the
00:47:33
genetics on AB you can see they're like
00:47:36
little purple tiny little purple
00:47:38
potatoes the genetics on CD are like
00:47:39
these little white you know tiny little
00:47:41
ball potatoes but when you put those two
00:47:43
together with boosted and you combine
00:47:45
all the DNA from Ab and all the DNA from
00:47:47
CD you get this crazy high yielding
00:47:49
potato ABCD which by the way is higher
00:47:51
yielding than the market variety that's
00:47:53
usually grown in the field on the far
00:47:55
right so why not just grow russet
00:47:56
potatoes then we are and so we're
00:47:59
working on doing this with russet we're
00:48:00
working on doing this with every major
00:48:02
potato line sorry um the the the the
00:48:04
Improvement you'll see is actually in
00:48:05
yield so it's not the size of the potato
00:48:07
it's the number of potatoes that are
00:48:08
being made um and so you'll see he acre
00:48:11
or something like that like the exactly
00:48:13
you know projects in the 60s and 70s sh
00:48:15
you know how you can tell freeberg is
00:48:17
onto something here he got David saaks
00:48:19
to pay attention during
00:48:21
it this is a this gonna be a deck oford
00:48:23
if saak is awake saak is like how do I
00:48:25
wet my be s is interrogating the potato
00:48:28
lines I've never what's going on I think
00:48:30
J is interesting but so have you tried
00:48:32
these potatoes do they taste different
00:48:34
oh no they're awesome yeah they're
00:48:36
they're potatoes and we do a lot of
00:48:38
analysis sprouted any horns yet or
00:48:40
anything like that
00:48:41
no I mean again one of the other
00:48:44
advantages of the system that we've
00:48:46
developed let me go back here and I just
00:48:48
want to take two seconds on this one of
00:48:49
the other things this unlocks is
00:48:52
creating actual seed that you can put in
00:48:55
the ground in CRS that you can't do that
00:48:57
in today so potatoes the third largest
00:48:59
source of calories but the way we grow
00:49:01
potatoes you guys remember the movie The
00:49:02
Maran you chop up potatoes and you put
00:49:04
them back in the ground because the seed
00:49:07
that comes out of a potato which grows
00:49:08
on the top in the flow every one of
00:49:10
those seed is genetically different
00:49:12
because of what I just showed on this
00:49:13
chart right you get half the DNA from
00:49:15
the mother half the DNA from the father
00:49:16
so every seed has different genetics so
00:49:18
there's no potato seed industry today
00:49:20
and potato is like a hundred billion
00:49:22
Doll Market with our system not only can
00:49:25
we make potatoes higher yield and make
00:49:27
them disease resistant what we also make
00:49:30
is perfect seed so farmers can now plant
00:49:33
seed in the ground which saves them
00:49:34
about 20% of Revenue takes out all the
00:49:36
disease risk and makes things much more
00:49:38
affordable and easier to manage for
00:49:39
Farmers so it creates entirely new seed
00:49:42
Industries so we're going to be applying
00:49:43
this boosted technology that we've
00:49:45
discovered across nearly every major
00:49:47
crop worldwide it'll both increase yield
00:49:51
but it will also have a massive impact
00:49:53
on the ability to actually deliver seed
00:49:56
and help farmers and make food prices
00:49:58
lower and improve
00:50:00
sustainability no it's it's actually
00:50:02
cheaper so higher Yi lower cost do you
00:50:04
need more water less water less land
00:50:08
less energy do you need more
00:50:11
fertilizer fertilizer usually scales
00:50:14
with biomass but these sorts of systems
00:50:15
should be more efficient so fertilizer
00:50:17
use per pound produced should go down
00:50:21
significantly as we get to to commercial
00:50:23
trials with all this stuff and we're
00:50:24
doing this across many crops so there's
00:50:26
a lot of work to do in terms of like how
00:50:28
do you scale the field tell us about
00:50:32
the the patents and how important
00:50:35
patents play a role in this because
00:50:37
isn't it like like one of monsanto's big
00:50:39
things like they just go and Sue
00:50:40
everybody into the ground or whatever
00:50:42
like I'm an answer you one second I'm
00:50:43
just goingon to switch my headset it
00:50:44
just died wow we went from sachs's Bots
00:50:47
to free BG's
00:50:49
crops epode I'm glad we're doing him
00:50:51
second cuz all of a sudden like group
00:50:53
chat doesn't seem very important yeah
00:50:56
wow he just
00:50:57
he just solved the whole Ukraine crisis
00:50:59
here we're going to be able to grow
00:51:01
wheat in the desert and in the
00:51:02
rainforest he solved the world food
00:51:05
problem yeah saaks what if you what did
00:51:06
you do for the last six months yeah we
00:51:08
made Enterprise chat a little better but
00:51:10
we added AI to Enterprise chat we
00:51:12
cleaned up your slack so yeah when you
00:51:14
invest we've invested a ton of money
00:51:16
this was stealth for five years we put a
00:51:18
ton of money into this business so when
00:51:20
you invest that um I mean north of 50
00:51:24
North North of 50 yeah 50 million five
00:51:27
years and you don't have a product in
00:51:28
market yet wow that's some we actually
00:51:30
have some product yeah so I haven't
00:51:31
talked about the way we've been making
00:51:32
money in some of the business we've been
00:51:33
doing okay let me just make sure this is
00:51:36
like clear so that last photo you showed
00:51:39
with the different types of
00:51:41
potatoes you had created the super huge
00:51:43
ones but you're saying that the the
00:51:46
yield benefit here is just you create a
00:51:48
much bigger hardier plant that's capable
00:51:49
of producing many potatoes but the size
00:51:52
of the potatoes doesn't change you can
00:51:54
control for that when you breed so the
00:51:56
selection of what plants you put
00:51:57
together in the boosted system allows
00:51:59
you to decide do you want small medium
00:52:00
large that's all part of the the design
00:52:03
of which plants do you want to combine
00:52:05
okay because your goal is not to turn
00:52:06
like a rusted potato into like a
00:52:07
watermelon or something like that no no
00:52:09
the goal is to make more russed potato
00:52:11
per acre so that we use less water we
00:52:13
use less land farmers can make more
00:52:15
money people pay less for food that's
00:52:17
the goal and so it's all about yield
00:52:19
it's not about changing the
00:52:20
characteristics there are some crops
00:52:22
where you want to change the
00:52:23
characteristics like you might want to
00:52:24
make bigger corn kernels and bigger cobs
00:52:27
on the corn which is another thing that
00:52:28
we've done and that's actually been
00:52:30
published in our patent and the Reason
00:52:32
by the way I'm talking about all this is
00:52:34
some of our patents started to get
00:52:35
published last week and so when that
00:52:37
came out the word started to get out and
00:52:39
that's why we decided to get public with
00:52:40
what we've done because it's now coming
00:52:42
out in the open you mentioned something
00:52:44
briefly there about where different
00:52:48
crops can be planted you know we had
00:52:51
these big talks about wheat and corn
00:52:53
they're only available in very specific
00:52:56
s you know north of the equator the camp
00:52:58
jungles camp in obviously polar or
00:53:00
desert extremes so if you're successful
00:53:03
what would this do for on a global basis
00:53:07
where these crops are made because
00:53:10
remember this whole discussion about
00:53:12
Ukraine question totally the Wheat Belly
00:53:14
of uh Europe the Cradle of wheat yeah
00:53:16
it's a great question I'm so glad you
00:53:18
asked it because that's one of the key
00:53:19
drivers for the business is that we can
00:53:22
now make crops adapted to all sorts of
00:53:24
new environments that you otherwise food
00:53:27
today there's close to somewhere between
00:53:29
800 million and a billion people that
00:53:30
are malnourished that means they living
00:53:31
on less than 1200 calories a day for
00:53:34
more than a year but on average we're
00:53:37
producing 3500 calories per person
00:53:39
worldwide in our a systems the problem
00:53:42
is we just can't grow crops where we
00:53:43
need them and so by being able to do
00:53:46
this sort of system where we can take
00:53:48
crops that are very drought resistant or
00:53:50
can grow in sandy soil or very hot
00:53:52
