Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
The business of biotech is produced by
0:02
LifeScienceConnect and its community
0:04
of learning , solving and sourcing
0:06
resources for biopharma decision
0:08
makers . If you're working on biologics
0:11
process development and manufacturing challenges
0:13
, you need to swing by bioprocessonlinecom
0:16
. If you're trying to stay ahead
0:18
of the Cell or Gene Therapy curve , visit
0:20
cellandgene . com . When it's time
0:22
to map out your clinical course , let
0:25
clinicalleader . com help , and
0:27
if optimizing outsourcing decisions
0:29
is what you're after , check out outsourcedpharma . com
0:32
. We're LifeScienceConnect and we're
0:34
here to help the
0:40
artificial intelligence and machine learning
0:42
drum beats are only getting louder
0:44
in biotech , and that trend will continue
0:46
for the foreseeable future . When
0:49
a new tech fires up the hype machine
0:51
, it's our job as industry observers
0:53
to suss out what's real . Who's
0:55
actually adopting and applying the technology
0:58
to affect outcomes in drug discovery
1:00
, development and , ultimately , patient
1:02
access ? When virtually every
1:04
biotech who's parsing a little data on
1:06
a computer calls itself AI-powered
1:09
, doing that sussing is
1:11
easier said than done . I'm
1:13
Matt Pillar . This is the business of biotech
1:15
, and on today's episode I'm sitting down with
1:17
the leader of one company that's decidedly
1:20
applying sophisticated computational
1:22
technologies to its biologic
1:24
development efforts . Flagship
1:26
Pioneerings Generate Biomedicines
1:29
led by ex Merck CMO of
1:31
Human Health , Mike Nally . He's
1:33
using machine learning to make big use of
1:35
big data and it's raising big
1:37
money along the way to support the initiative
1:39
. I sat down with Mike in San
1:42
Francisco to learn all about it . Let's
1:44
give it a listen . We're going to talk
1:46
about a lot , but I want to start with you because
1:49
when I look at your background , it's
1:52
not from an early point
1:54
, post academia
1:57
not like it wasn't written in the stars
1:59
. You're going to be a biotech CEO
2:02
. You were a financial analyst at Merck
2:04
. That came fresh off of your Harvard
2:07
MBA , which
2:09
came on the heels of an economics degree
2:11
. That's right . So that was
2:13
in 2003 . And in 2021
2:16
, you were still at Merck . You were CMO
2:18
of Human Health . So there's
2:21
many questions in there , totally One
2:23
of them being that's
2:29
a long run to be at that
2:31
company . So why
2:33
did you decide to leave the comforts
2:35
of a company like Merck , where you
2:37
surely could have rode out the entirety of your career
2:39
?
2:41
It's a really great question . I think
2:44
part of my own personal journey
2:47
. When I
2:49
graduated college , I
2:54
grew up outside of New York City and I thought the path of
2:56
salvation was to be an investment banker because that's what
2:58
all my friends and parents did . My father
3:00
had his own little business and
3:03
in many respects he was the anomaly within
3:05
the community that I grew up with . What
3:07
was his business ? He was in Sporting Goods
3:09
. Oh really , yeah , he was a Sporting Goods manufacturer
3:12
. He'd go over to China manufacturer and then sell
3:14
to the mass merchants and it was
3:16
a great business . But
3:18
he was an entrepreneur through and through
3:21
. But everyone
3:23
else in the town were vendors
3:25
and I
3:27
went and studied finance thinking that's
3:30
what you did , what you grew up
3:32
and I think , as
3:34
I stepped , back after
3:36
the first couple of years I realized I
3:38
was missing that sense of purpose . I was missing
3:41
to what end am I doing
3:43
all this work ? And at
3:45
least for me , the feeling of
3:47
slaving
3:50
away , working 100-hour
3:52
weeks to making money for
3:54
rich people was not how I wanted
3:56
to spend my time . My best friends are
3:59
investment bankers , so I don't mean any disrespect
4:02
for the field , but for me personally I
4:04
wanted a broader sense of purpose in my work and I was
4:06
fortunate in that time to touch on
4:08
the pharmaceutical industry . And when
4:10
I went back to business school I
4:13
went with a distinct purpose of transitioning
4:15
careers . Now I was really
4:17
lucky when I knew
4:20
some of the folks at Merck . One
4:23
of the ladies who was kind
4:25
of a financial Mozart , who's
4:28
Merck CFO the first female CFO
4:30
of a Fortune 500 company said Mike , if you come
4:32
to Merck I'll show you the whole company , and
4:35
this is a lady named Judy Lilliant , and Judy
4:38
basically was true to her
4:40
word . Every 18 months I
4:42
was doing a different sign . So I started
4:44
in the company as a financial analyst doing competitive
4:47
intelligence . Worked on a late-stage pipeline
4:49
. Did the pre-commercialization work
4:51
for a diabetes product called Genuvia . Was
4:54
put into investor relations during the Vioxx
4:56
trials . Went into business development . Did
4:59
the sharing fund merger . So
5:02
if you think about that
5:04
sort of broad exposure
5:06
, I'm a voracious learner
5:08
.
5:09
I was going to say like you're talking about this and I'm
5:11
thinking like you've got to be wired
5:14
a little bit differently , you've got to be a sponge , right
5:16
, I mean you've got to be a sponge .
5:19
Well , I love science
5:21
. I love
5:23
kind of the
5:26
learning process . There's
5:28
so much complexity in
5:31
the end-to-end business model for these
5:33
large pharmaceutical companies . I mean , if you think about
5:35
it , it's a miracle that
5:37
we're able to come up with medicines . Yeah
5:40
, and these medicines are harder
5:42
than putting a man on the moon . Sure
5:45
, in the scientific challenge
5:47
, the manufacturing
5:50
challenges , the commercial challenges , to make
5:52
sure that the world has access to these
5:54
medicines require really
5:57
dedicated , committed
5:59
professionals and as
6:02
I went on those different rotations through Merck
6:04
, I was just really blessed to see the end-to-end
6:06
process , working with
6:09
some of the best people in the industry and within
6:12
any organization , why join Merck I ? Thought it
6:14
was two to three years .
6:15
Right , and when you were
6:17
thinking like at the time , you talked about seeking
6:20
not to get shocked here , but you
6:23
talked about seeking a more meaningful
6:26
end right , rather than getting rich
6:28
and making rich people richer . What was
6:30
the pharmaceutical life
6:34
sciences At that time ? Was that like , well , yeah , that's
6:36
the place . Or why not some other
6:39
philanthropic endeavor ?
6:41
Well , I think combining
6:43
a background in business with
6:46
a endeavor
6:48
that , basically , was
6:50
set up to improve
6:52
the health and well-being of the population
6:54
was something that was really attractive to me . I also
6:57
love the global nature of
6:59
the industry . I loved the
7:02
challenges that the industry were facing
7:05
. The industry's been under kind
7:08
of seminal challenges for
7:10
the last 25 years R&D
7:13
productivity going down . You think about
7:15
pricing challenges , access
7:17
challenges these are things
7:19
that unfortunately have undermined
7:21
the reputation of what was once seen as
7:24
a very novel industry
7:26
and has made it on
7:29
par reputationally
7:31
with tobacco in the
7:33
late public and the
7:36
questions we have to answer and
7:39
solve for as an industry
7:41
are really , really profound . And as
7:43
someone who loves complex challenges
7:45
, as someone who loves these sort of puzzles
7:48
, trying to figure
7:50
out how you
7:52
write the ship so that you can solve
7:54
some of the world's greatest amount of medical
7:56
needs . And when we talk about
7:58
sense of purpose , I
8:00
had the privilege of working on some of
8:03
the most impactful programs
8:05
, products that the industry's
8:07
ever seen Ketrula , indicating
8:11
our 30 different tumor types . You
8:13
think about the number of lives you can
8:15
impact , you think about . Gardasil
8:18
for the HPV vaccine
8:20
, the prospect that we can now eliminate cervical
8:22
cancer from this planet , some of the HIV
8:25
medicines where you're
8:27
offering the prospect ultimately of stopping
8:30
the transmission dynamics in HIV . But
8:33
these are things that have global impact
8:36
and millions of
8:38
people are going to live longer
8:41
, healthier , happier lives as
8:43
a result of that work . And that's what's
8:45
fulfilling what we do and that's
8:47
why a year , two
8:50
years , three years at Merck became
8:52
18 . I was constantly learning . I was constantly
8:55
challenged . I had mentors who
8:57
believed in me in ways that I didn't
8:59
believe in myself . There were jumps that
9:01
I took in my career that I
9:03
probably wouldn't have taken without
9:05
steady encouragement . You know
9:07
, junie was one mentor . Ken Frazier
9:09
, merck's former CEO
9:12
, was another who constantly
9:14
said Mike , we want you to go to
9:16
Sweden and run our Swedish business
9:18
. I had basically been to Sweden for one
9:20
day in my life and knew nothing about the
9:23
Swedish healthcare system . It was asked to
9:25
run that organization and you
9:27
know those were opportunities for me
9:29
to actually become much
9:31
more self-aware as a
9:33
professional , to learn to grow , to
9:36
grow personally , with my family going
9:38
to Sweden , but also professionally
9:40
but also contribute to a much
9:42
greater cause , and that's part of what made
9:44
, you know , the 18 years at Merck so special .
