a16z Podcast - a16z Podcast: Shifting Risk Mindsets, From Tech to Bio

Episode Date: May 10, 2018

What challenges do first-time founders or tech founders encounter when building companies in the bio space, and how do they differ from traditional tech companies? In this hallway-style conversation episode of the a16z Podcast (originally recorded as a video), a16z bio team general partners Vijay Pande and Jorge Conde, with Jeff Low discuss the mindset shifts involved in building bio (particularly therapeutics) companies. They cover everything from different paths to market and different partnerships (including pharma) to different timelines and milestones for validating the product and business itself. But how do we get to a common language that bridges the worlds of tech and bio?

Welcome to the A16Z podcast. Today we're having another of our hallway style conversations based on videos that are also available on our YouTube channel at YouTube.com slash C slash A16Z videos. How can bio and therapeutics founders learn to speak the languages of tech and bio? In this hallway style conversation, A16Z bio team partners, including general partners Jorge Condé and Vijay Pondi with Jeff Lowe, discussed the mindset shifts involved in building biocompsies, therapeutics companies in particular. They cover paths to market, timelines, and more. So welcome.

Here today we have the A16Z bioteam, VJ Pande, Jorge Condé, and Jeffrey Lowe joining. And what we want to talk about today is some of the challenges for entrepreneurs investing and building companies in the biospace broadly. And especially in ways that that might differ a bit from the tech. the tech companies, traditional tech companies, and certainly I want to make sure we focus on many of the common pitfalls that we see entrepreneurs stumble into in this space. So with that, Jeffrey, you spent a lot of time thinking about that. In your mind, what do you think is the top issue that an entrepreneur in this space faces

when they start to think about innovating and building a company in the biospace? I think it's about speaking a new language. So you have tech entrepreneurs and technical founders, coming in and speaking a language that is very geared toward a tech audience, having a new technology. And I think what we're so interested in is bringing these new technologies to the biospace. And how do you translate that to your partners? How do you translate that to your potential funders? How do you translate that to the media? Because there's something new here and something new and different is something that they often have difficulty translating

to people doing something used to doing something in the same way. It's previously. Yeah, or even, I mean, some of these founders, while they're in computer science or engineering, they've actually never founded a company before. So they're not even familiar with how the tech side works. Either way, there's tons of pitfalls going into the biospace. And what do you think is different about this class of companies that makes it interesting in the first place? Yeah, I mean, first off is that often these companies have to prove that their technology works and is useful. And the elements necessary for proof are relatively high bar, high barrier to pass. And so, you know, you could get early proof of concept deals, and that will look good.

And you might have like five deals, and it'll be like a couple hundred K each. But in reality, those are so easy to get that maybe it sounds more impressive than it is. Getting really big upfronts and obviously big biobuck like deals, that's a whole different game. And what do you think is different this time? Computational techniques and in drug discovery have been around for quite a long time. You know, what's exciting about what's going on now and what's different this time? Yeah, you know, you think about it is that at people point. to AI and machine learning,

but really that's, I think, just a surrogate for the fact that data is being used very efficiently, and then we have so much more data, and we do have techniques that we didn't exist before, but in many ways it's the data that makes things different and that we see pharma companies and biotechs and startups

taking themselves not as biotex, but as eventually, I think, as data science companies. And that's a very huge sort of mindset shift. Yeah, and I think on that point it's a really interesting moment in time, because I think one of the big challenges that startups have had in the space traditionally is that if the time to get to a validation point or a value inflection point more specifically is very long and risky and time consuming, you do have this temptation to start with the pilot. And I think trying to get a bunch of pilot deals has multiple pitfalls associated with it.

First of all, they tend to be relatively small up fronts. And so you have to generate a lot of them to actually generate enough additional cash. runway to drive you forward. But second of all, most pilot projects come with expectations associated with them. And so, you know, I don't think it's unique to the biospace, but this concept of death by a thousand pilots is a real one. And so if you're trying to find a company in this space, and you're trying to build a company in this space, I should say, you have to be very careful about who you decide to work with on the pilot's sense, on the pilot side. Because not only do you run the risk of working on a lot of small projects, small projects with

large companies tend to have scope creep and timeline creep. And so if you thought you would finish pilot project and get paid X in Y amount of time, oftentimes you're getting paid X in something like two X, two times the Y amount of time, and you're also working a lot harder if there's been scope creep. So it gets really, really challenging to deliver on multiple pilot projects. in one given time frame. So what's the solution to that pitfall? I mean, do you just not do the pilot projects and wait for the bigger deals?