weather and adapt cooler climate crops
00:53:54
to those regions but through the system
00:53:56
we can actually move significantly where
00:53:59
things are grown and um and improve food
00:54:02
access in regions of me how freeberg
00:54:04
when you look at a potato how do you
00:54:05
figure out what part of their DNA is the
00:54:08
drought resistant part yeah and then how
00:54:11
do you make sure that that's turned on
00:54:13
so even if you inherit that chromosome
00:54:14
it's is is there some potential
00:54:16
interaction with the generally if we can
00:54:18
so these are what are called markers
00:54:20
genetic markers and so there are known
00:54:22
markers associated with known phenotypes
00:54:24
a phenotype is a physical trait of a
00:54:26
plant and so we know lots of markers for
00:54:29
every crop that we grow markers for
00:54:31
disease resistance drought resistance
00:54:33
markers for big plants short plants Etc
00:54:37
and so what we do is we look at the
00:54:39
genetics of different plants that we
00:54:40
might want to combine into the boosted
00:54:42
system and we say these ones have these
00:54:43
markers these ones have these markers
00:54:45
let's put them together and then that
00:54:47
that'll drive the results one of the
00:54:49
other interesting things we're seeing
00:54:50
which I didn't get too much into in the
00:54:53
slides it's not just about combining
00:54:56
traits but it turns out when you add
00:54:59
more genes together biology figures out
00:55:02
a way to create Gene networks these are
00:55:05
all these genes that interact with each
00:55:06
other in ways that are not super well
00:55:09
understood but it makes the the organism
00:55:11
healthier and bigger and live longer
00:55:14
this is like when you bring like why
00:55:15
muts are healthier and live longer than
00:55:17
purebred dogs because they have more
00:55:19
genetic diversity so there's a lot of
00:55:21
work now in what's called quantitative
00:55:24
genomics where you actually look at the
00:55:25
statistics acoss all the Gen you use a
00:55:28
model and the model predicts which two
00:55:31
crosses you want to make out of hundreds
00:55:33
of thousands or millions of potential
00:55:34
crosses that the AI predicts here's the
00:55:37
two best ones to to cross because you'll
00:55:39
get this growth or this
00:55:41
healthiness how do you want to make
00:55:43
money freeberg are you going to sell the
00:55:44
seeds are you going to become the direct
00:55:46
farmer are you going to become food as a
00:55:49
service like how do you make the most
00:55:51
money from this we're not going to farm
00:55:53
farmers are our customers and so they're
00:55:56
different ways to partner with people in
00:55:58
the industry who already have seed
00:56:00
businesses or already have genetics and
00:56:02
help them improve the quality of their
00:56:04
business and then there's other
00:56:06
Industries like in potato where we're
00:56:08
building our own business of making
00:56:09
potato seed for example so every crop
00:56:12
and every region is actually quite
00:56:14
different so it becomes a pretty
00:56:15
complicated business to scale we're in
00:56:17
the earlier days we're already Revenue
00:56:19
generating I would like a sweeter
00:56:22
blueberry no comment no comment yeah I
00:56:25
get tilted by the quality of the
00:56:27
driscolls blueberries let me tell you
00:56:29
something about the driscolls
00:56:30
blueberries also the Driscoll I've I've
00:56:32
had only one batch of a Driscoll
00:56:34
strawberry that was just off the charts
00:56:36
and every 19,8 47 other batches I bought
00:56:40
have been total yeah now you want the
00:56:42
European small ones or the Japanese ones
00:56:45
from H CU they're rich and sweet and
00:56:47
they're not these like monstrosity of
00:56:49
giant flavor of strawberries seedless
00:56:53
could you do a seedless
00:56:55
mango yeah you cut it cut it oh my God
00:56:59
how great would that be amount of work
00:57:01
per bite on a mango is like the worst
00:57:03
ratio yeah oh my god well somehow we
00:57:06
made it about us yeah no no look I think
00:57:08
that's it is all about you guys tell us
00:57:10
about the blueberries sorry every year
00:57:13
driscolls puts out a special labeled
00:57:16
package called sweetest batch and they
00:57:19
just had the sweetest batch of
00:57:20
strawberry and blueberri I don't know if
00:57:22
they're still in the stores but they
00:57:23
only last for like a week or two and
00:57:24
that's the best genetics only grown on a
00:57:26
small number of Aces really inred going
00:57:30
as soon as this is done see if they have
00:57:32
it still I got it a few weeks ago it's
00:57:34
quite delicious anyway we know let's
00:57:36
just say we know the berry Market very
00:57:37
well my co-founder CTO Jud Ward who's
00:57:41
whose brilliant idea boosted breeding
00:57:43
was many years ago who I met because
00:57:45
they had a New Yorker article on Jud I
00:57:47
cold called him and said hey will you
00:57:48
come in and give us a tech talk we
00:57:50
started talking and Jud came up with
00:57:52
this idea for boosted breeding and so we
00:57:54
started the business with Jud and Jud
00:57:56
ran molecular breeding at Driscoll so we
00:57:58
have a lot of Driscoll people that work
00:57:59
at ohal we know the market really well
00:58:01
can you go back to the patent stuff like
00:58:02
are
00:58:03
youed person so we spent we spent 50
00:58:06
million bucks on you know plus on this
00:58:09
business to date so we have filed for IP
00:58:12
protection that people can't just rip us
00:58:14
off but I would say I think that the
00:58:16
real Advantage for the business arises
00:58:19
from what we call Trade Secrets which is
00:58:22
not just about taking patents and going
00:58:23
out and suing people that's not a great
00:58:25
business the business is how do you
00:58:27
build a moat and then how do you extend
00:58:28
that moat the great thing about plant
00:58:31
breeding and genetics is that once you
00:58:32
make an amazing variety the next year
00:58:35
the variety gets better and the next
00:58:36
year the variety gets better and so it's
00:58:37
hard for anyone to catch up that's why
00:58:39
seed companies generally get monopolies
00:58:42
in the markets because farmers will keep
00:58:44
buying that seed every year provided it
00:58:46
delivers the best genetics and so our
00:58:49
business model is really predicated on
00:58:50
how do we build advantages and Moes and
00:58:52
then keep extending them rather than try
00:58:54
to leverage IP so I I'm big fan of like
00:58:56
building business model advantages this
00:58:58
is going to be incredible Sachs if you
00:58:59
think about you know geopolitically
00:59:02
what's going on in Somalia Sudan Yemen
00:59:04
Afghanistan those places have tens of
00:59:07
millions of people I think hundreds of
00:59:09
millions collectively who are at risk
00:59:11
for starvation if you could actually
00:59:12
make crops that could be farmed there
00:59:14
freedberg you would change humanity and
00:59:17
then all these people buying up Farmland
00:59:18
in America that could devalue that
00:59:21
Farmland if that wasn't as limited of a
00:59:24
resource you freeberg like no I think um
00:59:26
so first of all like Farmland in America
00:59:28
is mostly familyowned it's about 60%
00:59:31
rented actually so a lot of families own
00:59:33
it and then they rent it out because
00:59:34
they sto farming it but the great thing
00:59:38
that we've seen in agriculture
00:59:40
historically is that the
00:59:41
more calories we produce the more food
00:59:44
we produce the more there seems to be a
00:59:45
market it's like any other economic what
00:59:47
about system wheat and rice yeah so
00:59:50
those are calorie sources one and two
00:59:53
and there's certainly opportunity for us
00:59:55
to apply our boosted systems there the
00:59:58
big breakthrough with potato is we can
00:59:59
make potato seed using our boosted
01:00:01
system in addition to making better
01:00:02
potatoes McDonald's is the largest buyer
01:00:05
of potatoes yeah so in the US 60% of the
01:00:07
potatoes go to french fries and potato
01:00:09
chips McDonald's buys most of the fries
01:00:12
PepsiCo under fro buys most of the
01:00:14
potato chip potatoes 40% are table
01:00:16
potatoes in India 95% of the potatoes
01:00:20
are table potatoes they're eaten at home
01:00:23
and the Indian potato Market's three to
01:00:24
four times as big as the potato Market
01:00:27
in Brazil it's 90% table potato so all