9:47
Yeah , it's interesting the impact you talked
9:49
about , the impact that you were able to play a part
9:51
of in Merck
9:53
. Often when I in specific
9:55
examples , ketrude Gardasil I'm an
9:57
amazing often when I talk
10:00
with big bio turn biotech
10:02
CEOs , they tell me that , maybe
10:06
, ironically , the
10:09
tipping point for them wanting to jump into
10:11
biotech you know , small , nimble , agile
10:13
, scrappy is because they want to feel
10:16
like they have more of an impact than they did in big
10:18
bio .
10:18
So that's sort of an interesting dichotomy . Well
10:21
, I think you
10:23
have more control in
10:26
biotech . You
10:28
know , the litmus test that I had when I thought
10:30
about , you know , leaving Merck was
10:32
wherever I was going to go . I had a belief , fundamentally
10:35
, I could have a greater impact
10:37
on humanity , and
10:40
there aren't many organizations in the world that
10:43
can have a greater impact on humanity than the company
10:45
. If you think about the history
10:47
of the company , yeah , from the mass production
10:49
of penicillin in the 50s
10:51
, all the way through
10:53
a series of antivirals , vaccines
10:55
you know 70 , 11 , cdc
10:58
recommended vaccines were invented by Merck
11:01
the cardiovascular drugs , the
11:04
HIV drugs , all the way
11:06
now through an array of
11:08
cancer products . It's
11:10
a company that's had , you know
11:13
, a huge impact on society . And you
11:15
know , in the 90s Merck
11:19
was the IT company . It
11:22
was Fortune's most admired company seven
11:24
years in a row . And
11:26
that's because of an unparalleled
11:29
track record of not only
11:31
scientific innovation
11:33
but doing good for humanity
11:35
. Programs like the Mectizan donation program
11:37
, which you know had basically a limited
11:40
river blindness , were a huge
11:42
part of what made the company
11:44
a special place . And you know that
11:46
is , you know , this industry at
11:49
its best , where you can do good
11:51
for shareholders while doing
11:53
good for humanity . And that was a core
11:56
part of , I think , what I learned within
11:58
Merck . And you know , as I thought about
12:01
going to biotech , certainly
12:04
in a smaller organization , the
12:07
impact of your actions
12:09
and decisions . You know , at Merck
12:11
you had to influence a 72,000 person
12:14
organization In a biotech
12:16
organization . When I joined Generate , it was 30
12:18
people . Obviously
12:20
, you know each of those 30 employees had
12:23
a huge impact on
12:25
the success of the organization . Sometimes it was harder to
12:27
see that impact in a big company
12:29
. But I think the impact of the
12:31
collective work of the 72,000 was
12:34
enormous and you know
12:36
, depending on how you're measuring , you
12:39
come up with different viewpoints .
12:41
Yeah . So I mean
12:43
, if listeners didn't know that your CEO partner
12:46
, flagship CEO at Generate
12:48
, they might think you're still an executive
12:51
worker . They might think you're passionate
12:53
about the work you did there . Why did you leave ? You
12:56
could have stayed there until .
12:58
I've believed . Merck
13:00
to you . It's a company that is near
13:02
and dear to my heart
13:05
. It's a company with some
13:07
of my best friends , some of my
13:09
proudest professional moments . I
13:12
was really just fortunate to
13:14
uncover an amazing
13:17
opportunity , what I think is a generational
13:19
technology . And
13:22
you know , when I was at Merck one of the
13:24
things that I was fortunate enough to do
13:26
was lead Merck's vaccine business . Well
13:29
, I was leading Merck's vaccine business . We had
13:31
an early partnership with Moderna and vaccines
13:33
. During that time I got
13:35
to know Stefon
13:38
Bunsell and Newbar Fan . So
13:40
Stefon is the CEO of Moderna and
13:42
Newbar is the founder and
13:45
CEO of flagship pioneer and
13:47
we
13:50
had had conversations over the years around
13:53
. You know , if I were to leave Merck , what would I
13:55
want to do ? And you
13:57
know , over time I became
14:00
increasingly convinced that
14:02
the future of the industry
14:04
was going to lie at the intersection
14:07
of science and technology . And
14:09
you know I became , you know , increasingly
14:12
frustrated in many respects
14:14
that the
14:17
industry wasn't having enough success
14:19
at changing underlying productivity curve . We've
14:23
talked about it forever that in four
14:25
decades in a row , industry-wide
14:27
R&D birth have been going down . I
14:31
think somewhat interrelated not entirely
14:33
that has led
14:35
to bad pricing practices
14:38
within the industry . Those
14:41
pricing practices have undermined patient
14:43
access to medicines . That's what creates
14:46
some of the reputation conundrum
14:49
that the industry finds itself with them
14:52
, I became the belief
14:54
on why science and technology
14:56
would be this potential
15:00
answer . It's simply
15:02
the fact that I think it's in our
15:04
view . We'll just look at the
15:07
fact that , despite staring
15:09
at biology in the eye for all of our existence
15:11
as humanity , we only
15:13
know a small fraction of
15:17
the majesty of biology . With
15:20
the advances in technology that
15:22
we were seeing all around us
15:24
, we were entering an age
15:26
where the combination
15:29
of brilliant human ingenuity
15:31
and computation
15:33
and technology was
15:36
going to start to help us understand the language
15:38
of biology in ways that the human mind alone
15:40
couldn't understand it . Fortunately
15:45
, the team at Flagship
15:48
was working on a company called
15:51
Generate Biomedicines . Generate
15:53
was founded five years ago . When
15:56
I first received this call roughly
15:58
three years , a little over three years ago Generative
16:02
artificial intelligence was not part of my lexicon
16:05
. I
16:07
mean , it's still a black box to most
16:09
. This
16:12
team recognized that
16:15
by applying a top-down approach
16:17
to the language of proteins and
16:19
using these sort of generative technologies , we
16:22
would be able to understand that language
16:24
in a way that would allow
16:26
us to ultimately learn how to program
16:28
novel protein-based therapeutics
16:30
with desired attributes . In
16:33
the 20 years I've been part of the industry
16:38
, I had never seen anything like the
16:40
early data . If you think
16:42
about how we discover protein therapeutics
16:44
today , we immunize a human and
16:46
a mouse , a llama we call
16:48
them plasma . We find things that
16:51
stick to interesting targets and
16:53
then we manipulate them in a
16:55
pure trial and error process . We
16:58
have no control over the antibodies our
17:00
immune systems create . We
17:02
then try and make them into drug-like entities
17:05
. Think about the world
17:08
where we say give
17:10
me a molecule with
17:12
this code to
17:15
treat this disease . It
17:19
sounds science fiction In
17:22
many ways . We're not by any means there yet
17:24
, but we're on a path toward
17:28
that destination . This technology was helping
17:30
us start on this to understand that
17:33
language or proteins in really profound
17:35
ways . Some of the
17:37
things that we were able to do were
17:40
just beyond my wildest imagination
17:42
. Some
17:45
of the interesting tidbits
17:47
that I picked up as I was looking into the opportunity
17:50
were the average
17:52
protein . We
17:54
have 20 natural immunosets , the
17:56
average protein may be 200 immunosets long
17:58
. If you think about the
18:01
combinatorial possibilities of proteins
18:03
, it's 20 to 200 pounds
18:05
. That's atoms in the universe
18:07
cubed . The
18:09
search space is extraordinary
18:12
. Oftentimes
18:14
we think about nature as being this all-powerful
18:17
, all-knowing being
18:19
. The reality
18:21
is that nature has always
18:24
surveyed a small fraction of
18:26
that search space . It's just too
18:28
large . If
18:31
you have a technology that will allow you to survey
18:33
those oceans , you're going to find some
18:35
really interesting things . We're
18:37
finding that on a daily basis . These
18:40
technologies are helping
18:42
us understand biology better
18:44
if they're helping us come up with better answers to
18:47
existing conditions , but also giving
18:49
us new tools now to prosecute
18:52
disease in very unique ways we're
18:54
going to get back to the science again .