You try to execute them differently? Well, one thing I noticed is I think this is where a lot of these companies start is. They start with pilots and they start as a service company. They start by saying, hey, I'm going to go to Big Pharma. I have a new technological breakthrough. I'm going to sell that as a pilot as a service to Big Pharma companies. And then they can't generate the economics that it would make for a sustainable business. And then they're really in a sort of tough spot.

And what ultimately, you know, they go through this idea maze and, you know, what they decide to do often is, hey, I need to develop assets in-house. That is because in the biospace, value is created in these huge technical milestones in the very beginning where a lot of these services are used. These are pre-clinical assets in which there's really not too much value until you bring a drug candidate into the clinic. So these companies go and say If I want to capture the value for my technology I need to advance this asset a little bit further down And then they get into the space We say, well, I now am developing drugs.

Yeah, and I think So I think that's exactly right To go back to VJ's comment or question rather And how do you address this issue of the death by a thousand pilots? The first one is you have to be very selective with who you work with. In other words,

it has to be with and this could be a different company in any given in different context but it has to be a company that fundamentally believes in what you're doing and and that if you can get to a proof point for them you are immensely valuable for them the big risk is that everyone does the you know I'll just do you know the free sampling yeah I'll try a little bit of everything yeah that's always fun and that's yeah it's fun of your sampling it's hard of you have to make all the samples right and so I think that's the the big challenge. You have to be very careful with who you, you know, you align with on, I think Jeffrey's point, which I think is exactly right, is you also have to be very thoughtful

about where you are in the value chain. And so, you know, there were a lot of bio companies, for example, that came around around really, really neat science, but that the end for the science was to, for example, identify a novel target. And in this case, we're talking about a novel target we're talking about the piece of biology that you want a drug to um to intersect with to impact the the progression of the disease and so the problem with that is pharma companies are swimming in targets right the problem they're trying to solve for is how do i fill in my pipeline yeah often they think targets should be free anyways exactly and then there are a lot of competitive efforts to make targets free and basically win on on the therapy on the drug development process

piece and so if your platform is to identify novel targets it's going to actually be very hard If that's sort of the end of the line for what your technology can do, it's really hard to capture value there and certainly hard to get that from a collaboration standpoint. If on the other hand you somehow had a magical tool that could predict which drug was likely to fail in a phase three trial, you can imagine that you'd be able to capture a lot of value there. So it really is where you are in the value chain. Another pitfall I see is that you, and I see those many different types of technologies.

It could be, you know, machine learning or compute, or it could be like a new type of structural biology technique, whatever the technology is. They got the cool technology. They think this will change drug design. And so then they go to pharma and they try to sell it. And pharma is not convinced yet. It's like my kids with new foods and they don't want to try it until they really know that they really like it. And then they love it and it goes, but the beginning is really hard. And so then you decide, okay, we can't try it. We can't sell it to pharma for what we think it's worth. So we're going to design drugs. But we don't know anything about designing drugs. And then that can also be a very dangerous road. So I don't know how you deal with that pitfall. Do you just sort of try to

stay away from that from the beginning and honestly never crossed that line? Or do you plan to do that from the beginning? Yeah. Well, I, you know, Jeffrey alluded to this a few minutes back. I think if you look at the history of sort of the biotech startup ecosystem, you know, a relatively common journey is that is, say, I have this interesting technology or platform. I'm going to do a couple of early stage business development deals and then my grand vision and hope is that eventually so I can develop drugs for them for my partners and capture some piece of the economics of a downstream value. But my ultimate vision and goal for my company is that I'll eventually get into developing my own drugs and where I'll keep, you know, all of the economics