01:00:30
around the world potato different the US
01:00:31
is you know unusually large consumers of
01:00:35
french fries and potato chips I speak on
01:00:37
behalf of J celvin I said we will gladly
01:00:41
invest the million at a 10 cap in both
01:00:43
of your businesses absolutely yes we
01:00:45
will grift our way into this J J Cal and
01:00:48
I will do the deal wire the money we
01:00:50
wire the money a little million to each
01:00:51
of you guys at a 10 cap thank you
01:00:53
absolutely you're it may not be cap
01:00:56
though but yes breing news jam and jcal
01:00:59
have secured the bag so breaking news
01:01:01
jam and jcal have secured the bag from
01:01:04
the bes actually doing work yeah well I
01:01:07
appreciate you guys letting me talk
01:01:08
about it exced to both of you I love it
01:01:11
it's been yeah for yeah building stuff
01:01:14
is hard there's always risk it's a lot
01:01:16
of work and a lot of setbacks but man
01:01:19
when you get stuff working it's great
01:01:21
we're each doing the things we do best
01:01:23
freeberg is solving the world's hunger
01:01:26
problem and I'm making I'm cleaning up
01:01:28
your
01:01:30
slack making your Enterprise chat a
01:01:32
little better all progress
01:01:34
C all right Stanley dren Miller has got
01:01:38
a new boyfriend dren Miller's got a
01:01:41
boyfriend and his name is Javier and
01:01:44
they alop to Argentina dck Miller
01:01:47
professed his love like Tom Cruz on
01:01:49
Oprah couch in a CNBC interview this
01:01:52
week the only free market quote leader
01:01:55
in the world right now bizarrely is in
01:01:57
Argentina of all places he cut Social
01:01:59
Security 35% after he came to office
01:02:02
they've gone from a primary deficit of
01:02:03
like four or five% to a 3% Surplus
01:02:06
they've taken a massive hit in GDP
01:02:08
basically a depression for a quarter and
01:02:10
his approval rating has not gone down
01:02:13
truck and Miller has explained how he
01:02:16
invested in Argentina after seeing
01:02:18
mala's speech at Davos which we covered
01:02:21
here's a 30 second clip play the clip
01:02:23
Nick by the way do you want to hear how
01:02:25
I invest in our artina it's a funny
01:02:26
story I wasn't at Davos but I saw the
01:02:30
speech in Davos and it was about 1:00 in
01:02:32
the afternoon in my office I dialed up
01:02:34
perplexity and I said give me the five
01:02:38
most liquid adrs in Argentina Argentina
01:02:41
it gave me enough of a description that
01:02:43
I follow the old Soros rule invest and
01:02:46
then investigate I bought all of them we
01:02:48
did some work on them I increased my
01:02:50
positions so far it's been great but
01:02:53
we'll see yeah that's quite interesting
01:02:55
he um quick note you hear Dr Miller
01:02:57
mentioned adrs for those of you who
01:03:00
don't know and I was one of them they
01:03:01
stand for American depository receipts
01:03:04
basically a global stock offered on a US
01:03:06
exchange to simplify things for
01:03:10
investors yeah I mean he didn't sign a
01:03:13
prenup here he just went all in and he
01:03:14
bought the stock chamath and then he's
01:03:16
going to figure it out later tell us
01:03:18
your thoughts on this love affair this
01:03:20
Bromance there's a great clip of mle he
01:03:23
goes on this talk show in Argentina and
01:03:26
the talk show host she's just so excited
01:03:29
and greets him and then they start
01:03:31
making out have you guys seen this what
01:03:34
they're just out of control full on
01:03:37
French kissing each other it's
01:03:39
hilarious yeah I mean like Soros has
01:03:42
been very famous for this invest and
01:03:43
investigate thing it's a it's like a
01:03:46
smart strategy for very very liquid
01:03:49
Public Market investors that have the
01:03:52
Curiosity that he does I mean I don't
01:03:53
have much of a reaction to that I think
01:03:55
that
01:03:56
the the thing with Argentina that's
01:03:57
worth taking away is when you've spent
01:04:01
decades casting about and misallocating
01:04:04
capital and running your economy into
01:04:06
the ground the formula for fixing it is
01:04:10
exactly the
01:04:11
same you cut entitlements and you
01:04:15
reinvigorate the economy and so the
01:04:18
thing we need to take away is if we
01:04:19
don't get our together that's probably
01:04:21
what we're going to have to do Sachs the
01:04:23
influence of mle on American
01:04:26
politics Will there be any it seems like
01:04:29
he has paralleled what Elon did at
01:04:32
Twitter Facebook uh and and zucked it at
01:04:35
Facebook do you think that this you know
01:04:38
experiment he's doing down there of just
01:04:40
cutting staff cutting departments will
01:04:42
ever make its way into American
01:04:46
politics probably not I mean not until
01:04:48
we're forced to but what mle did he
01:04:51
comes in and they've got a huge budget
01:04:53
deficit and they've got runaway
01:04:54
inflation and they're deba ing their
01:04:56
currency and just practically overnight
01:04:57
he just slashes government spending to
01:05:00
the point where he has a government
01:05:02
surplus and then as soon as he gets
01:05:03
credibility with the markets that allows
01:05:05
them to reduce interest rates inflation
01:05:07
goes away and people start investing in
01:05:09
the country magic It's Magic there is a
01:05:12
path it's obvious listen I mean you
01:05:15
can't
01:05:16
run deficits forever you can't
01:05:19
accumulate debt forever it's just like a
01:05:21
household if your spending exceeds your
01:05:24
income
01:05:26
eventually you got to pay it back or you
01:05:27
go broke and the only reason we haven't
01:05:29
gotten broke or experienced
01:05:30
hyperinflation is because we're the
01:05:32
world's Reserve currency so there's just
01:05:34
a lot of room for debasement and there's
01:05:37
not a ready alternative yet I mean
01:05:39
everyone's trying to figure out what the
01:05:39
alternative will be so we've been able
01:05:42
to accumulate more and more debt but
01:05:43
it's it's reaching a point where it's
01:05:45
unsustainable and what we've already
01:05:46
seen is that the feds had to jack up
01:05:48
interest rates from very low practically
01:05:51
nothing to 5 a half% and that has a real
01:05:54
cost on people wellbeing because now
01:05:58
your cost of getting a mortgage goes way
01:06:00
up I mean mortgage rates are over what 7
01:06:02
and a half% now yeah six s% depending on
01:06:05
how much net worth and your credit rting
01:06:08
yeah right and so it's much harder to
01:06:10
get a mortgage now it's harder to make a
01:06:12
car payment if you need to borrow to buy
01:06:14
a car and if you have personal debt the
01:06:16
interest rat is going to be higher the
01:06:18
inflation rate actually doesn't take
01:06:19
into account any of those things
01:06:21
remember Larry Summers did that study
01:06:23
where he said the real inflation rate
01:06:24
would be 18 % or would have peaked at
01:06:27
18% if you included a cost of borrowing
01:06:30
that's why people don't feel as well off
01:06:32
as the unemployment rate would normally
01:06:34
suggest so people are hit really hard
01:06:38
when interest rates go up in terms of
01:06:41
big purchases they need to make with
01:06:43
debt and then of course it's really bad
01:06:45
for the investment environment
01:06:47
because when interest rates are really
01:06:49
high that creates a higher hurdle rate
01:06:51
and people don't want to invest in Risk
01:06:55
assets and So eventually the pace of
01:06:57
innovation will go down and Dr Miller
01:06:59
made this point in his next set of
01:07:01
comments he said that treasury is still
01:07:03
acting like we're in a
01:07:04
depression it's interesting because I've
01:07:06
studied the depression you had a private
01:07:08
sector crippled with debt basically with
01:07:10
no new ideas so interventionist policies
01:07:13
were called for and were effective he
01:07:15
said the private sector could not be
01:07:16
more different today than it was in the
01:07:18
Great Depression the balance sheets are
01:07:20
fine they're healthy and have you ever
01:07:21
seen more innovation ideas that the
01:07:23
private sector could take advantage of
01:07:25
like blockchain like AI he says all the
01:07:28
government needs to do is get out of the
01:07:29
way and let them innovate inste theyve
01:07:30
spend and spend and spent and my new
01:07:32
fear now is that spending and the
01:07:35
resulting interest rates on the debt
01:07:37
that's been created are going to crowd