18:56
I'm just clearly very excited about it . We're
18:59
going to get back to that because I
19:01
want to pause for a minute at the moment of transition
19:03
. You were excited about it then
19:05
and you remain excited about it . You
19:08
join Flagship . You come
19:10
on as CEO at Generate
19:12
2021 . How
19:17
aware were you that you were signing up
19:19
for a post that
19:21
was going to face the headwinds at that
19:23
time ? How aware were you that you were not
19:26
on the scientific and exegesis standpoint You're
19:28
coming with ? Energy . I get it All
19:30
of a sudden . You're like yeah , I'm the CEO of this new
19:32
company we're
19:35
destined to solve all these problems , and then , all of a sudden
19:37
, the whole space
19:40
goes to shit .
19:43
We were really lucky . If
19:45
you think about the peak of the biotech
19:47
market , we started with 2021 . I
19:50
started March of 2021 . I
19:52
don't think people realized that . I think people
19:54
thought it was going to be a momentary blip for
19:57
a while in 2021 . During
19:59
that period , we raised $370
20:02
million as part of our series B financing . We
20:05
raised it from an extraordinary group of
20:08
investors who
20:10
shared our vision
20:12
for the future
20:14
promise of the technology . As
20:17
I think the downturn started to be
20:19
fully realized , we
20:22
were in a very strong position for my capital
20:24
.
20:27
Did you know that coming in , did
20:31
you know this was risky ? There's
20:34
always inherent risk in
20:36
joining a biotech and leading a biotech , but
20:38
we're , relative to the downward
20:41
curve , we're
20:43
in a good position .
20:45
I think this whole idea of
20:47
risk is grossly overstated . Every
20:52
independent biotech endeavor has
20:54
inherent risk . If
20:57
you think about the unmetting and the
20:59
number of opportunities across the biotech ecosystem
21:01
. Failure is part
21:03
of our model . I
21:07
think it's almost your failing for
21:10
valid scientific reasons
21:12
. The biology just is not going to work out . If
21:15
you're giving it a shot , there's going to be
21:18
plenty of opportunity on the other
21:20
side , as long as you manage
21:23
the business side of things Certainly
21:28
in 2021, . If you thought about the
21:30
biotech ecosystem , companies
21:32
were stringing up last and right . Opportunity
21:35
was plentiful . The
21:38
risk of generating not working
21:40
out to me was okay
21:42
. I could do another opportunity
21:45
. I think you've seen
21:47
that across this ecosystem for a long
21:49
time . We're in a glorious
21:52
era of scientific innovation
21:54
. You've
21:57
heard this from , I'm sure , countless people
21:59
. Science has never been better .
22:03
Science needs to be better to
22:06
sustain because we've seen over
22:08
the course of the last 18 , 24 , 36
22:11
months , a lot of great science gets
22:13
shell back burner flush
22:16
. You hit
22:18
an important point as
22:20
long as you manage the business people
22:22
are the rate lender , it's not capital
22:24
.
22:26
Great teams are the rate lender
22:28
. I think there's plentiful
22:30
science and not
22:32
always will the science work out . You
22:35
can have an extraordinary biological
22:37
hypothesis , one of the things that
22:39
you learn this is part of
22:42
. If you think about my time at Merck , it's like
22:44
I got a PhD in this
22:46
industry from Merck . The
22:51
best scientific minds that we've ever
22:53
encountered in life sciences are
22:56
wrong 80 to 90% of the time
22:59
, and
23:01
I'm certainly not in that category . Failure
23:06
is part of this model To
23:11
me , as soon as you acknowledge that the
23:14
way you manage that to
23:16
a certain extent is you build an extraordinary
23:19
team really very people in
23:21
a cross-disciplinary manner . If
23:26
you have an opportunity , you
23:28
find a portfolio or
23:30
platform , and part of what attracted
23:32
me to the opportunity of flagship and opportunity to
23:34
generate was that it wasn't a single
23:36
asset , because , again
23:39
, I'm not smart enough to pick the
23:42
winner , but
23:44
if I have a portfolio , I can then
23:46
manage the
23:49
odds in a very different way . In
23:52
the 2010 assets , the
23:54
expected value starts to become greater
23:56
than one and
23:59
in that sort of a world , I think , a place
24:01
of the strengths of someone who has
24:03
more of a business background
24:05
and
24:07
to me , the period
24:11
of 2021, . The
24:13
biggest issue that we actually encountered
24:16
was delusion of talent . We
24:19
saw this skyrocketing number
24:21
of biotech companies , of public
24:23
biotech companies , of private biotech companies
24:25
, of new startups , and
24:27
that created position and
24:30
title inflation across
24:32
the industry and the world doesn't
24:34
need the sixth and seventh and eighth-in-class
24:37
approach to the same problem . There
24:40
was actually too much . Too
24:43
many companies started and
24:45
unfortunately , I think we've had to go
24:48
through a pretty rough
24:50
datum turn over the last couple
24:52
of years and
24:54
we've seen a number of those companies struggle
24:57
to sustain financing
24:59
through that period . But
25:02
over time , that raw
25:04
scientific foundation is in place , there
25:07
still is a lot of capital and
25:09
certainly the assets
25:12
that show real
25:14
clinical proof are getting funded
25:16
and I think we'll see
25:18
the pendulum . This pendulum swings back and
25:20
forth on product versus
25:22
platform . Right In 2021
25:25
, it was in vogue to
25:27
be a platform , largely because a
25:29
platform technology in mRNA was
25:32
saving the world Sure , and
25:34
in 2022 and 2023 , it
25:36
was much less in vogue to be
25:38
a platform Over time . That
25:40
, oh , the RNA space still .
25:42
There's still plenty of opportunities . There's plenty of platform
25:45
.
25:45
Absolutely , but those things will
25:48
go back and forth in terms of sentiment , and
25:50
so I think for
25:53
me , for the industry
25:56
, this period has
25:59
been one where , I think you rightly point out
26:01
, some really great science will
26:03
get caught up in this downturn . Some
26:07
really great teams will get caught up in this downturn . Some
26:10
teams will get caught
26:12
up in a valuation
26:14
trap in some respects where , if
26:16
you raised in 2021 , 2020
26:18
, at the peak of the market , major
26:21
down rounds would be required to
26:24
sustain organizations , and
26:27
that's really hard , and
26:30
so it's been really a rough
26:32
period . But I think , as we enter this
26:34
year , certainly we now have had the two
26:36
best months on record in
26:38
the biotech industry over the last couple of months . That
26:42
gives us a lot of green shoots as we enter 2024
26:45
to start to make a big difference again .
26:47
Yeah , you
26:49
talked about the fact that you probably wouldn't have signed
26:51
on for such an endeavor had it been
26:54
not a single asset company . And on that perfect
26:56
science , it's logical . In
26:58
the platform space
27:00
or the platform play , what
27:04
sort of thing ? You know you got a single asset . You've
27:06
got to hit milestones to make . You know you got
27:08
to make a money , stretch to the
27:10
next milestone , get more money . You know it's kind of
27:12
that leap probably . So
27:16
I understand where the platform
27:18
play gives you more to work with . But
27:21
how does that change ? Sort of the milestone
27:24
picture
27:26
, right Like you've still got to produce something to reach
27:28
a milestone .
