00:00hi and welcome to the a 16z podcast I'm
00:02Hannah and we're here today for the
00:04first time ever with the bio team all
00:06together to take a pulse on where we are
00:08on the intersection of bio and
00:09engineering and what it makes possible
00:11the conversation includes general
00:13partners Vijay Pandey and Jorge Conde
00:16and Malenko well Ali a day interviewed
00:18by Jeffrey Lo and covers everything from
00:20what the shift away from empirical
00:22science towards engineering means for
00:24bio to what we're looking for in
00:26entrepreneurs now that we've announced
00:27our second bio Fund today we're going to
00:30talk about what we've seen at the
00:31intersection of biology and computer
00:32science and how engineering and biology
00:34is changing how we think about bio
00:36investments let's start out with
00:38computational biomedicine VJ how do we
00:40think about this at the start of the
00:42first bio Fund actually this was very
00:43much an inspiration for the first fund
00:45in the first place that we saw the
00:46existence of companies that are really
00:49tech companies at their heart but that
00:51can be built in the biology and
00:53healthcare space and that machine
00:54learning was a key means towards that
00:56end and what have we seen it over these
00:58past two years one way to divide
01:00medicine up traditional ways between
01:01diagnostics and therapeutics in the
01:04diagnostic space here there's a very
01:06natural trend that you take some new
01:07data source whether that would be
01:09genomics or wearables or other new
01:12technologies and then you marry that
01:14with artificial intelligence or machine
01:15learning some means to go through all
01:18this data and to gain insight faster and
01:20higher accuracy with continued learning
01:23in a way that we really couldn't do
01:24before and then finally to toward some
01:27ends that is actionable so a great
01:29example this is something like free
01:30gnome where you take genomics and the
01:33data from what the DNA in your blood
01:34tells you about your immune system but
01:36you know we don't understand their
01:37immune system so we use AI to be able to
01:39sort of tell us what this means with
01:41high accuracy and then it's very
01:42actionable that you would get the
01:44appropriate cancers procedure done
01:45especially if you could catch cancer
01:47early this is where AI could really be a
01:49part of the key missing link towards the
01:51cure cancer two years ago there weren't
01:53dominating examples of prominent
01:55journals publishing AI and healthcare
01:58pieces where it was you know shown to
02:00work in healthcare context and and now
02:01it feels like every month there's a new
02:03nature paper or something coming out
02:05where you know it's been shown to be
02:07demonstrated to be much more accurate
02:08than than a human physician the
02:10application of AI or machine learning
02:12actually doesn't allow you to just do
02:15something better or faster or cheaper it
02:17actually allows you to do something that
02:19previously was impossible one of the
02:21biggest challenges in developing
02:22therapeutics has been understanding
02:25where to target specifically a disease
02:27what mechanism what what target to go
02:29after and how to develop chemical assets
02:32against that and I think one of the
02:34things that machine learning can help us
02:35with is to unravel the inherent
02:37complexity of biology in a way that
02:39we're not depending on our human
02:41understanding our human minds
02:43understanding of the complexity of the
02:45disease in order to determine how best
02:47intervene so Diagnostics therapeutics
02:49these are two areas that we've seen AI
02:51applied is that where the future is or
02:53is it in a third space it's really just
02:56one subset of what we think is the
02:57broader theme here which is this shift
02:59away from biology being primarily an
03:02empirical and experimental science to
03:05becoming more of an engineered
03:06discipline and so machine learning
03:08artificial intelligence is one example
03:11of the application of an engineering
03:12approach to biology but there are many
03:15others and what do you mean by an
03:16engineering approach we lack a sort of
03:17fundamental foundational understanding
03:19of what drives most biology and
03:21specifically what drives disease biology
03:22you have to essentially say I have a
03:25hypothesis that this particular target
03:27may be the you know optimal point of
03:29intervention for you know for
03:30intervening in a disease process and
03:32that requires experimentation and as we
03:35know with experimentation sometimes it's
03:37successful oftentimes it fails so when
03:39you take more of an engineering based
03:40approach it's the discipline of saying
03:42well let's take the science risk out of
03:44it and there many ways that you can do
03:46that one way is as we just covered is to
03:48use machine learning or artificial
03:51intelligence so that the algorithm
03:53itself can learn the underlying biology
03:55in a way that we can't as we discovered
03:57but another way is to focus on areas
03:59where the biology is relatively well