or the majority of the economics. And it's a, that's a really hard sort of transition to make. And frankly, not many companies have done. And the teams aren't set up for it. That's right. And they aren't set up for it. So I would say there's two things here. One is going back to the pilot concept that who you partner with matters, what they want, how they value and how they see what value you bring to the table matters, which I think leads to a second point, which is business development in the biospace is a fundamentally strategic advantage to have. If you have a team that is good at structuring business development deals, in other words, that has experience in the space, that allows you to actually help bridge what is often a fatal chasm for life. of companies how to figure out how to set up early stage deals with pharma companies where um you're getting value from the larger partner because the larger partner obviously can deliver a lot of value how you're figuring out ways that you can capture some of the downstream value if your approach works

right to this you know downstream milestones and royalties and all of those things and importantly how to make sure that you structure these things in a way that you haven't fully encumbered your platform so that you can in fact do things on your own in the future if you choose to or do things with other partners in the future if it makes sense. Right. I mean, well, you know, the pitfalls that the platform sometimes is a thing that's, we find very exciting. But then the company sort of has finally actually get through all the things we talked

about. And now they got to deal with the fact that they were successful and with this one asset. And now maybe they're tempted to ditch the platform. And I could see the value and sort of just pushing a lot with what works. But, you know, how do we sort of balance that? Especially nowadays where the platform really could be really valuable. if these technologies are as powerful as we think they could be. Well, I've certainly seen that in close, the first-hand experience,

where you have a platform that could be very valuable, and precisely because it's a productive platform, you find an asset, in this case, a specific potential drug, and very quickly all focus goes towards, you know, how we make sure you maximize the value, minimize the risk in successfully commercializing said drug. And it's a really interesting thing, And it happens almost overnight, that as soon as you have that, all of the conversations start to focus on, well, we could always use more resources on making sure that the drug program succeeds.

And there's only a fixed pool of resources, generally speaking. And so what ends up happening is the platform gets started. And I think that happens time and time again. I think one way to address that is, of course, through business development partnerships. And even then, you don't entirely alleviate the pressure on funding the program ahead of the platform. unless you've specifically structured your business development agreement to fully fund the program, right? I mean, that's the, and that's hard to do, especially if you're an early stage company. But a second way to do that is through innovative structures.

And so we've seen this in a couple of examples where companies establish an LLC structure. And they basically are able to say for investors that are really the believe in the future of this platform and the productivity of the platform to not just develop one drug asset, but multiple drug assets over time, that becomes sort of the parent company of the mothership and then you have separate LLCs under that that you basically say for drug asset A I'll bring in new investors and new investors are essentially betting

primarily on the success of that drug asset and so the parent company is taking a much smaller piece of the economics but they can now replicate it many many times because they have the resources to invest in the platform so ensure that future products let's be clear why we do this a unvalidated platform

really doesn't have a lot of value. And that's why there's on the first asset, you know, the whole company's value may be riding on this asset because not only is that asset in itself valuable, but it also validates the efficacy and usefulness of the platform itself. Then you're kind of in this state where you might have to say, well, I am now a single asset bet because of my first asset from my platform fails. Well, maybe I throw out a good platform with a bad asset, or maybe I just have a bad platform, you know, and a bad asset. Some of these other legal structures are kind of ways to get around that. But it doesn't get around this problem of, hey, the first asset validates the platform, and its success or failure means a lot for

this business going forward. Yeah, and we've spoken a little bit about this before, and I think it's an important point to bring up, is that there are platforms that generate a drug asset, and therefore you don't know that they're unvalidated by definition. because you have one drug acid, so until that drug asset is approved, you don't really know if the platform was valuable or not. And so therefore, all the value does accrue and all the risks does accrue to that lead asset. But then there are platforms that could be so fundamental in understanding biology and so generalizable that, you know, that you actually want to structure a deal in such a way that someone that only believes in assets will fund that and you still leave room for folks that believe in the platform to support that. Now, it's hard to know a priori, which is which, and I think that's one of the big challenges here.