01:07:39
out some of the Innovation that
01:07:41
otherwise would have taken place I
01:07:42
completely endorse duck Miller's view of
01:07:44
bionomics and actually I mean this is
01:07:46
what I said way back in 2021 Victory lap
01:07:50
here we go Little David saaks Victory
01:07:53
lap we need a little graphic for that Dr
01:07:55
used the word biomics and said I give
01:07:57
these guys an F because they're they're
01:07:59
still printing money and spending money
01:08:01
like we're in a depression even though
01:08:02
we're in a riping economy and when they
01:08:04
started doing this back in 2021 you know
01:08:07
I tweeted bionomics equals pumping
01:08:09
trillions of dollars of stimulus into a
01:08:10
roran economy I'm not going to pretend
01:08:11
like I know what's going to happen next
01:08:13
but never tried this before what
01:08:14
happened next was a lot of inflation and
01:08:16
that jacked up interest rates according
01:08:18
to even Keynesian economics the reason
01:08:21
why you have deficit spending is because
01:08:23
you're in a recession or depression and
01:08:24
so use the government
01:08:26
to stimulate and balance things out you
01:08:28
don't you don't do deficit spending when
01:08:30
the econom is already doing well so this
01:08:32
spending there's no reason for it yeah
01:08:34
it's like showing up to like a party
01:08:36
that's going crazy and being like
01:08:38
putting gasoline on the fire yeah I mean
01:08:40
more importantly it should limit the
01:08:43
approval or action of certain
01:08:46
programs that you might otherwise want
01:08:48
to do in a normal environment but in an
01:08:51
inflationary environment you don't have
01:08:54
the flexibility to do them student loan
01:08:56
forgiveness is a really good example is
01:08:58
now the time of course not to do student
01:09:00
loan forgiveness or do we wait for
01:09:02
inflation to temper a bit is now the
01:09:04
time you know so so there's just a lot
01:09:06
of these examples that actually the
01:09:08
opposite should be true yeah but none of
01:09:10
all of those things get you votes before
01:09:13
we move on from this look what we have
01:09:14
coming out of Washington here is a
01:09:15
contradictory and therefore
01:09:17
self-defeating policy you've got the FED
01:09:19
jacking up rates to control inflation
01:09:21
you move across town and you've got
01:09:23
Capitol Hill in the White House spending
01:09:25
like there's no tomorrow which is
01:09:26
inflationary why would you do both those
01:09:28
things choose what your policy is going
01:09:30
to be it's like driving with your foot
01:09:32
on the break and the gas at the same
01:09:33
time it's not a great idea for the car
01:09:35
let me just make one comment J Cal
01:09:36
before we move on about the dren Miller
01:09:38
investment statement of course and I I
01:09:40
just wanted to say like I think what it
01:09:42
highlights about dren Miller and call it
01:09:45
a rift in investing philosophy or skill
01:09:47
is the difference between precision and
01:09:50
accuracy what I mean by that is
01:09:52
precision really references that you do
01:09:54
a lot of detail analysis to try and make
01:09:56
sure you understand every specific thing
01:09:59
that is going right or could go wrong
01:10:02
but the problem and so that means you
01:10:04
for example might do a ton of diligence
01:10:05
on a company and make sure you
01:10:07
understand every dollar every point of
01:10:08
margin all the specifics of the
01:10:11
maturation of that business and where
01:10:12
they are in their cycle but you could be
01:10:15
very precise but be very inaccurate for
01:10:18
example if you miss an entire Trend
01:10:20
someone could invest in Macy's back when
01:10:23
Amazon was taking off and have done a
01:10:25
lot of precise analysis on Macy's margin
01:10:28
structure and performance and said this
01:10:29
is a great business but they missed the
01:10:31
bigger Trend which is that e-commerce
01:10:34
was going to sweep away Macy's and
01:10:35
consumers were simply that's not
01:10:37
possible in the analysis that they were
01:10:39
doing let's be honest freeberg nobody
01:10:41
could make that stupid of a trade to say
01:10:43
Macy's versus Amazon over the next 10
01:10:46
years well yeah and so like and um J you
01:10:49
want to show that who no no no no do not
01:10:51
poke the tiger let's not get into with
01:10:54
other podcasters the worst spread trade
01:10:56
in history yeah let me just finish the
01:10:57
statement yeah but the other one is
01:11:00
being accurate and accurate means you
01:11:01
get the yeah the right bet the right
01:11:04
sentiment yes the right Trend the
01:11:06
problem with being accurate you could
01:11:07
have said in the year 2000 hey the
01:11:10
internet's going to take off and you
01:11:12
could have put a bunch of money in but
01:11:14
the problem was you were right you just
01:11:16
had to have the necessary patience and
01:11:19
so accuracy generally yields better
01:11:22
returns but it requires more patience
01:11:25
because you can't necessarily time how
01:11:27
long it will take for you to be right so
01:11:29
a guy like drunen Miller is making an
01:11:31
accurate bet he bets correctly on the
01:11:34
trend on where things are headed he
01:11:36
doesn't necessarily need to be precise
01:11:38
but he has the capital and his capital
01:11:40
structure that allows him to be patient
01:11:42
to make sure that he eventually gets the
01:11:43
return and to build on your thoughts
01:11:45
having watched this movie a couple of
01:11:46
times and you know I overthought the
01:11:48
Twitter investment as but one example I
01:11:51
had the opportunity to invest in Twitter
01:11:52
when it was like a singled digit
01:11:54
Millions company I just thought you know
01:11:56
what this thing is only like the
01:11:59
headline and I told like it's the
01:12:01
headline it's not like the entire blog
01:12:03
post it's be a cacophony of idiots this
01:12:05
thing is going to be chaos and I was
01:12:06
right but I was wrong right great bet
01:12:09
but my wrong analysis right and so you
01:12:11
can add Precision to other aspects like
01:12:14
when you sell your shares or when you
01:12:16
double down but you have to get the
01:12:18
trend right which is Evan Williams great
01:12:19
entrepreneur Jack great entrepreneur
01:12:21
Twitter taking off like a weed just make
01:12:23
the BET right you knew too much about
01:12:27
journalism you knew too much about the
01:12:29
space they were trying to disrupt and
01:12:30
that can be a mistake correct we did
01:12:32
yeah PayPal none of us knew anything
01:12:34
about payments that was one of the
01:12:35
reasons we were successful all the
01:12:36
payments experts told us it couldn't be
01:12:38
done right absolutely so that happens a
01:12:40
lot I had never even know I didn't even
01:12:42
know what a Facebook was when I joined
01:12:44
Facebook it's an American college
01:12:45
phenomenon no serious you don't have
01:12:47
that in Canada but you knew Zuck and you
01:12:49
you saw some growth charts and you saw
01:12:51
some Precision in his ability to build
01:12:53
product and
01:12:55
I mean the great thing about the great
01:12:56
thing about Network effect businesses is
01:12:58
there's a trend line that sustains
01:13:00
because it builds if if it's an
01:13:01
appropriate Network effect so you can be
01:13:04
accurate about buying into the right
01:13:06
Network effect business you don't need
01:13:08
to use all of this diligence to be
01:13:10
perfectly sound around the maturation of
01:13:14
the revenue and the margin structure and
01:13:16
all that stuff as long as the trend line
01:13:17
is right and you're willing to be
01:13:18
patient to hold your investment I think
01:13:20
dren Miller's point is incredible he
01:13:22
took a look he very quickly made a macro
01:13:24
assessment from a macro perspective what
01:13:26
mle is doing is significantly different
01:13:29
than what we're seeing in any other
01:13:30
Emerging Market let alone mature Market
01:13:32
with respect to fiscal austerity and
01:13:34
appropriateness in this sort of
01:13:36
inflationary Global inflationary
01:13:37
environment and he said you know what I
01:13:39
don't see any other leader doing this
01:13:41
this is a no-brainer bet let me make the
01:13:43
BET and as long as he's willing to hold
01:13:45
this thing for long enough eventually
01:13:47
the markets will get there and call it a
01:13:49
spread trade against anything he'll be
01:13:51