04:01understood and now find ways to improve
04:04our ability to intervene so one example
04:07on the therapeutic side might be in
04:08areas where the cause the disease is
04:11very well understood so in sickle cell
04:13anemia it's long been understood that
04:15the cause of the disease is a mutation
04:18in hemoglobin that causes red blood
04:20cells to fold or sickle and so if you
04:23wanted to develop a therapy for sickle
04:26what you would need to do is to find a
04:27way to either replace or repair
04:30hemoglobin so that your red blood cells
04:32don't fold and there are many
04:34engineering focused approaches to doing
04:36that namely CRISPR or gene therapy that
04:40are focused on a very specific way to
04:42solve what is it already known as the
04:44biological problem to give you a
04:46contrast to that take a disease like
04:48Alzheimer's where we actually don't
04:50really understand what is driving the
04:53disease and if you don't understand the
04:55biological basis of the disease then you
04:56have to hypothesize on what you think
04:58might be the best point of intervention
05:00and that's very challenging that's a
05:02very science-based approach you
05:03experiment you test and as we've seen
05:06very often unfortunately many of the
05:08therapies that have been developed for
05:10Alzheimer's fail and often they fail
05:12very late in the process in a phase 3
05:14trial which of course is very expensive
05:15and very time-consuming and so that's
05:18how we sort of differentiate it how can
05:20we think about focusing on opportunities
05:23where there's not a lot of science risk
05:25a novel biology risk and focus instead
05:29on opportunities where there is perhaps
05:31engineering and scale risk and that's an
05:34area that we think we're well positioned
05:35to help support entrepreneurs in that
05:37space for decades anything in this bio
05:40space biotech space would be really
05:41dominated by how to mitigate science
05:43risk and that the new opportunity now is
05:46to new technology either machine
05:48learning the computer or otherwise that
05:49there are more and more areas where that
05:52science risk is no longer the issue and
05:53that's the opportunity I think that we
05:55were looking to go after it's remarkable
05:57to have seen how broadly applied it is
05:59so we've seen this in dermatology
06:00analysis of the retina we've seen this
06:03in applications of actually using facial
06:05features to identify genetic diseases
06:07we've seen this in analyzing pulse from
06:11an Apple watch to diagnose heart
06:13conditions so this is a remarkable sea
06:15change in the way we think about how we
06:17can collect data and make sense of it
06:19using this intersection of sensors and
06:21artificial intelligence to really make
06:23sense of what's driving disease the
06:25other big changes is that when you have
06:27these new data sets AI allows you to
06:29create whole new areas much like you
06:32create new apps so like you know in the
06:34old days with before the smartphone you
06:35had your GPS box and your camera and
06:38your phone and you have to have a whole
06:40note a piece of hardware similarly in
06:42Diagnostics if you have your PSA tests
06:43you're not gonna do a Brockett test or
06:46genomics could be used for so many
06:47different things you have data for
06:48wearables can be used for so many
06:50different things and often it's not sort
06:52of whole new science it's just repeating
06:54the exact same engineering process you
06:55had to get the first test for freedom to
06:57do one type of cancer and do a second
06:59type of cancer from genomics is really
07:01essentially running the exact same you
07:02can contrast you know a company is able
07:04to develop a completely new test from
07:07its data compared to old diagnostic
07:09companies that you know have to develop
07:11new things you know science is something
07:13that you can't schedule creativity or
07:15the ability to come to a breakthrough so
07:17if you have understood the science of
07:19colorectal cancer and what you've
07:20learned about a stool test probably is
07:22not gonna be very useful for a breast
07:23cancer test it'd be a completely
07:25different biology but if you're using
07:27genomics and you can be able to learn
07:29from training from samples it's
07:30basically just getting new data and new
07:32samples and repeating the exact same
07:34process I mean this speaks to sort of
07:35the different ways in which this has
07:37shifted from science engineering and the
07:39ways that you know engineering and AI
07:41makes it more accurate and more
07:43repeatable there's also been non
07:45artificial intelligence means of
07:47engineering where has engineering and
07:49biology come in that has not involved
07:51artificial intelligence one of the key
07:54areas where we see biology itself as a
07:56tool for engineering has of course been
07:59in the gene therapy space in in the
08:02CRISPR space the technology that allows
08:04us to do very precise gene editing or
08:07genome-wide editing and we've also seen
08:09at the cell level with the cell
08:11engineering love you can now use these