27:30
I think , in terms of the executional
27:33
rhythm that you want to instill within the company
27:35
, I don't think it changes anything . There's
27:38
different milestones , right ? I mean , I've
27:40
got to be as
27:43
accountable to my shareholders
27:46
as a single product . So
27:49
if I say I'm going to advance the technology
27:51
on X , y and Z dimension
27:54
, you know the
27:56
. You know I'm going to
27:58
have to do some of the things that
28:00
I've been doing . So I have to do that and
28:03
if I say I'm going to , you know , have
28:05
two programs into the clinic this
28:08
year . I've got to show products
28:10
on that front and so you know it
28:12
just gives you more levers . And
28:15
sometimes those levers , you
28:18
know , when something you know runs
28:20
a file , you can not you can replace it with
28:23
something else . So
28:25
it gives you a bit more flexibility , a bit more optionality
28:27
, but I think the executional
28:32
rigor has to be the same
28:34
, irrespective of whether you're in a one product
28:36
company or , you
28:38
know , a platform organization . Yeah , I
28:40
mean , that's what builds trust with your
28:43
board , with other stakeholders , with your employees
28:45
, for that matter .
28:47
You mentioned your 2021 series
28:49
B . Yeah , so you raised
28:51
the $280 million series
28:54
C , so I'm hoping
28:56
for some insight on that . How
28:58
you made that happen , what was key to it , how
29:00
it was executed .
29:02
I think it was really key to what was . When
29:06
we look back at what we suggest that we
29:08
would do with our financing
29:10
from our Series B . We delivered
29:13
on milestones so
29:15
we had earned the trust of
29:18
all of our existing investors participating in our
29:20
Series . We had 16 Series B investors
29:23
and all 16
29:25
of them stepped into our
29:27
Series C financing and
29:29
as you can imagine , given market
29:31
conditions , everyone
29:34
was in different situations . They
29:37
all decided to continue to
29:39
put the trust in the team at
29:41
Generate and that was really
29:44
, really meaningful to me , to the entire organization
29:46
. At the same time , we were able to
29:48
attract an array
29:50
of different new
29:53
investors in the process . It
29:55
was much . I mean
29:58
, the level of diligence that went into the
30:00
round was night and day
30:02
between 2021 and 2023 .
30:04
Yeah , that's a huge and overused
30:06
term . Let's unpack that a little bit . That's one of the questions
30:08
I have for you . Is you know
30:10
in more of a rudimentary way to
30:13
put it , like what not
30:15
? Necessarily a frustrate to the way the deals
30:18
are being done today , but what's
30:20
different about it ?
30:21
Well , listen , I think you know again
30:23
, I think that's a big deal , I think
30:26
, of these things . You know , sometimes we hit
30:28
different polls in 2021 and may have been a
30:30
little loose In 2023
30:32
, it was probably , you know
30:34
, the orientation 2021 was
30:36
how do I find a way to deploy
30:38
capital In 23,
30:40
? The kind of orientation
30:43
that the starting point for many investors was
30:45
I
30:47
don't want to deploy capital , but let me
30:49
, I want to look , and
30:51
so you're almost finding reasons not to invest versus
30:54
finding reasons to invest . Yeah , and
30:57
you know , listen that that you know within . You
30:59
know the round that we were able to put together
31:02
the . A big key of it was
31:04
our ability not only to show
31:06
the tracker curve of execution but
31:09
also show the enormous
31:11
promise over the next 24 to
31:13
36 months . And
31:16
you know , as we were in
31:18
the process of raising , you
31:20
know we talked about having , you know , multiple programs
31:22
homegrown programs enter the clinic . These
31:24
are programs that were started by
31:27
generate a couple of years ago using
31:29
the platform . The first one
31:31
that went into the clinic was 17 months from concept
31:34
to clinic . The second one was about
31:36
two years . They're
31:38
addressing major medical needs , their
31:41
technology , their approaches that are
31:43
very distinctive , you know
31:45
, for the individual targets that we've gone after . You
31:47
know , in many ways , you know
31:50
we were able to show promise
31:52
to both our
31:54
existing and future investors that they don't
31:56
said . You know this is a great time , yeah
31:59
.
32:00
Is there a difference in your mind ? I mean , obviously
32:03
there are a lot of variables behind
32:06
the decision to invest in a company
32:08
. Yeah , like generate , any company
32:10
for that matter and you mentioned the importance
32:13
of the support of your existing investor
32:16
base , and then you mentioned
32:19
they got some new ones . Do you see any
32:21
, I guess , lines of demarcation between what
32:23
turned the new ones
32:26
to on the generate
32:28
versus those that you already had ? Like , yeah
32:30
, we're the tire kickers more interested
32:32
in the future ? Were they happy to get
32:34
on board with great investors
32:36
who are happy to see results over the past
32:38
18 ? Months . Like what was sort of ?
32:41
Well , I think you know , every
32:43
investor follows a slightly different form . All
32:45
right . So what you said is going to be
32:47
somewhat of a generalization , but the I
32:50
think you know the commonalities
32:52
that I would pull out right . One
32:55
is you know
32:59
people usually back people and
33:02
you know I've been really fortunate to be able to recruit
33:05
a world-class team and
33:07
that team , I think , instills
33:09
confidence in both new and existing investors
33:12
right and
33:14
but on the jacking of the horse and
33:16
I think there's an element to that
33:18
. I think to the horse
33:20
.
33:22
We don't want to discredit the horse , we
33:24
want to discount the horse , but generally they have a pretty
33:26
cool space . Yeah , right .
33:30
I think you know , certainly you know , over the course of
33:32
the past couple of years it's
33:34
become part of the common lexicon
33:36
and I think we started to see you
33:39
know on an individual
33:41
and personal level , the
33:43
transformative impact that these sorts of technologies
33:45
can have . You know the chat GPT
33:47
raised generative AI in
33:50
the conscious and
33:53
it started to show people the
33:55
enormous potential . I you
33:57
know , before the kind
34:00
of generative AI revolution , most
34:02
people thought of AI and all come
34:05
or technologies
34:07
as largely efficiency
34:09
plays . It was
34:11
about speed and efficiency right . What
34:15
generative AI has shown now ? Is
34:17
it aids the creative process ?
34:20
I can have chat GPT be a thought partner
34:23
while maintaining the speed and efficiency as
34:26
well , or at one hand .
34:28
Yeah , all right , but
34:30
the real value in our industry
34:32
comes from
34:35
novelty and creating
34:38
the best assets . You
34:41
know , if I'm a month faster
34:43
than someone and I'm in fear , ultimately
34:47
clinicians shouldn't use my . If
34:50
. I'm a month faster and superior
34:52
. I'm going to
34:54
have a whole lot of use of that product
34:56
. Yeah , and I think what we're finding
34:59
with these sort of technologies
35:01
and why I think you know part of you know
35:03
you go back to your personality generative
35:05
AI is a really interesting force in
35:08
applying it to biology , because the
35:10
early signals and
35:12
some of the early clinical programs and some of the next waves
35:15
of clinical programs are suggesting
35:17
that we're getting better answers
35:19
than historical approaches enabled
35:22
, and I think that's something
35:24
that if you kind of just keep pulling
35:26
that thread , if you've got a capability
35:29
that can give you differentiated products consistently
35:32
, that's a pretty
35:34
exciting proposition to
35:36
get behind Right
35:38
now I don't want to pull .
35:40
I had a conversation with a guy a bit earlier
35:42
today and I don't want to pull words in his mouth so I'm not
35:44
going to name him , but he was getting
35:46
to a point around we were talking about the
35:49
impact of artificial intelligence and
35:51
its early impact on molecular discovery
35:53
, right , and he kind of alluded
35:55
to a point where he was kind of saying
35:57
like to change the paradigm
35:59
you talked about it earlier the costs
36:02
and access paradigm . There
36:04
needs to be more than faster
36:06
, better molecular discovery , because things
36:10
get really tight and expensive
36:12
, totally complex at the clinical level
36:14
. So
36:16
I guess a big picture . If you look at the
36:18
impact of artificial intelligence and the
36:20
way that you're using it to generate , how
36:23
far along the continuum
36:25
does it actually do envision
36:28
it making an impact on better
36:31
and faster ?
36:33
So I would . I'll
36:38
challenge that underlying premise , but let me
36:40
kind of just start with the end to end
36:42
. View right Artificial
36:46
intelligence , machine
36:48
learning , our tools , the
36:52
tools to apply to
36:54
all parts of our lives . They
36:57
will be applied to all parts of our lives and
37:01
all pieces of
37:03
the drug discovery and development continuum
37:05
will ultimately be transformed by
37:08
these underlying technologies . Today
37:12
we've found the sweet spot
37:14
for our corporate company is
37:17
in molecularism . We're
37:19
starting to use it on target
37:21
discovery approaches and
37:23
we're thinking about how we can use
37:26
data-driven digital
37:28
data approaches in
37:31
trial execution and
37:34
some of the biomarker discovery . But
37:39
I think in
37:42
the future , consistently
37:46
, humans who
37:48
embrace AI co-pilots
37:50
are going to explore our research
37:53
alone , and
37:56
that will be true across all
37:58
parts of all industries .