proven right well and so not to but
01:13:54
speaking of bets J you told me this week
01:13:57
that you just made your largest
01:13:58
investment ever tell us about that yeah
01:14:00
so I've gotten very lucky now because a
01:14:03
lot of my Founders from the first couple
01:14:04
of
01:14:05
cohorts of investing I did when I was a
01:14:08
sequoia Scout have come back and created
01:14:10
second and third companies and so you
01:14:12
know that happened with TK Uber and then
01:14:14
Cloud kitchens it happened with Raul
01:14:17
from report of then superum and then it
01:14:20
happened recently just in the past year
01:14:22
my friend Jonathan who's the co-founder
01:14:24
of thumbtack asked me to come to dinner
01:14:26
he said hey you know you were the first
01:14:27
investor in Thumbtack will you be the
01:14:29
first investor in our next company
01:14:30
Athena and I said sure what do you do
01:14:32
and he explained it to me and and we put
01:14:33
a seven figureure b in which is rare for
01:14:35
us as a seed fund right normally our bet
01:14:37
sizes are 100K 250 you know it's a $50
01:14:40
million fund why did you do it yeah it's
01:14:43
very simple it's the fastest growing
01:14:44
company I've ever seen and I'm including
01:14:47
Uber in that it has been growing at uh
01:14:50
you
01:14:51
know a rate that I'll just say is faster
01:14:54
than Uber and Robin Hood when when we
01:14:56
were investing in them tens of millions
01:14:57
of dollars it's a very simple
01:14:59
concept when
01:15:00
Thumbtack was building their market
01:15:03
place they used researchers in places
01:15:07
like Manila Etc in the Philippines
01:15:09
knowledge workers and what they realized
01:15:10
was the 0.1% of those knowledge workers
01:15:14
were as good or better than say
01:15:16
Americans at doing certain jobs and so
01:15:18
they've created this virtual EA service
01:15:21
you can go see it at Athena wow.com and
01:15:25
we now have two of them inside of our
01:15:27
company it turns out Americans don't
01:15:28
want to do the operations rle so it's
01:15:31
kind of like AWS you just give them
01:15:32
$336,000 a year they give you
01:15:34
essentially an operations or an EA and
01:15:37
they have ones that are kind of chief of
01:15:39
staish and this company is growing like
01:15:43
a weed so I am working with them on the
01:15:45
product design as well so imagine having
01:15:48
you know two or three of these
01:15:51
incredibly hardworking people who are
01:15:53
trained with MBA class level curriculum
01:15:58
they spend months training these people
01:16:00
up they pay them two or three times what
01:16:03
they would make at any other company and
01:16:04
then they pair them with Executives here
01:16:06
and it's kind of been an underground
01:16:08
Secret in Silicon Valley because it's
01:16:11
only By Invitation right now because
01:16:12
they can only train so many people but
01:16:14
if you've tried to hire an executive
01:16:16
assistant I don't know if anybody's
01:16:17
tried to do that recently you hooked me
01:16:19
up so I will be Guinea picking this
01:16:21
service yes soon and I have two of them
01:16:24
and so it is just the greatest that you
01:16:27
can have an operations are these people
01:16:29
powered by AI tools as well yeah so
01:16:31
that's the kind of Secret Sauce here is
01:16:33
they're training them and they watch you
01:16:35
work and then they will learn how you do
01:16:38
your job and then how quickly you can
01:16:40
delegate and get stuff off your plate is
01:16:42
the name of the game so we have an
01:16:44
investment team with researchers and
01:16:45
analysts in it we have a due diligence
01:16:47
team and then you have like executive
01:16:49
functions in our fund they have now
01:16:51
started shadowing you know you know
01:16:54
highly paid Americans in an investment
01:16:57
firm ours and then train them up and now
01:17:00
our due diligence our first level
01:17:02
screening you know and our tracking of
01:17:04
companies is being done by these
01:17:05
assistants for what I'll say is a third
01:17:08
to a fourth of the price I was paying
01:17:10
previously so what that does in an
01:17:12
organization is we're just delegating
01:17:14
away and then moving our investment team
01:17:16
to doing in-person meetings and doing
01:17:19
higher level stuff and so yeah you're
01:17:21
you're 8090 the so at 8090 we have this
01:17:24
funny thing where we've made it a verb
01:17:25
whenever you see somebody doing high
01:17:28
quality work at a quarter to a tenth of
01:17:30
the cost we say oh you just 89 it
01:17:33
correct so you're you're 8090 in the
01:17:35
investment team I'm 8090 in the
01:17:36
investment team and you know what it was
01:17:38
scary as hell for them because they're
01:17:39
like am I GNA lose my job it's like no
01:17:41
you now get to instead of doing a check
01:17:42
and call once a month you can do a check
01:17:45
and call every other week or every week
01:17:46
or instead of doing 15 first round
01:17:49
interviews a week you can do 25 because
01:17:51
you have this assistant with you way
01:17:53
doing all the repetitive work the way
01:17:55
that companies will work in five and 10
01:17:57
years I don't think guys any of us are
01:17:59
going to recognize what it's going to
01:18:00
look like yeah so I go I mean like
01:18:03
watching Sax's demo earlier how much
01:18:06
progress and how seamless that product
01:18:09
works with the features it has enabled
01:18:11
by the underlying models you just get to
01:18:14
thinking how all of these vertical
01:18:17
software applications become completely
01:18:21
personalized and quickly rebuilt around
01:18:23
AI you know
01:18:25
it's so obvious can you imagine how long
01:18:27
it would have taken John to write a
01:18:28
letter to Lena KH to invite like if we
01:18:30
said John invite Lena KH But be sure to
01:18:33
reference all the nice things we said
01:18:34
about her on episodes of the Pod oh it
01:18:36
be 10 hours of work got find find all
01:18:39
those you got to listen to them and
01:18:40
figure out what the best quotes are you
01:18:42
got and you got it done in five seconds
01:18:43
it's incredible totally and then imagine
01:18:45
building that same sort of capability
01:18:47
into a very specific vertical
01:18:49
application that's specific to some
01:18:51
business function and you can probably
01:18:54
spend a couple or an hour building that
01:18:55
function and then it saves you hours a
01:18:58
day in peret you know I think I think
01:19:01
that's why these tools companies or the
01:19:03
tools products that
01:19:06
Google Microsoft Amazon and a few others
01:19:09
are building are actually incredible
01:19:11
businesses because so many Enterprises
01:19:13
and so many vertical application
01:19:15
Builders are going to be able to
01:19:16
leverage them to rewrite their entire
01:19:18
business functions I got myself and my
01:19:20
co-founders at 8090 we get this stream
01:19:23
of emails of companies that are like
01:19:24
like or people that are like we have
01:19:27
this product idea or we have this small
01:19:29
product one of the emails I got this was
01:19:32
crazy was from a guy that's like oh
01:19:34
we've 8090 Photoshop so like we have
01:19:37
like a much much cheaper version of
01:19:38
Photoshop and the guy was doing like a
01:19:39
few million bucks of of AR and growing
01:19:42
really nicely but then it turned out
01:19:44
that somebody saw that and then 809 it
01:19:47
so then there was version of that thing
01:19:51
and so to your point freeberg none of
01:19:53
these big companies a chance yeah yeah
01:19:56
it's everything cheap faster not because
01:19:59
they're not because the products aren't
01:20:01
good but like J's going to go off and
01:20:03
experiment with this sack's going to go
01:20:04
off and build a product you know as
01:20:06
every time that you're at a boundary
01:20:08
condition we're all going to explore
01:20:10
well maybe we could do this with AI
01:20:12
maybe we shouldn't hire a person not
01:20:14
because we're trying to be mean about it
01:20:16
but it's because the normal natural
01:20:18
thing to do and the Opex of companies is
01:20:21
just going to go down which means the
01:20:22
size of companies are going to shrink
01:20:24
which amount of M is going to and that's
01:20:28
just going to create the ability for
01:20:29
these compan to sell those products
01:20:30
cheaper so M it's a massive deflationary
01:20:34
tail we had the same thing happen with
01:20:36
compute and now it's happening inside of