08:12very precise tools to either replace DNA
08:16or edit DNA or in some cases even write
08:19DNA and that's very much an engineering
08:21take for example next-generation
08:23sequencing for DNA today because of
08:26improvements that were born of the
08:29application of engineered disciplines
08:30and three very specific ones we can now
08:32sequence a human genome and what used to
08:34take 13 years in a matter of hours and
08:36what used to cost three billion dollars
08:38for less than a thousand and really what
08:41drove that wasn't some fundamental
08:43discovery on the biology or the science
08:45what drove that was aluminium was able
08:48to apply three very specific in
08:51engineering disciplines and converts
08:53them one was the use of micro fluidics
08:56the just a fancy word for saying to use
08:59piping to move around tiny amounts of
09:01liquids and chemicals so you can run
09:03experiments in a very very dense
09:04environment the second one is the use of
09:06optics so you can in fact detect
09:09chemical reactions again happening in
09:12very dense environments and the third
09:14one is the use of compute to enable you
09:16to actually string together all of this
09:17data to actually recreate what would be
09:19a representative genome and it was
09:21really those three engineering based
09:22disciplines that gave us the ability to
09:24do genomics as it exists today it takes
09:27something from being impossible to now
09:29possible to eventually become in routine
09:30and then from there indispensable we see
09:33the same thing happen in applied
09:35engineering approaches to how we design
09:37monoclonal antibodies we've seen this
09:40now with mRNA and the ability to produce
09:42that on demand and we've seen this again
09:44and again one of the very interesting
09:46things about this particular space is
09:48these companies will look very tech like
09:51in terms of their ability to create new
09:53markets and scale and dominate those
09:56markets and those are obviously very
09:57attractive and interesting opportunities
09:58for us one really interesting aspect on
10:00the engineering biology side is oh
10:02there's this entirely new wave of cell
10:04therapies that's coming out which is an
10:05entirely new modality of drugs I mean we
10:08started with small molecules then
10:10proteins in the 70s or 80s with
10:12Genentech and now there's this entirely
10:14new modality coming out which of course
10:15means that there's a lot more demand for
10:17types of for delivery methods and in
10:19terms of tools for this modality yeah we
10:21had small molecules then we have large
10:23molecules engineered cells represent
10:25essentially living drugs which is a
10:28remarkable advancement and I think one
10:30of the things that we're seeing is that
10:32now when we can start to design cells we
10:35can actually start to design cells that
10:37have logic that will know where to go in
10:39the body what to do when they encounter
10:41disease and how to essentially terminate
10:44themselves when the disease state has
10:45been resolved so we're on the cusp of
10:48not only going from having living drugs
10:49but having intelligent drugs and that is
10:52an incredible step forward well let's
10:54move on to digital health what's the
10:56state of the field of digital health now
10:57and and where's digital health going one
10:59of the huge opportunities there is that
11:01one can actually first off design
11:05therapeutic so that don't have any
11:06toxicity when we think about therapeutic
11:08week think about something like an
11:09antibiotic and an antibiotic is kind of
11:12a magical thing right take this pill and
11:14like a week later you're done and you're
11:16fine but there's no equivalent for that
11:18in so many areas so there's no pill that
11:20you take for a weekend you have no more
11:22PTSD or you have no more anxiety or you
11:24have known depression or you have no
11:26more type-2 diabetes so what these
11:29things have in common first off is that
11:30the biology very complicated it's not
11:32something where there's some invading
11:34pathogen that you need to get rid of but
11:36we actually do have therapies for them
11:38they're just not pills these therapies
11:40are behavioral therapies and often have
11:42pretty good efficacy but they're very
11:44expensive and hard to scale a great
11:46example of this is the CDC's diabetes
11:48prevention program that existed and was
11:50validated as a science was de-risk by
11:53the CDC and others and so that was very
11:55appealing and you know companies like
11:57Hamada now can take that engineer it
11:59scale it and the intriguing thing is
12:01that they can do the equivalent of
12:03clinical trials except in computer land
12:05the clinical trials a B testing and they
12:07can do this AV testing you know if they
12:09wanted to once a week at scale they can
12:12constantly iterate to make their
12:14therapeutic have higher efficacy without
12:16having any issue with toxicity the
12:18challenge has been very much to find
12:20areas where you you could make that
12:22similar impact in areas other than type
12:232 diabetes we've seen Network effects as