38:02
You bring this to that point brings up an interesting
38:05
conversation that I've had many times with AI
38:07
for this via text , and
38:09
that's this concept of especially
38:12
those who maybe were born , you
38:14
know , pre-ai acceptance
38:16
or understanding . I've
38:18
had many conversations about the
38:21
nuance of successful integration
38:23
of traditional biology
38:25
and computational biology
38:27
. Total At a physical level , at a
38:29
personal level , at a , you
38:32
know , intellectual
38:34
property level , that's right . So don't
38:36
you manage that ? A generator Like what's , what
38:38
do you ? Is there a challenge
38:40
there ? Or is it a situation where
38:43
you know from the company's outset
38:45
it was we're looking for these very
38:47
open-minded biologists
38:50
who have some computational skill ? Or are
38:52
you marrying two different disciplines in a , you
38:54
know , throw them in a room and make them sorted out , kind
38:56
of way ?
38:57
Well , our two co-founders who still
38:59
are very active at the company , molly Gibson and
39:01
Gavore Gorgon . Both
39:03
are computational
39:06
scientists that
39:09
have studied biology , right
39:11
so . Gavore was a professor at Dartmouth , but
39:13
in both computer science and
39:15
biology Right . Right Molly
39:18
was a PhD in computational systems
39:20
biology from Washington University
39:22
, with a computer science undergraduate
39:24
degree . These folks
39:27
have , you
39:29
know , done a brilliant job
39:31
of creating
39:34
an organization and an
39:36
ethos as of an
39:38
organization where our
39:41
computational scientists are
39:43
not in service of our experimentals
39:47
, Our experimentals are not in service of our
39:49
competition . We are collaborative
39:51
across the domain and we embrace
39:54
people who have
39:56
that multidisciplinary training and
39:58
the single biggest
40:01
predictor of success within January
40:03
is intellectual curiosity , yeah
40:05
, and people who are voracious
40:08
learners , who are interested in
40:11
the domains that none
40:13
of us are experts . In
40:16
every component of
40:18
this complex you know
40:20
way maze that we call drug
40:23
discovery , development , manufacturing , commercialization
40:25
, we
40:27
all are dependent on certain handoffs , but I think
40:29
the people that can stretch across multiple parts
40:32
of this Biological problems
40:34
are really profound
40:36
that we're trying to address . Biology
40:38
is extraordinarily complex and unfortunately we've
40:41
seen , you know , while we've had , you know , a
40:43
lot of organizations with extraordinary data
40:46
in individual disciplines , we've
40:48
only made so much progress , right , having
40:51
people that can span these different domains . I
40:53
think you're going to give us a different set of answers . It's
40:55
not a solve everything .
40:56
No , no , but it's certainly so . I mean
40:59
, I think about it in the context of , like you know
41:01
, a three year old , nine or four year
41:03
old generate had
41:06
the company launch with the same mindset
41:09
, even three years earlier , much less 10
41:11
years ago . Right , that
41:14
talent deficit that you talked about earlier would
41:16
have been even more exacerbate
41:18
, right yeah . Like
41:21
, in what you're doing , are you seeing improvement in
41:23
terms of , you know , replicating Molly
41:25
, and who else did you mention ? Yeah , are
41:27
you seeing , like , are you seeing improvement
41:30
in replicating those skill sets down
41:32
the line , not just at the founder level , all down the line
41:34
? So coming out of academia .
41:36
I mean there were about 100 people
41:38
in the world when I started that generate . They
41:40
were skilled at the intersection of machine learning and protein
41:42
and cheese . We were all under it by
41:44
name . Yeah , I mean
41:46
it was a small niche , community , Right , Good
41:49
place to be .
41:49
I think for those folks , right
41:52
, but you know where ?
41:53
they actually all worked . Yeah
41:56
, deep mind , microsoft
41:58
research , facebook , salesforce
42:01
all had protein teams . Yeah
42:03
, that's interesting and you know , obviously somewhere
42:05
within academia as well , but it
42:07
wasn't a classical path of training and
42:10
you know market dynamics
42:13
are very powerful , supply
42:15
and demand dynamics are very
42:17
powerful . Let's just say , when tech
42:19
enters into the bidding war for
42:22
those types of talent , salaries
42:25
, well , they should change
42:27
Right . And so
42:29
you know there has
42:31
been an influx of
42:33
people , you know
42:36
, that are now entering these , these areas
42:38
, to the point where we
42:40
had a . Our ML team had
42:43
a , an internship program last year
42:45
, and we had five
42:47
spots . We
42:49
had 2000 applications .
42:51
Really yes , and
42:53
these kids like it . I said kids because
42:55
, I'm just these young
42:58
aspiring professionals that are interested
43:00
in a position like that , are they ? Are they coming , these
43:02
2000 people coming to the , the
43:05
, the opportunity with like
43:08
?
43:08
biology , phd , phd
43:11
learning , math biology
43:13
experience .
43:14
That's what I'm wondering , and is it if they're coming
43:16
at it with that scientific
43:18
background and curiosity that
43:20
you would require in an ideal world and
43:23
in the same , like some of our team
43:25
you know , may just be brilliant
43:27
mathematicians that are extraordinarily versatile
43:29
.
43:30
Others will be extraordinary biologists with
43:33
a natural predisposition toward computation
43:35
Right . Some are
43:37
computer scientists . You know that you
43:39
are very versatile , and so
43:42
I don't think there's one single phenotype
43:44
. But you know the
43:46
we're
43:49
really fortunate to have some extraordinary people
43:51
on our team . You know
43:53
our both our computational team
43:55
, our experimental teams , are
43:59
filled with
44:01
brilliant people , and even our experimentalists
44:03
oftentimes have competition , because
44:06
this doesn't work . You know , one
44:08
of the things that you know has been super important
44:11
from the inception of generate
44:13
was how
44:16
do you view data , and
44:19
unfortunately , in many organizations
44:21
, data is power . It's
44:24
my experiment , then
44:27
I have the data , I see what you're getting at and
44:30
you know , when you have a machine learning approach
44:32
, data has to be a collective
44:34
asset and
44:36
you need an orientation
44:39
on the experimental side that recognizes
44:42
that every data point is
44:45
captured to refine the computational approach
44:47
, and that is not your data
44:49
, it's our data , because
44:52
you know the whole machine operates
44:56
on that currency and
44:58
so you know there's a
45:00
huge cultural piece that I think is really
45:03
important for these companies that are starting at this intersection
45:05
. And you know , one of the debates I had for
45:07
a long time with colleagues at Merck was
45:09
will it be the incumbents or the insurgents
45:11
that actually craft this cup
45:14
of tea ? I think incumbents have huge advantages
45:16
, right . Whether you're attacking
45:18
incumbent or you're a pharmaceutical incumbent , you
45:20
know , if you're attacking , you have some of the best computational
45:22
talent , you have extraordinary
45:25
resources . Right , if
45:27
you're on the pharma side , you
45:30
have some of the best biological talent in the world
45:32
and extraordinary
45:34
resources . What you know these insurgent organizations
45:37
can do , though , is create a culture that
45:40
actually blends the best of both worlds
45:42
, and I think it's really
45:44
, really important that
45:46
you know , in pharma , oftentimes
45:49
the computational scientists are in service of the
45:51
biologists , and in
45:53
tech , if you have biologists , they're
45:56
oftentimes in service of the technologists
45:58
. In these sort of
46:00
organizations , the
46:03
ability to create
46:05
alignment and
46:08
you know , kind of a parody that you know these
46:10
two one . You know it's a
46:12
symbiotic relationship in many respects
46:14
between you know , the computational theme and
46:17
the experimental theme , and you know that
46:19
is something that I think has been , you know , really
46:22
, really extraordinary to
46:24
watch , to cultivate , and
46:26
if you're an experimentalist and you don't subscribe
46:28
to , one of our
46:30
five core values is
46:33
around being digitally native and
46:36
around how data has to be a collective
46:38
asset , and if you don't subscribe to that generally
46:41
, it's just not the right place for you . It doesn't mean
46:43
you're a bad person .