01:20:39
organizations I I wrote a blog post
01:20:41
about this on my substack called add
01:20:42
this is the framework I came up with I
01:20:45
told my entire team look at what you got
01:20:47
done every week and I want you to ask
01:20:48
three questions how can I automate this
01:20:50
how can I deprecate this how can I
01:20:53
delegate it and you know the automate
01:20:55
part is AI and what you're doing David
01:20:57
the delegate part is atheno wow.com and
01:21:00
then the deprecate is hey just be
01:21:02
thoughtful what are you doing that you
01:21:03
don't need to do and that's 8090 and
01:21:06
something like there are things inside
01:21:07
these products that you don't actually
01:21:09
need what's the core functionality of
01:21:11
the product you know make it as
01:21:13
affordable as possible and then what's
01:21:15
going to happen for people who think
01:21:16
this is bad for society you've got to
01:21:18
completely wrong we're g to have more
01:21:20
people be able to create more products
01:21:23
and solve more problems the unemployment
01:21:25
rate is going to stay very low we're
01:21:27
just going to have more companies so the
01:21:29
idea like there was somebody who was
01:21:31
working on very small like software I
01:21:34
want to I get pitched on very Niche
01:21:35
ideas I want to create something where
01:21:37
people can find people to play pickle
01:21:39
ball with right like a pickle ball
01:21:40
marketplace now that didn't wouldn't
01:21:43
typically work because you would need $5
01:21:45
million a year to build that product but
01:21:47
if you can build it for $500,000 a year
01:21:49
well now you've only got to clear that
01:21:50
number to be profitable so a lot more
01:21:52
smaller businesses a lot more
01:21:55
all these little niche ideas will be a
01:21:57
to be built and a VC who says I'm not
01:21:59
giving you $5 million to build that app
01:22:00
will be like but I will give you 500k
01:22:03
and that's what I'm seeing on the ground
01:22:05
in startups the same startups that had a
01:22:07
request of $3 million in funding five
01:22:09
years ago are now requesting 500 to a
01:22:13
million it's deflationary all the way
01:22:15
down did you incredible did you see the
01:22:18
Google thing did you guys see the Google
01:22:20
AI Gemini stuff chat GPT Omni launched
01:22:23
at the same time or perhaps
01:22:25
strategically right before Google
01:22:27
dropped its latest AI announcements at
01:22:30
IO the biggest announcement is that they
01:22:34
are going to change search this is the
01:22:36
piece of the puzzle in the Kingdom that
01:22:38
they have been very concerned with and
01:22:40
they're going for it the new product and
01:22:42
they have like 20 different products you
01:22:45
can see them at labs. gooogle where they
01:22:46
put all their different products but
01:22:48
this is the most important one they call
01:22:49
it AI overviews basically it's
01:22:51
perplexity for most users by the end of
01:22:54
the year they're going to have this
01:22:55
here's how it works and you can see it
01:22:56
on your screen if you're watching us go
01:22:58
to YouTube here they gave an example how
01:23:00
do you clean a fabric sofa this normally
01:23:02
would have given you 10 Blue Links here
01:23:04
it gives you step-by-step guide with
01:23:06
citations and links so they're
01:23:08
preempting you know the issue of people
01:23:10
getting upset and as I predicted they're
01:23:14
going to have targeted ads here's the
01:23:16
things you need in order to clean your
01:23:19
couch you can only use this if you're
01:23:20
using your Gmail account if you use like
01:23:22
a domain name on Google it won't work
01:23:25
there so go to labs. gooogle but they're
01:23:28
doing citations and I think that we're
01:23:31
going to see a major lawsuit here those
01:23:33
people who are in those boxes are going
01:23:34
to look at the answer here and realize
01:23:36
maybe they don't get the click through
01:23:37
and that this answer was built on that
01:23:39
and now we're going to have to have a
01:23:40
new framework there's going to need to
01:23:41
be sax a new company that clears this
01:23:45
content so that Google can do answers
01:23:48
like this the workflow stuff in Gmail
01:23:50
also kicked ass the demo that they
01:23:52
showed was you get a bunch of receipts
01:23:55
and the person giving the demo she said
01:23:57
something the effect of well wouldn't it
01:23:59
be great if like you know the AI
01:24:01
assistant were able to find all the
01:24:02
receipts and then aggregated them and it
01:24:05
put them in a folder and then also
01:24:07
actually generated an expense report or
01:24:09
like a a spreadsheet on the fly it it's
01:24:13
crazy yeah I gotta say I think that it's
01:24:15
free to change your mind and so it's
01:24:18
good to do that oh and I think that
01:24:20
jamath in a rare moment of reflection
01:24:24
might do a are we gonna have a re
01:24:25
underwriting is this a re underwriting I
01:24:27
change my mind all the time I just I
01:24:29
mean you know because I'm gentlemen
01:24:32
breaking news is rewriting his Google
01:24:35
trick sorry I don't to blow your ears
01:24:38
out I think the Google thing is pretty
01:24:40
special between last week's announcement
01:24:43
of isomorphic labs which let's be honest
01:24:45
that's a that's just a multi hundred
01:24:49
billion dollar company so you're saying
01:24:51
there might be many think about it this
01:24:53
way right m opportun sitting there
01:24:55
dormant inside of Google that AI unlocks
01:24:57
look at a company like royalty Pharma so
01:24:59
if royalty Pharma with a pretty it's a
01:25:02
phenomenal business run by a phenomenal
01:25:04
entrepreneur Pablo laretta but what is
01:25:06
that business that's buying two and
01:25:07
three% royalties of drugs that work and
01:25:11
you can see how much value that those
01:25:13
guys have created which is essentially
01:25:15
90% eitaa margin business it's
01:25:18
outrageous because they're in the
01:25:19
business of analyzing and then buying
01:25:21
small
01:25:22
slivers I think something like is
01:25:24
morphic ends up being of that magnitude
01:25:26
of margin scale but at an order of
01:25:28
magnitude or two orders of magnitude
01:25:30
higher Revenue so if you if you fold
01:25:32
that back into a
01:25:34
Google if you think about what they're
01:25:36
doing now on the search side these guys
01:25:38
may be really kicking some ass here so I
01:25:42
I I think that the the reports of their
01:25:44
death were premature and exaggerated
01:25:46
absolutely and the report of their death
01:25:48
freeberg was based upon people don't
01:25:50
need to click on the ads but as I said
01:25:52
on this very Bas my belief is that this
01:25:55
is going to result in more searches and
01:25:58
more knowledge engagement because once
01:26:00
you get how to cook your steak and get
01:26:03
the the right uh temperature right for
01:26:04
medium rare it's going to anticipate
01:26:07
your next three questions better so now
01:26:09
it's going to say hey what wine pairing
01:26:10
would you want with that steak hey do
01:26:12
you need steak knives and it's just
01:26:13
going to read your mind that you need
01:26:14
steak knives and chamath likes to buy
01:26:16
steak knives but maybe you like to buy
01:26:18
mock meets whatever it is it's going to
01:26:20
drive more research and more clicks so
01:26:23
while the monetization per search may go
01:26:25
down we might see many many more
01:26:27
searches what do you think freeberg you
01:26:29
work there and um when we look at the
01:26:32
the the future of the company and the
01:26:34
stock price Nick will pull it up man if
01:26:37
you had held your stock yeah I don't
01:26:39
know did you hold I think I bought some
01:26:43
um original stock during the stu oh no I
01:26:46
sold all my stock back when I started
01:26:48
climate because I was a startup
01:26:51
entrepreneur and needed to
01:26:53
live so which you know I I Rec I did the
01:26:57
math on it it was pretty it'd be worth
01:27:00
it' be worth a lot it would be worth
01:27:02
billions or tens of billions no no would
01:27:05
it have been a billion or okay no no
01:27:07
okay you know I was not like a I was not
01:27:09
a senior exact or anything I I think
01:27:12
what you said is is probably true so
01:27:15
that's a creative I think the other
01:27:17
thing that's probably true is a big
01:27:19
measure at Google on the search page in
01:27:22
terms of search engine perform