12:26an area that has changed a lot of the
12:27tech ecosystem how those played out in
12:30health care space we've talked about
12:31network effects at length before but the
12:33quick just stays a network effect is any
12:35company that's able to build a network
12:37where the more people that join the
12:38network the more powerful and defensible
12:40that company is there's a lot we've
12:42learned in the pure tech consumer tech
12:45side of the world in terms of network
12:46effects that we've seen in the bio space
12:47so an interesting example is this
12:49company called patient paying which
12:51there's something called care
12:52you know hospitals are increasingly
12:54taking financial risk on their patients
12:56they are financially in the hook for
12:58excessive care care coordinates at this
13:00hospital actually take risk on patients
13:02who leave the hospital and go to
13:03external facilities but they have zero
13:04idea where these patients actually go
13:07but somehow they're financially liable
13:08for these people which is insane if you
13:10think about it and so what patient pain
13:12does is they actually connect all these
13:14providers together and care coordinate
13:17is in the hospital and other places get
13:18paying anytime one of their patients
13:20goes in checks and someone else it's a
13:21very simple product but what's really
13:23powerful there is they form a really
13:24really powerful network effect and the
13:26more providers join the network the more
13:29valuable the network is because the more
13:31granularly they can track their patients
13:33and so this is a very much a concept
13:35from the traditional attack and consumer
13:37world that's been brought into
13:39healthcare it's it's difficult to
13:40develop a one they do happen they're
13:41very powerful that's said now this is
13:43just one type of network effect this is
13:44a people network effect there's there's
13:46others of course the real heart of
13:48network effects is to try to create two
13:50things one is a barrier to entry but
13:51also there's this powerful fact that you
13:53know as you get more customers this
13:55beared entrant grows and one way to do
13:57this is to are so-called data a network
13:59effect and we've seen this with
14:00companies like free no more cardiogram
14:02in these diagnostic spaces where as each
14:05of them gets more data they get more
14:06accurate the tests go from like 85 to 90
14:09and 95 97 and hopefully you know even
14:12higher percent accuracy and then as they
14:15get higher because you tests they get
14:16more customers because they have a
14:17better product which of course gets them
14:19more data and it just spins up from
14:20there what's intriguing is that this is
14:22all about barrier to entry and in
14:24traditional biotech Behrendt entries
14:26largely due to patents and patent
14:29windows shrinking it's such a challenge
14:30but you know data network effects never
14:33go off patent they just get stronger and
14:35stronger and help companies grow even
14:36after even decades after one of the
14:38things that I've seen since joining is
14:39we're looking more and more into
14:41therapeutics we're looking for companies
14:43where the science risk has been removed
14:46or greatly de-risked and so that often
14:48sort of removes a lot of traditional
14:51type of therapies companies companies
14:53where there's a huge amount of science
14:54risk but I think there will be examples
14:56of therapeutics where are driven by a
14:59much more engineering approach whether
15:00at home but engineering cells using AI
15:03for new ways to find new small molecules
15:05for biologics I think there's a lot of
15:07potential there what's fascinating about
15:08you know what we would call traditional
15:10therapeutics is that it's historically
15:13been a very sort of bespoke effort so
15:15you're focused on a disease area you I
15:18try to identify a target where you want
15:20intervene and then you make a very
15:22specific thing a specific molecule that
15:24will hit that target and ideally only
15:27and so by extension the next time you
15:30want to make a second drug or another
15:32drug in many ways there's not a whole
15:34lot you can take from your experience
15:36from having discovered and developed the
15:38first drug because again it's a very
15:40bespoke thing so I think one of the
15:42things that we think about from an
15:43engineering based approach is where are
15:45there examples where you can actually
15:48transfer knowledge and experience from
15:51the first to the second to the third and
15:52so on and the application of AI and drug
15:55discovery I think is one example and I
15:58think there are other examples that we
15:59look at that are biologically
16:00engineering based approaches whether
16:02it's you know CRISPR or mRNA or any
16:05other number of things where the tool is
16:07a biological tool where really you're
16:09using a modality and you're just
16:10swapping out you know the code that
16:13you're actually inserting into the drug
16:14in the case of CRISPR or into the case
16:16of of mRNA and so those are the kinds of
16:19things that I think are potentially more