46:45
When you say collective asset , I'm curious about your
46:47
take on collective meaning
46:49
. You know within the four
46:51
walls , so to speak , of generate
46:54
versus data being a collective asset to
46:57
the industry , and
47:00
where the lines fall and
47:03
I'm sure it's probably both right
47:05
. Yeah , but how do you like
47:07
when you look at it from an industry-wide perspective
47:10
? Open source data
47:12
, publicly available data , which is incredibly
47:15
important , Huge . How
47:18
does that kind of shake out in terms of generates
47:20
, approach to differentiation and competition
47:22
?
47:22
directly I mean . So when the company started
47:25
, you know , the ratio
47:27
of public to proprietary data obviously was
47:29
one to zero . Right , it's
47:31
all public . Yeah . And you know the
47:34
reason generally it was actually found
47:36
to dissolve the protein challenge in summer's facts
47:38
was because there were two extraordinary
47:40
public data sets . One was
47:42
the sequencing revolution has
47:45
given us 180 million amino acid
47:47
sequences across all species . And
47:50
then , number two , the structural
47:52
biology community gave
47:54
us an extraordinary database called the Protein
47:56
Data Bank , where 200,000
47:59
high-quality , high-resolution structures
48:01
of proteins were exceptionally
48:03
well curated . And
48:05
what that allowed us to start to do is establish
48:07
the interrelationship
48:09
between protein sequence and protein
48:12
structure . And
48:14
if you think about technologies like alcohol or , from deep mind
48:16
, those in some ways were trained
48:19
on these two data sets . So
48:21
the ability to go from a sequence to
48:23
a structure is a prediction challenge , right ? And
48:26
if I start to understand those two data sets I
48:28
can start to make predictions around different sequences
48:30
and different you know structure
48:32
, structural or holding patterns . Generally
48:36
it was always oriented toward adding a third leg to
48:39
the stool . As a therapist
48:41
companies . Sequence and structure are interesting
48:44
. Functions is
48:47
where the business is right . We
48:49
care about how you know
48:51
proteins interact , how they drive
48:54
biological function , right , and
48:56
unfortunately there
48:58
is no public database
49:01
function and there's
49:03
a number of inherent
49:06
limitations in
49:08
the industry to
49:11
create a kind of an open source functional database
49:13
. Primary among them is
49:15
inherent variability of measurements
49:18
. So if you were to run
49:20
an assay and I were to run an assay in
49:22
different labs
49:24
. We make it different values and
49:28
, as we all know , with machine learning based
49:30
technologies , the quality
49:32
of data is paramount . Garbage
49:34
in , garbage out . So
49:37
part of what we've tried to do within the company
49:39
is create proprietary data sets
49:41
of extraordinarily well
49:43
curated data
49:46
. We use a
49:48
lot of automation , we
49:51
use miniaturization techniques
49:53
, microfluidics to do experiments
49:56
at greater volumes , so that
49:58
we can . You know , when we run experiments
50:01
we want to collect
50:03
as much data as possible , and
50:06
that's allowed the company , over the course of the last five
50:08
years , to go from a starting
50:11
point of 100 examples
50:14
to an individual experiment , to
50:16
the biggest experiment we've now run as a million defined
50:19
variants and every one of those
50:21
data points either biophysical
50:23
, measurement or
50:26
functional measurement are captured and
50:29
fed back to refine the computational approach . So
50:31
you know , we've now generated
50:33
sequence and built into full
50:36
length proteins and tested
50:38
around five million proteins and
50:40
all that data now is informing
50:43
our models . And that's helping us now to understand
50:46
. You know what does it mean to
50:48
be a . You know go from a protein to
50:51
a drug . You know key parameters
50:53
affinity how tightly
50:55
do I bind to the target Potency
50:58
, right Immunogenicity
51:01
All of the manufacturability
51:04
and developability parameters that
51:06
are really important . Those are things that
51:08
are measured to actually
51:10
help inform our modeling , to say
51:12
, how do I optimize to go from
51:14
protein to optimal
51:16
therapeutics as quickly , as quickly
51:19
as time , as fashion as possible ? And so the
51:21
data sets are accumulating and
51:23
there are major data gaps that we still see
51:25
, that we're making either
51:27
sizable investments ourselves or we're looking to
51:30
partner right . And
51:32
you know , on the internal
51:34
investment , we've made an investment
51:36
in cryoEM microshills . So
51:39
these are four . We have four cutting
51:42
edge , world class high
51:44
resolution cryoEM microshills . The
51:47
reason we bought those were in
51:50
that protein data bank . Of
51:52
the 200,000 high quality , high resolution
51:54
structures that I mentioned , only
51:57
about 2,000 are
51:59
protein-protein interactions . Most
52:02
of them are pre-standard proteins and
52:04
in medicine or
52:07
in the pursuit of medicines you
52:09
want the interaction . You
52:11
understand what residues are driving function
52:14
in that antibody , antigen
52:16
interaction , and
52:19
we
52:21
decided that let's figure out how to supplement that dataset
52:24
because that data is really valuable for our models
52:26
. On the other side
52:29
, there's a whole array of data that we're
52:31
still very data poor as a company
52:33
and as an industry . Part of why we
52:35
did the partnership we have
52:37
with MD Anderson MD Anderson
52:39
Cancer Center , one of the world's
52:42
busiest cancer center
52:44
in the season many patients at any center in the United
52:46
States was
52:49
we wanted to have understandings
52:52
of the drivers of certain cancer types
52:54
. We understood that at a
52:57
foundational level our technology
53:00
would allow us to drug those
53:03
targets and
53:05
it's been a great partnership because it brings the data
53:08
from a world-renowned
53:10
organization with our unique capability
53:13
to create medicines for patients and hopefully get
53:16
better medicines for patients faster .
53:18
I've been trying to learn about spatial biology . Yeah
53:20
Right , this is a new trend
53:22
. Is that ? Spatial . Are
53:25
you guys part of the player ? Not
53:29
a huge part .
53:31
We combine all components of these sort
53:33
of technologies , but the reality
53:35
with us , for us , is when
53:38
we step back and we think
53:42
about the opportunity that's before us
53:44
. There are these data sets that you rightly
53:46
point out that are missing
53:48
. Ultimately
53:50
, what we're trying to do is
53:52
be able to master
53:55
sequence , the function in
53:58
a relationship , because
54:00
if I know the sequence
54:02
, the function in a relationship
54:05
, what ? That allows me to do
54:07
is program . I
54:11
can sit there and type in the
54:13
code to drive
54:15
the desired function . That's
54:18
what if you ? think about the
54:20
real , long-term
54:22
massive transformation
54:24
taking drug discovery
54:27
from the artisanal craft that is today and
54:30
making biology engineer yeah
54:32
, making it programmed . We're
54:37
on a path toward that .
54:42
We've spent a lot of time talking about AI , and
54:45
that's good . You're giving great concrete examples of
54:47
the work that it's doing . I'm
54:49
curious about your take on the
54:51
investor appetite form because there's
54:53
no doubt a curve a high curve right there's
54:56
a high curve going on right now . For
54:59
the past three years , we've been talking about investor
55:02
discernment . I've got
55:04
all these reasons to be discerning . Why
55:08
, why , right , I don't want to invest
55:10
in you , I should . I'm
55:12
like AI is one of those things that I feel like there
55:15
are so many companies that are riding that high
55:17
curve right now . That's an area where investors
55:19
should be discerning . If you come out , you're
55:21
coming out at a conference like this and
55:23
you say , well , machine learning and
55:25
AI , let's
55:28
pull back covers and have a closer look . Give
55:31
me some color around
55:33
. How , if
55:35
you're speaking to the investment community , which you
55:38
are , I get a lot of VCs listening to
55:40
you how they should
55:42
be discerning from reality
55:44
around artificial
55:47
intelligence forward .