01:27:24
performance was the bounceback rate
01:27:25
meaning someone does a search they go
01:27:27
off to another site and then they come
01:27:29
back because they didn't get the answer
01:27:30
they wanted and then the one box
01:27:32
launched which shows a short answer on
01:27:34
the top which basically keeps people
01:27:37
from having a bad search experience
01:27:38
because they get the result right away
01:27:40
so a key metric is they're going to
01:27:42
start to discover which vertical
01:27:44
searches uh meaning like hey cooking
01:27:48
recipes that kind of stuff like you're
01:27:49
referencing travel there's lots and lots
01:27:51
of these different types of searches
01:27:53
that will a snippet or a one box that's
01:27:55
powered by Gemini that will provide the
01:27:58
user a better experience than them
01:28:00
jumping off to a third party page to get
01:28:02
that same content and then they'll be
01:28:04
able to monetize that content that they
01:28:06
otherwise were not participating in the
01:28:08
monetization of so I think the real
01:28:10
victim in all this is that long tale of
01:28:13
content on the internet yeah that
01:28:14
probably gets cannibalized by the
01:28:16
snippet one box experience within the
01:28:18
search function and then I do think that
01:28:20
the revenue per search query in some of
01:28:23
those categories actually has the
01:28:24
potential to go up not down explain
01:28:26
explain give me an example you keep
01:28:28
people on the page so you get more more
01:28:30
search
01:28:31
volume there you get more searches
01:28:33
because of the examples you gave and
01:28:35
then when people do stay you now have
01:28:37
the ability to better monetize that
01:28:39
particular search query because you
01:28:41
otherwise would have lost it to the
01:28:42
third party content page so for example
01:28:45
selling the steak knives is another you
01:28:46
know it's kind of a good example or
01:28:48
booking the travel directly and so on so
01:28:51
by keeping more of the experience
01:28:53
integrated they can monetize the search
01:28:55
per query higher and they're going to
01:28:59
have more queries and then they're going
01:29:01
to have the quality of the queries go up
01:29:02
so I think it's all in there's a case to
01:29:05
be made I haven't done a spreadsheet
01:29:07
analysis on this but I guarantee you
01:29:09
going back to our earlier point about
01:29:10
precision versus accuracy my guess is
01:29:13
there's a lot of hedge fund type folks
01:29:14
doing a lot of this Precision type
01:29:16
analysis trying to break apart search
01:29:18
queries by vertical and try to figure
01:29:21
out what the net effect will be of
01:29:22
having better AI driven box and Snippets
01:29:25
and my guess is that's why there's a lot
01:29:26
of buying activity happening in the
01:29:28
stock right now and I think they're
01:29:29
probably all missing to's point a lot of
01:29:32
these call options like isomorphic Labs
01:29:36
I can tell you meta and Amazon meta and
01:29:38
Amazon do not have an isomorphic lab and
01:29:40
wh sitting inside their business that
01:29:42
suddenly Pops to a couple hundred
01:29:44
billion of market cap and Google does
01:29:46
have a few of those so so other bats
01:29:48
could actually pay off these there there
01:29:50
may be look I mean there's Calico no one
01:29:51
talks about calico I don't know what's
01:29:52
going on over there exension yeah let me
01:29:54
get saaks involved in this discussion
01:29:55
saxs when we show that example it's
01:29:59
obvious Google is telling you where they
01:30:01
got these citations from and how they
01:30:02
built they how to clean your couch how
01:30:04
to make your steak those they were in a
01:30:06
very delicate balance with content
01:30:08
creators over the past two decades which
01:30:10
is hey we're going to use a little bit
01:30:12
of your content but we're going to send
01:30:14
you traffic this is going to take away
01:30:17
the need to send traffic to these places
01:30:19
they're going to benefit from it to me
01:30:21
this is the mother of all class action
01:30:23
lawsuits because is they're putting it
01:30:24
right up there hey we're using your
01:30:27
content to make this answer here's the
01:30:29
citations we didn't get your permission
01:30:30
to do this but we're doing it anyway
01:30:32
what do you think is the resolution here
01:30:34
does all these content go away because
01:30:37
there's no model does Google try to make
01:30:39
peace with the content creators and cut
01:30:40
them in or license their data what's
01:30:42
going to happen to content creation when
01:30:45
somebody like Google is just going to
01:30:47
take wire cutter or these other sources
01:30:49
that are not behind a pay wall and just
01:30:51
give you the goddamn
01:30:52
answer well look this is the same
01:30:54
conversation we've had two or three
01:30:55
times where we're going to need the
01:30:57
courts to figure out what fair use is
01:30:59
and depending on what they come up with
01:31:00
it may be the case that Google has to
01:31:03
cut them in by licensing by doing
01:31:05
licensing deals we don't know the answer
01:31:06
to that yet uh by the way I do know a
01:31:08
Founder who is already skating to where
01:31:11
the puck is going and creating a rights
01:31:13
Marketplace so that content owners can
01:31:16
license their AI rights to whoever wants
01:31:18
to use them I think that could be SM I
01:31:20
was I had a call with him yesterday and
01:31:22
you and I will be on that table together
01:31:24
once again and I yeah yeah so I don't
01:31:26
want to say who it is because going to
01:31:27
let him announce his own round but um
01:31:29
I'm going be participating in the seed
01:31:30
round look stepping back here it's
01:31:33
interesting if you go back to the very
01:31:34
beginning of Google the OG Google search
01:31:37
bar had two buttons on it right search
01:31:39
and I feel lucky I feel lucky was just
01:31:42
tell me the answer just take me to the
01:31:43
best result and no one ever did that
01:31:46
because it kind of sucked then they
01:31:48
started inching towards with one box but
01:31:50
it wasn't you didn't get the one box
01:31:51
very often it's very clear now that
01:31:54
powered one box is the future of Google
01:31:56
search people just want the answer I
01:31:59
think that this feature is going to eat
01:32:01
the rest of Google Search now it's a
01:32:05
little bit unclear what the financial
01:32:07
impact of that will be I think like you
01:32:09
guys were saying there'll be probably be
01:32:10
more searches because search gets more
01:32:12
useful there's fewer Blue Links to click
01:32:14
on but maybe they'll get you know
01:32:16
compensated through those like relevant
01:32:18
ads hard to say you're probably right
01:32:20
that Google ultimately benefits here but
01:32:24
let's not pretend this was a deliberate
01:32:26
strategy on their point they got drag
01:32:28
Kicking and Screaming into this by
01:32:29
innovation of perplexity and other
01:32:31
companies yep they had no idea they got
01:32:34
caught completely flat-footed and
01:32:36
they've now I guess caught up by copying
01:32:39
perplexity and sucks for perplexity I
01:32:41
think they're kind of screwed now unless
01:32:42
they get it's over an acquisition deal
01:32:45
but perplexity came up with the idea of
01:32:48
having
01:32:50
citations in your having a comprehensive
01:32:52
search result yeah which was something
01:32:55
search result with citations and related
01:32:57
questions and right they did it
01:32:59
extremely well and quite frankly all
01:33:00
Google had to do was copy them now
01:33:02
they've done that and I think it does
01:33:04
look like a killer and by the way this
01:33:06
was all something that I saw 15 years
01:33:08
ago when I did Mahalo which was my human
01:33:10
power search engine and which I had
01:33:12
copied or been inspired by neor and D in
01:33:15
Korea they were the first ones to do
01:33:16
this you know it shamama because there
01:33:18
were only three or four markets where
01:33:20
Google couldn't displace the number one
01:33:22
Korea Russia
01:33:24
Japan Russia had um what was the Russian
01:33:27
search engine God I can't remember now
01:33:30
Japan had Yahoo Japan uh which masi
01:33:32
yoshian had carved out it was never part
01:33:34
of it and they were loyal to that and
01:33:36
very nationalistic Koreans and very
01:33:38
Innovative Folks at D and neor just made
01:33:42
search that was so amazing you do a