16:21interesting to us because it takes drug
16:23development from being a very bespoke
16:24thing to being something that's more
16:26generalizable once you understand the
16:28underlying disease biology are you
16:30interested in in modalities with a tool
16:32like CRISPR is the therapeutics itself
16:34or are there other ways where crispr RNA
16:36could be involved in a therapeutics
16:39company well I think CRISPR is a great
16:41example that shows that innovation in
16:43our space is accelerating CRISPR is a
16:45concept barely registered just a few
16:47years ago and now it's an indispensable
16:49tool for drug discovery for actually for
16:52understanding biology as we've seen the
16:54cycles of iteration in biology
16:56accelerate as we move biology more and
16:59more into an engineered based discipline
17:00I think we're going to see more powerful
17:02modalities like CRISPR and in fact
17:05CRISPR already has emerged and to
17:07various different flavors and I wouldn't
17:10be surprised if we see sort of the next
17:12new thing emerged just think about
17:13what's emerged over the last two years
17:15in biology it's kind of breathtaking you
17:17know compared to let's say the previous
17:18ten years and I think we're going to
17:20continue to see those opportunities
17:21emerging ok you know as we go from hype
17:24to creating real companies these
17:25companies all will have to go to market
17:27so what's been working what channels
17:29have been working and what have the
17:30company's been doing there's a couple
17:31different stages and you have to do
17:33within different types of people that
17:35you have to get to so first off often
17:37you're doing with the CMO and you have
17:38to convince him or her that this
17:41demonstrates some degree of efficacy but
17:43clinical trial some other study or
17:45something to point to but then actually
17:47beyond that then you have to get to a
17:48point where you get reimbursed and
17:50demonstrate there's real value and
17:51that's the next step that's when you can
17:53start to get commercial traction so we
17:55start with evidence then go to
17:57reimbursement and yet you know who are
17:58they getting this money from a very
18:00common pathway when we first started was
18:02going to self-insured employers so most
18:05employers over 500 employees will self
18:07insure what that means is they hold the
18:10financial risk the employer actually
18:12pays for the employee's health care and
18:14then they work with a third party
18:16administrator or an administrator
18:18services organization like an Aetna or
18:20Cigna whatever to actually do the claims
18:23processing on the back end when a
18:25company gets large enough they can
18:26self-insure and all the financial risk
18:27so a lot of startups started with
18:29self-insured employers and they that was
18:32actually a pretty popular approach early
18:34on and here's why they started with
18:36self-insured employers it's because as a
18:38healthcare company when you sell two
18:40pairs when you can share healthcare
18:41service two pairs there's two different
18:43value props that you can sell against
18:45one is you actually save them money on
18:47healthcare costs and the other is it is
18:50an engaging product that can be used by
18:52their members or their employees which
18:54makes those members employees like the
18:56employer or health insurance plan more
18:58now health insurance plans care a lot
19:00more about saving money than offering an
19:02engaging product and so for an
19:04early-stage startup it's a lot harder
19:06for them to do the saving money piece
19:07because that takes a long study to prove
19:09and they do it eventually
19:10but it's hard to do that initially but
19:12they can show that it is an engaging
19:13product early and therefore employers
19:15who cared about both of those pieces a
19:17lot they can go and sell their employers
19:19on the engaging product piece and that's
19:21why health care self-insured employers
19:23were often considered early adopters
19:25also you often had very forward-thinking
19:28benefits people at these self-insured
19:30employers for example Sean Leavitt of
19:32Comcast who's been you know long
19:34considered a thought leader in the
19:35benefits space who will work with
19:37startups early on and you know this
19:39should be those benefits to the
19:40employees now unfortunately that
19:42approach got a little too popular and
19:44those channels got a bit too full and
19:47benefits leaders were getting pitched by
19:48these vendors all the time so at this
19:50point there's really a few large digital
19:53health estores in the space that
19:54being able deploy those employers and
19:57the next step now is for them to
19:59actually start working directly with
20:00plans and distributing to those plans
20:03because you get obviously much much
20:05wider distribution fortunately many of
20:07the administrators at plans and also
20:09much more sophisticated about this space
20:10is just a matter of time and they're
20:12working directly with startups to do and
20:13this is something we've seen through a
20:14mod as well so once you've gone through