55:48
Yeah , Biotechs right now . I
55:51
mean I always start with the people Right
55:54
. I mean I carefully
55:56
scrutinize the team of
55:59
the organization that we would
56:01
potentially work with partner , with the
56:04
people . Yeah , don't
56:06
look at , we've got an act , we're
56:09
outsourcing to a Look
56:11
at the individuals that are actually driving that
56:13
technology , and are
56:16
these folks among
56:18
the world class contingent
56:22
in those domains ? I
56:25
look at the publications that they've done
56:27
, neither while
56:29
they're within their organizations or in
56:31
prior lives , to
56:34
understand the quality of
56:36
their science . You
56:39
know , we recently published
56:41
one of our models . This is a model called
56:43
from up , and it
56:45
was in nature Right
56:47
. So when people ask me like you know , how
56:50
do I know the quality
56:52
of your machine learning ? You
56:55
can see the major publication Right
56:57
, you have a readout
56:59
Right and
57:02
you'll see . You know
57:04
, we're really blessed with an
57:06
extraordinary team of machine learning
57:09
scientists . I
57:11
think then you know you obviously have
57:14
to look at the technology as well , because
57:16
you know it's great to have great people , but if they're
57:18
not translating that into a technology
57:20
that that's really meaningful , then you
57:22
know the technology , though
57:24
, or the concept , I
57:27
think you know we can all use the same
57:29
words . The question then becomes
57:31
the translatability , and
57:34
you know , what I ultimately want to be judged
57:36
by is
57:39
the output . Are
57:41
we creating better drugs that we could have
57:43
otherwise ? Yeah , and
57:46
you know there are going to be a series of interim markers
57:48
until we hit the real marker . Do
57:51
these drugs make a difference in patients' lives ? And
57:55
that is , you know , the definition
57:58
of success that I
58:00
aspire to , that Generate aspires to . Because
58:03
of the nasancy
58:05
of some of these technologies , we're
58:08
just early in that journey and I think , unfortunately
58:11
, sometimes the long
58:13
lead times in pharmaceutical discovery
58:15
and development obscure
58:18
progress or
58:21
the view of progress . Right
58:23
, the revolution is happening right now . Yeah
58:26
, the outcome of
58:28
that revolution , given the
58:30
regulatory paradigm that
58:33
we have in front of us , is
58:35
maybe a decade away .
58:44
So that long lead time will remain in the interim
58:46
.
58:48
I don't think we clap clinical trials under
58:50
the current FDA guidance in
58:52
the very short term . So we're getting two clinical
58:54
trials quicker . We're getting two clinical trials
58:56
quicker . I mean you're taking , you know , a 10
58:58
to 15 year process coming
59:01
off a couple of years . Ideally
59:05
and theoretically you should
59:07
be getting the better answers with changes probabilities
59:09
, but again we're going to have to
59:11
wait time to judge that Right
59:13
. The
59:16
, I think you know
59:18
, will there be opportunities for further compression
59:20
? Absolutely , will we find
59:22
ways of compressing the clinical trial dynamics
59:25
? I believe so . I believe both . There'll
59:27
be better preclinical models and
59:29
ways of enriching clinical trials to
59:32
do that better . But you
59:34
know the nature of our business
59:36
right now . You know is
59:39
, you know , that
59:41
is the path that we have to
59:43
take and these technologies should
59:46
be held to the same level of scrutiny . As
59:49
you know , a technology that was discovered through more
59:51
traditional means .
59:54
Yeah , so I'll look
59:56
at your crystal ball right Questions
1:00:00
. Everyone hates that answer
1:00:02
. How long
1:00:04
before you and I are sitting here having a conversation
1:00:07
about the impact of these technologies on
1:00:09
clinical trials ? Because
1:00:11
that is at the inflection point
1:00:13
that we're at right now in terms of its
1:00:16
impact on molecular discovery
1:00:18
and design .
1:00:19
I think we're going to start to see impact over the next
1:00:21
few years . Yeah , I
1:00:24
think we'll look
1:00:26
back in the
1:00:28
very near future and
1:00:31
these technologies will be pervasive in
1:00:33
everything we do . Yeah Right , I mean
1:00:35
you know we're just learning
1:00:37
how to interact with
1:00:40
the chat GPT , Do you ?
1:00:42
have any concerns ?
1:00:43
Oh yeah , well of course I
1:00:47
mean these are extraordinarily
1:00:49
powerful tools . But
1:00:52
we already have some pretty extraordinary powerful
1:00:54
tools . These are just imbobial
1:00:56
You're dressed up with right , that's right . And
1:00:59
so you know , I mean like all
1:01:01
of these things have to be pursued with
1:01:03
due caution and due consideration
1:01:05
. But you know
1:01:07
, the reality is , I think the good outweighs
1:01:09
the concerns in a pretty
1:01:11
massive way . You
1:01:13
know , for us , you know it's
1:01:16
interesting . You know , one of the things that people say to me all
1:01:18
the time is like well , mike , how do you deal
1:01:20
with the hallucinations of generative
1:01:22
AI ? Well
1:01:24
, in a construct where
1:01:26
you have this tight integration of dry and wet lab
1:01:28
, we experimentally validate
1:01:31
everything . So
1:01:33
the hallucinations actually are enormously
1:01:36
powerful for our modeling
1:01:38
Because it's iterative .
1:01:40
Yeah , because we learn .
1:01:42
We learn from failure in a way that the industry's
1:01:44
never learned from failure . And
1:01:47
if the model is ultimately hallucinating
1:01:49
that protein doesn't express
1:01:51
and it doesn't work and that data
1:01:53
is already fed back , right
1:01:55
and so it helps us get to a better answer
1:01:57
faster , yeah , and you know
1:01:59
, those sort of bugs are actually features
1:02:01
in many respects in
1:02:04
this sort of world where you're experimentally valid
1:02:06
, I think if you're just thinking that computation
1:02:08
is going to solve everything , you know
1:02:10
I greatly discount organizations
1:02:14
in biology that
1:02:16
think computation will solve everything . Our
1:02:21
workforce is skewed . Everyone
1:02:24
considers us an AIML , a drug discovery company
1:02:26
. Our workforce
1:02:28
is skewed toward experimentalists
1:02:30
. We
1:02:32
have about two-thirds of our workforce that
1:02:35
are experimentalists . That's
1:02:38
just because we have to validate , we
1:02:40
have to measure , we have to generate the data . The
1:02:42
only way to do that is
1:02:45
through experimentation , to
1:02:47
do it reliably For
1:02:50
us . Certainly in the short , medium
1:02:52
and I'd say medium
1:02:54
to long-term , the
1:02:58
paucity of
1:03:00
data , functional
1:03:02
data , the data that we really care
1:03:04
about across a number of
1:03:07
domains in life
1:03:09
sciences , will lead
1:03:11
us to a place where we
1:03:13
are constantly
1:03:15
needing more and more information to
1:03:18
refine these computational approaches , to get to the best
1:03:20
answers to the status as possible .
1:03:22
Yeah , it's fascinating
1:03:24
. I'm curious about where
1:03:26
you are . If you want to share some information on
1:03:28
where you are right now and what the next
1:03:30
step might be for generating
1:03:32
.