01:33:45
search and be like here's music Here's
01:33:47
images here's answers here's Q&A it was
01:33:50
awesome but you know it just shows you
01:33:52
like you need to have a lot a wherewith
01:33:54
all and timing is everything as an
01:33:55
entrepreneur my timing was 10 years too
01:33:57
early and the wrong technology I used
01:33:58
humans not AI because AI didn't work 15
01:34:00
years ago one thing I would say about
01:34:03
big companies like a Google or Microsoft
01:34:06
is that the power of your Monopoly
01:34:08
determines how many mistakes you get to
01:34:10
make so think about Microsoft completely
01:34:14
missed iPhone remember and they like
01:34:16
they screwed up the whole smartphone
01:34:17
mobile phone era and it didn't matter
01:34:19
didn't matter SAA comes in blows this
01:34:22
thing up to A3 trillion doll public
01:34:23
public company same thing here with
01:34:25
Google they completely screwed up AI
01:34:27
they invented the Transformer completely
01:34:29
missed llms then they had that Fiasco
01:34:32
where you know they have black George
01:34:33
Washington black George Washington
01:34:35
doesn't matter they can make 10 mistakes
01:34:37
but their Monopoly is so strong that
01:34:39
they can finally get it right by copying
01:34:41
the innovator and they're probably going
01:34:43
to become A5 billion doll company now
01:34:45
sorry five trillion dollar company it
01:34:46
reminds me you know the greatest product
01:34:48
creation company in history I think we
01:34:52
all know who that was and take a a look
01:34:54
down memory lane here are the 20 biggest
01:34:56
felt Apple products of all time the
01:34:59
appal Lisa Macintosh portable we all
01:35:02
remember the Newton which was their PDA
01:35:05
the 20th anniversary Macintosh super
01:35:07
sexy people don't remember they had
01:35:09
their own video game I I was at a
01:35:11
conference a couple years ago that Jeff
01:35:15
spoke at I think he's given this talk in
01:35:18
a couple other places you could probably
01:35:19
find it on the internet yeah but he
01:35:20
talks about Amazon's Legacy of failure
01:35:24
and how they had the fire phone and the
01:35:25
fire this and the fire that and he's
01:35:28
like our job is to fail swings we have
01:35:30
to make these blunders but what makes us
01:35:32
successful is that we learn from the
01:35:34
failures and you know we make the right
01:35:36
next decision yeah yeah but I say if
01:35:39
you're a startup and you make big
01:35:40
failures you usually just go out of
01:35:41
business one and done but this is how
01:35:43
you this is how you but this is how you
01:35:45
stay competitive if you're a big founder
01:35:48
Le tech company the only way you're
01:35:50
going to have a shot at staying relevant
01:35:53
is to take big shots that you're going
01:35:55
to fail at I
01:35:57
just you have you have to do things
01:35:59
you're to fail right remember this boom
01:36:01
box is one of the huge difference
01:36:02
between startups and big companies is
01:36:04
that big companies can afford to have a
01:36:06
portfolio of products they have a
01:36:07
portfolio of bets some of them will work
01:36:09
and that keeps the company going startup
01:36:11
really has to go all in on their best
01:36:12
idea totally I always tell Founders just
01:36:14
go all in on your best idea they're
01:36:16
always ask me for permission to Pivot
01:36:19
and I always tell them do go for the
01:36:21
best idea don't don't hedge don't try do
01:36:23
five things at once just go all in on
01:36:25
your best idea yeah yeah and if it
01:36:27
doesn't work out you reboot and start
01:36:29
with a new c table you're going to go
01:36:31
all in so to speak another amazing
01:36:34
episode is in the Ken the boys are in a
01:36:36
good mood you got your great episode no
01:36:38
guests this week just all bestie all the
01:36:42
time and very important the march to a
01:36:46
million continues halfway there you got
01:36:48
us there fans we hit 500,000 subbies on
01:36:52
YouTube which means y all earned a live
01:36:54
Q&A with your besties coming at you in
01:36:57
the next couple of weeks we're going to
01:36:58
do it live on YouTube so if you're not
01:37:01
one of the first 500 get in there now so
01:37:03
you get the alert we're going to take
01:37:04
your questions live it's going to be
01:37:06
dangerous any questions no questions uh
01:37:08
are dangerous who knows what could
01:37:10
happen on a live show and by the way I
01:37:12
just want to let you know that Phil
01:37:14
helmouth breaking news Phil helmouth and
01:37:16
Draymond Green just resigned from open
01:37:18
AI we didn't get into that but the open
01:37:19
a resignations continue Phil H has
01:37:22
tweeted he is no longer with open
01:37:26
AI you guys like my baby cashmir pink
01:37:29
sweater it's pretty great we gonna get
01:37:31
Summer chamat soon are the buttons
01:37:32
coming down are you gonna go linen the
01:37:35
when does linen chamat show up the
01:37:36
unbuttoning is about to happen in the
01:37:38
next two or three weeks great
01:37:40
unbuttoning this is how you know it's
01:37:41
kind of like Groundhog Day you know that
01:37:43
Summer's here when we lat's buttons
01:37:46
almost it's Memorial Day when after
01:37:49
Memorial Day the button can come down
01:37:51
yeah we're going to go three buttons
01:37:52
down be wearing my black tea Sachs will
01:37:56
still be blue blaze or blue shirt red
01:37:58
tie and freeberg in Fields of Gold look
01:38:02
at freeberg in Fields of Gold taking us
01:38:04
out Sting Fields of Gold coming at you
01:38:06
two for Tuesday see you all in the next
01:38:08
all in pod for the Sultan of
01:38:11
science raain Man David saxs and
01:38:13
chairman dictator I am your Z100 Morning
01:38:16
Zoo DJ we'll see you next time love you
01:38:19
boys byebye
01:38:22
byebye will let your winners
01:38:25
ride Rainman
01:38:29
David and in said we open source it to
01:38:32
the fans and they've just gone crazy
01:38:34
with it love queen
01:38:36
[Music]
01:38:42
of
01:38:45
Besties my dog taking your
01:38:48
driveway man oh man my habiter will meet
01:38:53
me
01:38:53
we should all just get a room and just
01:38:55
have one big huge orgy cuz they're all
01:38:57
this useless it's like this like sexual
01:38:58
tension that they just need to release
01:39:00
[Music]
01:39:06
somehow we need to get
01:39:11
[Music]
01:39:15
mer I'm
01:39:18
going and now the plugs the all-in
01:39:21
Summit is taking place in Los Angeles on
01:39:23
September 8th through the 10th you can
01:39:26
apply for a ticket at summit. Allin
01:39:29
podcast.co scholarships will be coming
01:39:32
soon you can actually see the video of
01:39:34
this podcast on YouTube
01:39:37
youtube.com Allin or just search Allin
01:39:39
podcast and hit the alert Bell and
01:39:42
you'll get updates when we post and
01:39:44
we're going to do a party in Vegas my
01:39:47
understanding when we hit a million
01:39:48
subscribers so look for that as well you
01:39:50
can follow us on xx.com
01:39:53
the Allin pod Tik Tok is allore inor
01:39:57
talk Instagram the all inod and on
01:40:00
LinkedIn just search for the all-in
01:40:02
podcast you can follow chth x.com chth
01:40:05
and you can sign up for a substack at
01:40:07
cho. substack docomo freeberg can be
01:40:10
followed at x.com freeberg and ohal is
01:40:13
hiring click on the careers page at ohal
01:40:15
genetics.com and you can follow saxs at
01:40:18
x.com davit saxs saxs recently spoke at
01:40:21
the American moment conference and
01:40:23
people are going crazy for it it's
01:40:24
pinned to his tweet on his X profile I'm
01:40:27
Jason kakanis I am x.com Jason and if
01:40:30
you want to see pictures of my Bulldogs
01:40:32
and the food I'm eating go to
01:40:34
instagram.com Jason inth firstname Club
01:40:37
you can listen to my other podcasts this
01:40:39
week in startups just search for it on
01:40:40
YouTube or your favorite podcast player
01:40:42
we are hiring a researcher apply to be a
01:40:45
researcher doing primary research and
01:40:47
working with me and producer Nick
01:40:49
working in data and Science and being
01:40:51
able to do great research Finance Etc
01:40:54
Allin podcast.co resesarch it's a
01:40:56
full-time job working with us the
01:40:57
besties and really excited about my
01:41:00
investment in Athena go to Athena wow
01:41:03
Athena wow.com and get yourself a bit of
01:41:07
a discount from your boy jcal beow
01:41:09
wild.com we'll see you all next time on
01:41:12
the Allin podcast