20:17the early adopters on a self-insured
20:18employer side you then to get to the
20:20mass market need to go to the regular
20:22insurance plans yeah yes it reminds me
20:25of the saying that you know a federal
20:27versus state that the then states are
20:30the laboratories of democracy yeah
20:32before it actually gets up to the
20:33federal level and I think that's
20:35probably true in this case and that's
20:37actually a good point because that once
20:39you go through plans actually the next
20:41step oftentimes is working with CMS and
20:44going through Medicare and Medicaid now
20:45people can change those steps up but
20:47that's a progression that we've seen
20:48yeah so it sounds like you know a lot of
20:51times they go from the insurance plans
20:53to the government to CMS to Medicare
20:56Medicaid we've also seen a lot of
20:57developments and other part of the
20:59government namely the FDA you know there
21:01have been a lot of new regulations out
21:03the first FDA approved digital
21:05therapeutic and in you programs for pre
21:08can you talk about how these might have
21:10affected some of the companies and
21:11investments that we might make in space
21:13well I think from a regulatory
21:15standpoint we're gonna see across the
21:17entire spectrum incredible advances in
21:20terms of what's happening at the FDA
21:21level will see more generics approved
21:23this year than in any year in history
21:25previously we saw this year the first
21:28cell engineered cell therapy approved
21:30and cartee historically you know FDA
21:33risk or regulatory risk was really used
21:35as this barrier and this essentially
21:38fear factor but I think as the therapies
21:41become more powerful as we target things
21:43where we understand fundamentally what's
21:46driving the disease biology and
21:47therefore understand fundamentally how
21:49best to intervene I think what becomes
21:51pretty clear that regulatory risk in
21:53many ways was just a euphemism for
21:55scientific risk for experimental risk
21:58gene therapy an area that historically
22:01has been considered to be very risky has
22:03been shown to be so effective in this
22:05form of treating in heritable form of
22:07- that that passed the fda panel
22:10recommendation unanimously a 13 to zero
22:12vote Carty similarly also passed the FDA
22:16panel at a 13 to zero vote I mean those
22:19are two things that two modalities to
22:21therapeutic modalities two treatment
22:23modalities that were considered to be
22:25very very risky but they were shown to
22:26be so effective that that sort of risk
22:28goes away and I think when we see a lot
22:31of these therapies emerging the benefit
22:33the effect is so powerful that it's very
22:36clear that the therapies work and
22:38therefore the regulatory risk is low so
22:40traditional therapeutics companies gone
22:42to market through a regulated process of
22:45clinical trials and they've kind of
22:46looked the same way for quite a long
22:48time as we see this a new type of buyout
22:50company will those look any different
22:51there are emerging technologies that are
22:54going to change the way we think about
22:56testing and ensuring that our drugs are
22:58safe and effective so just to take a
23:00step back in history or at least to
23:01where we are today even the way we
23:03ensure a drug is safe and effective is
23:05first we determine whether or not we
23:07think a drug might work and the way we
23:10do that is by testing it in animal
23:13models and cells in the petri dish and
23:15once we convince herself that a drug
23:17might be safe and might be effective in
23:19those models we then go to human
23:22clinical trials where we of course need
23:24human volunteers to confirm that a drug
23:26is in fact safe and effective one of the
23:28big challenges that we've seen
23:30historically is that that paradigm is
23:32not particularly effective we still have
23:34massive failure rates and clinical
23:36trials we still have massive failure
23:37rates and phase 3 clinical trials at
23:39which point hundreds of millions of
23:41dollars and years and years of years of
23:43investment have already been made and a
23:45part of that is because there's so much
23:46science risk in some of the therapies
23:48that have been developed historically
23:49but a part of it is also that the
23:52preclinical models using animal models
23:54using cells and petri dishes is not a
23:56particularly effective way to predict if
23:58something's going to work in a human
23:59being and one of the exciting things
24:02that we're seeing emerging at this
24:03intersection of biology of engineering
24:05is new approaches to essentially obviate
24:09the need to use animal models as a
24:11replicate as an avatar for what might
24:13work in a human and we're seeing the
24:15development of organs on chips or
24:17eventually humans on chips that will
24:19allow us to really test something
24:21more human-like engineered system that
24:23we hope and believe can be much more
24:25predictive of what in fact will happen
24:27in a human being so I think that part of
24:29the paradigm were starting to see shift
24:31the other part of the paradigm were