1:03:32
Yeah Well , I mean excitingly
1:03:35
, our first two molecules entered the clinic in 2023
1:03:39
. You
1:03:41
know that is a
1:03:43
major milestone for any company . We're
1:03:46
extraordinarily proud that we're
1:03:49
now starting to see the benefits in
1:03:53
people . Those
1:03:57
programs we think could
1:03:59
address major areas of unmedical
1:04:01
need . Behind it , though , we have
1:04:03
roughly 20 programs . A
1:04:06
number of those are in partnership . We have six
1:04:08
in partnership with Amgen . I mentioned
1:04:11
the MD Anderson partnership , where we're now working on
1:04:13
three different targets . We're working with Roswell
1:04:15
Park Cancer Center on
1:04:18
a number of other targets . Then
1:04:21
we have an internal pipeline for generate
1:04:25
around 8 to 10 programs that we're working
1:04:28
on from a proprietary standpoint
1:04:30
. We're in this transitional
1:04:32
state of an organization where for
1:04:34
the first few years we were really a
1:04:36
pure technology company . We
1:04:39
were developing a technology . Now
1:04:43
we're morphing
1:04:45
into a technology
1:04:47
company that discover drugs
1:04:50
and that
1:04:52
develop drugs . You're
1:04:55
seeing a bit of a transformation within the organization
1:04:57
to bring in some
1:04:59
of the skills and capabilities I
1:05:03
think from building the organization that
1:05:06
creates some interesting cultural dynamics with
1:05:09
what we're working through . Clearly
1:05:12
we need
1:05:15
the experience of seasoned
1:05:17
drug hunters and drug developers
1:05:20
with the pioneering
1:05:22
spirit of the original technology
1:05:25
developers . It's something that we spend a lot
1:05:27
of time thinking through . Then
1:05:30
, from a technology perspective , we're
1:05:32
scratching the surface of application
1:05:35
. I always say to the team the
1:05:38
one thing I know definitively is
1:05:40
I don't know the best use of the technology
1:05:42
at this point . I
1:05:45
know there are multiple multi-billion dollar applications
1:05:47
for the technology , but
1:05:51
where this will have the greatest impact on
1:05:53
humanity we're still searching
1:05:55
and that leads to a bit of
1:05:57
a dynamic where
1:05:59
you're constantly balancing exploration
1:06:02
and exploitation . When
1:06:04
you find those areas where we can make
1:06:06
impact , how do we do more
1:06:09
of those versus expanding the
1:06:11
aperture ? The technology
1:06:13
is protein modality agnostic . It works
1:06:15
for antibodies , peptides , enzymes
1:06:18
, cell gene therapy , all our protein
1:06:20
, those
1:06:22
ABCs , bi-specifics . That's
1:06:25
a good place to be Right
1:06:27
. You have a ton of real estate
1:06:29
to cover . Which
1:06:32
of these will have . Which of those domains
1:06:34
will we have greatest advantage over traditional
1:06:36
approaches ? We'll still learning . My
1:06:38
guess is it's going to
1:06:40
be a more complex biology . If
1:06:43
you think about where
1:06:46
computers routinely outperform
1:06:48
humans , it's
1:06:50
in complex multiparticular optimization
1:06:52
. The more
1:06:54
complex the biological
1:06:56
construct in menu or stacks , I think the
1:06:58
computer ultimately will
1:07:01
show inherent advantage . Sure
1:07:05
, those challenges as we go forward
1:07:08
are super interesting
1:07:10
. You're trying to do that while you're trying
1:07:12
to execute on these clinical programs , bring the next generation
1:07:14
of programs to the fore and
1:07:17
developing a technology that I
1:07:19
believe will ultimately change
1:07:22
the way the
1:07:24
whole industry creates large
1:07:26
molecules . I
1:07:28
don't say that lightly
1:07:31
. We can debate how long
1:07:33
that may be , but
1:07:36
I think it's inevitable that over time
1:07:38
these approaches will
1:07:41
provide a
1:07:44
foundationally better
1:07:47
way of coming up with the right answer
1:07:49
than the traditional artisanal
1:07:51
approach .
1:07:53
I know we're running short on time here so I can ask
1:07:55
you questions Well into the evening
1:07:57
, but I'm curious about one of the things you mentioned
1:08:00
is the cultural change
1:08:02
of generating to
1:08:04
a drug development company , not away
1:08:07
from a tech but being built on a technology company
1:08:09
. I'm curious about
1:08:11
how , if you're doing that , if you work leading a
1:08:13
company that's doing that on an island
1:08:15
unto itself , the
1:08:17
challenge , it occurs to me might be more daunting
1:08:20
than it would be for flagship
1:08:22
pioneering partner CEO . How
1:08:25
much support like how advantageous is
1:08:27
it for you I'm
1:08:30
not insinuating that perhaps you're snuggling
1:08:32
drug development telling from other flagship companies
1:08:34
but how advantageous is it for you to be
1:08:36
part of that ecosystem ?
1:08:38
I mean , listen , it's a special
1:08:40
ecosystem . What
1:08:42
flagships we're doing for the bill is
1:08:45
really extraordinary .
1:08:47
It's not for the faint of heart .
1:08:51
Most innovation occurs in an incremental fashion
1:08:53
. Most
1:08:55
innovation from most organizations
1:08:58
is around the edges . Flagship
1:09:01
lives in the white spaces and
1:09:04
even the name of the organization is
1:09:06
highly symbolic , right Flagship
1:09:09
, pioneering yeah . A
1:09:12
lot of the time we spend thinking about
1:09:14
is what does it mean to be a pioneer
1:09:16
? What does it mean
1:09:18
to push into
1:09:20
new areas
1:09:22
, frontiers ? How
1:09:25
do you orient and how
1:09:27
do you make it inhabitable for
1:09:29
other settlers ? How do you keep
1:09:31
?
1:09:31
pushing boundaries Right , the
1:09:33
immigrant mentality I agree 100% .
1:09:35
Every time I talk with a flagship Absolutely
1:09:38
, and I think
1:09:40
one of the other things that we spend a lot of time thinking
1:09:42
about are dualities
1:09:44
of leadership . Oftentimes
1:09:47
we think about the
1:09:50
world in black and white terms
1:09:53
and every once says well
1:09:55
, no , let's live in the gray zone
1:09:57
. Oftentimes
1:09:59
, what we believe is that you need
1:10:02
polar experience
1:10:04
, you need to be able to balance between the two poles
1:10:06
, not in between , and
1:10:08
it's those sort of mindsets
1:10:10
that , I think , become super
1:10:12
valuable in these sort
1:10:15
of domains . And
1:10:19
you probably have heard a new bar talk about paranoid optimism
1:10:21
. That's one of those
1:10:23
dualities . Sometimes we need to be
1:10:25
paranoid , oftentimes we
1:10:27
need to be optimistic , and
1:10:31
rather than sitting there in between , how
1:10:33
do we embrace both of the poles as
1:10:36
we develop these sort of organizations ? And so
1:10:38
certainly it's
1:10:42
really nice when you're doing
1:10:45
things that people haven't done before
1:10:47
and you have
1:10:50
guides that have
1:10:52
been on similar journeys that
1:10:55
can help you understand from their own experiences
1:10:57
, and the team
1:10:59
at Flagship , as
1:11:02
I mentioned , stefan , who is
1:11:04
a member of the board who's created
1:11:07
, obviously , moderna . They've
1:11:09
been huge , I
1:11:13
think teachers For
1:11:15
me , for the team that generates , sharing
1:11:17
their experiences in a very
1:11:19
open , transparent way and
1:11:22
, more importantly , not just the successes
1:11:25
but the failures
1:11:27
, because we oftentimes talk
1:11:29
a lot about what went well
1:11:31
, but you sometimes learn the
1:11:33
most from
1:11:35
the mistakes that organizations
1:11:38
make on these sort of journeys .
1:11:39
Sure , it's stated
1:11:41
in the outside it's a failure
1:11:44
, is inherent , and if you're not taking it away from that
1:11:46
, then that's right , that's on you .
1:11:48
You're losing . And
1:11:51
the team at Flagship New
1:11:53
Bar , jeff Aba
1:11:56
, that we're kind of , the founding team of
1:11:58
Generate have been
1:12:00
extraordinary
1:12:03
mentors , ambassadors
1:12:06
, teachers , guides
1:12:08
. As we've gone on this journey and
1:12:11
for someone that is coming from a
1:12:13
big company doing
1:12:15
this for the first time , being
1:12:18
able to surround myself with those
1:12:21
sort of people and a whole cohort of other
1:12:23
CEO partners that are
1:12:25
also doing this journey
1:12:27
for the first time , with an array
1:12:29
of different backgrounds- that's
1:12:32
an enormous asset Because , as you know , you
1:12:34
talk to a lot of folks in these sort
1:12:36
of roles . These roles can be really
1:12:39
lonely , and having
1:12:41
that sort of ecosystem to pull upon
1:12:43
is of huge , huge
1:12:45
value and it
1:12:47
gives you a sense of
1:12:49
calm and
1:12:52
massive places of
1:12:54
uncertainty , and I'm enormously
1:12:56
grateful for that .
1:12:58
Well , I'm enormously grateful for you taking time
1:13:00
on this busy week to spend with me
1:13:02
on the business biotech , so I really
1:13:04
appreciate it . I'm Matt Pillar and you just listened
1:13:06
to the Business of Biotech , the weekly podcast
1:13:09
dedicated to the builders of biotech
1:13:11
. We drop a new episode with
1:13:13
a new exec every Monday morning and
1:13:15
I'd like you to join our community of subscribers
1:13:18
at bioprocessonlinecom
1:13:20
Apple Podcasts , spotify , google
1:13:22
Play or anywhere you get your podcasts . You
1:13:25
can also subscribe to our Never
1:13:27
Spammy , always insightful monthly
1:13:29
newsletter at bioprocessonline
1:13:32
. com/bob . If
1:13:34
you have feedback or topic and guest suggestions
1:13:36
, hit me up on LinkedIn and let's chat and
1:13:39
, as always , thanks for listening .
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More