24:33starting to see shift is that
24:34historically clinical trials have been
24:36very large because you needed to recruit
24:39a lot of patients to determine whether
24:41or not a drug would work with more
24:43targeted drugs we need less patients
24:46because the signal-to-noise is likely
24:47going to be higher the second thing
24:50that's been a big challenge in clinical
24:51trials is that they take a very long
24:52time to find the patients to recruit
24:54them to get the center set up and this
24:56is an area where we get a major assist
24:58from technology you know a great data
25:00point that I heard recently was a
25:02partnership that Facebook made with the
25:04Michael J Fox foundation to find ways to
25:07pilot the use of social networks to
25:10recruit patients for Parkinson's disease
25:12trial using social media they were able
25:15to reduce the cost and the time
25:17associated with recruiting patients by
25:19like 96% that's transformative and so we
25:22start to see compression how we ensure
25:24that drugs are safe and effective and
25:26they start to become more predictive
25:27we're gonna see failure rates go down
25:29and we're gonna see our ability to
25:30innovate in therapies accelerate
25:32dramatically so traditionally we've seen
25:34tech investing go for companies where
25:36the technical risk was quite low but the
25:39market risks quite high they're
25:40investing in growth and they see profits
25:42over time biotech investors looking at
25:45places where a science risk is high but
25:47the market is known if you make that
25:49drug it's going to sell if we see this
25:51new type of company how do tech and
25:53biotech investors come together and work
25:56together or you know work separately on
25:58these kind of companies you know we're
26:00we live is really at the intersection
26:02between the two and so obviously
26:04there'll be the need for connecting on
26:06the deep technical domain of whether it
26:08be computer science or machine learning
26:10as well as on the biology side
26:12collaborations are cases where there may
26:14be as some major biological advance
26:16that's just into the engineering area
26:17now these become tech companies or the
26:20biological advance comes hand-in-hand
26:21with a machine learning advanced and
26:23they synergistically be able to push
26:25things forward what we're really seeing
26:26emerging our new founders that also live
26:29in both of these worlds founders that
26:31have deep experience of the biology and
26:33deep experienced in computers
26:34and actually for these types of
26:36companies I think that's going to be
26:37critically important you know you can
26:38have companies where each of this
26:40expertise are in two people but you know
26:42that only works if these two people or
26:44basically can read each other's minds
26:45and can be telepathic having in one
26:47person really changed at the game and
26:49but I think that's going to be one of
26:51the major trends going forth okay let's
26:52do a lightning round of new technologies
26:54on the frontier what's the coolest new
26:56technology that you think will change
26:58biology in the future you know we spoke
27:00so much about engineering biology I
27:02think what we're going to start to see
27:03is the literal engineering of biological
27:06circuits inside cells the ability to
27:08design circuits much like we design
27:10electronic circuits and the reason why
27:12this is important is that there's only
27:13so much you can do by hand and by having
27:15essentially all the lessons from the
27:18electronic design automation ad a tool
27:21flow on the electronic circuit side
27:23applied to biology could really unleash
27:26the same type of effect that we've seen
27:28in electronics over the last 50 years
27:30but now in biology yeah and I think just
27:32to play off that I think one of the most
27:33fascinating things that we've seen here
27:35is how biology is no longer an industry
27:38in and of itself it's in something
27:40that's going to touch every single
27:41industry and we're already seeing
27:42applications in energy and textiles in
27:45food of course in health and in data
27:47storage and so I think computation
27:49itself is at some point going to be
27:50affected by biology and I think it's a
27:52wonderful time to be an entrepreneur and
27:54an innovator in this space malinka
27:56what's the craziest thing can imagine
27:58EHRs that actually talk to each other
28:03between epic and Cerner dream big I also
28:07think we're gonna see a lot of
28:08interesting things in the
28:09non-therapeutic CRISPR space so in a
28:11therapeutic CRISPR side they're a bunch
28:13of public companies that are going
28:14directly after using CRISPR to cure
28:16disease I think it's just the beginning
28:18for non-therapeutic CRISPR that is using
28:21CRISPR for areas like Diagnostics
28:23to find out infections using CRISPR to
28:26discover new drugs as a platform for
28:29discovery rather than the direct
28:30treatment of disease itself and I think
28:32we're just starting there thanks so much
28:34for joining the a 16z podcast this is
28:37the a16 Z bio team signing off