Our belief, or at least I think I speak collectively for the team, is that, you know, for platforms that have an engineering-like bent to them, they're more likely to fall in the second camp than the first, but that's obviously the hope and the bet that we're making and the entrepreneurs and the companies that we're supporting. But yeah, I think that's, it remains to be seen how over time you make sure that you're very clear on what kind of platform you're dealing with when you're making an... We've been talking to some generalities here. You know, Vijay, you're involved pretty early on in Schrodinger, which, you know, had a very

successful collaboration with NIMBUS. You know, maybe you could talk about that example as one where the legal structure was, you know, really successful, made, helped make that company successful. Yeah, I mean, Nimbus was, I think, one of the early companies that thought about this LLC structure. And it's kind of interesting that pitfalls can even be just in how you structure the company. You would think SICOR would be a pretty standard thing to do these days. And so, I mean, I think it's still early and I think it's a new thing.

And therefore, the venture community and start bunch ofmers will need to get socialized to this and that will happen in time. You know, I think we've been spending a lot of time talking about therapeutics. I think, you know, there's also analogous pitfalls in other areas, like in diagnostics. And, you know, especially in the biofund, we're interested in biology quite broadly and other sort of applications. When I think about that, I mean, there's some applications that might look more like tech companies, depending on how they're built. but some things like diagnostics will still have to get through regulatory agencies, either Cleo or FDA, and there's a whole bunch of pitfalls just there. And I think a lot of times what we're seeing is that it's very important to address those

pitfalls by understanding them early and handling them. So, I mean, we could talk a little bit about the pitfalls that we see in that diagnostic space. So why don't we start there? What do you think is the biggest risk in diagnostic specifically? Yeah, ironically, I think, you know, most people think about the FDA or Clea being your big concern. I think reimbursement is probably the first place to start because I wouldn't want to sort of be designing a test without having a confidence that I'll get reimbursed. Otherwise, why bother doing the whole thing? And so, you know, when I've been involved in drug design, we usually

start with thinking about how we're going to run the clinical trial and then work backwards. On the drug design side, I would want to do the same thing on the diagnostic, but start even further downstream, downstream talking about how we're going to get reimbursed, what's the value that we're going to add? And if that's there, then the rest actually we can make arguments for. And when there's time early on, especially when tech is so powerful that you could go of lots of different avenues, and it's early, it's sort of directing where you want to take the ship. Thinking about that will get you to the right place, long term, rather than getting to the wrong places than having to figure out how to pivot from there. Maybe it would be helpful if you backed up and said, well, how is, what would a company that's a diagnostics company that's tech-driven or built like a tech company look like, and how would that be different from a biology-driven diagnostics company? Yeah, I mean, that's a great point.

I think one of the things that's, we see that's really different is that when you can engineer biology, you don't have this very bespoke process of sort of having scientific discovery, but you have the ability to engineer a process and then repeat this in different indications. So if you're having a cancer test, it could be for whichever cancer, you know, a indication you want based on the data that you have. So why not pick an indication where you feel like the go-to-market is strongest? And for that, it's now thinking about the go-to-market at the very earliest of stages. So just to take your words out of context a little bit, earliest of stages in

diagnostics, and you're talking about go-to-market with reimbursement, one of the challenges, and I think this is more a thought experiment than sort of hard data, but one of the challenges that at least I've heard about the early-stage screening type diagnostics and reimbursement is that a lot of patients don't stay on a plan for a long period of time. and so getting broad-based reimbursement for early screening may not make sense from an ROI perspective for the payer. How do we get around that question? Yeah, this is one of the fundamental questions in healthcare because how do you pay for things where the ROI is backloaded like five or ten years? And so if your insurance company, maybe that doesn't make sense for your economics, there's a couple different ways.

One is that, I mean, outside the U.S., obviously things are different. So that's like the cheating answer that you don't get, they only get partial credit for. I think the, maybe a somewhat deeper answer is to think about that self-insured payer, sorry, some insured employers might be a little more motivated because while people may change plans, they change jobs slightly less frequently. But I think, you know, I think we've talked about this. We've seen other sort of more interesting financial mechanisms that are coming on board where insurance companies can find ways to make these incentives.

And I think we'll have to change how paying is done, but at least there's been proposals for that. Yeah, and I think that's, that was one of the interesting ones that I've sort of heard thrown around is the idea, you know, we talk a lot about doing pilot projects with, you know, pharma companies. If you're trying to develop a therapeutic, there is this sort of tantalizing, um, potential to do pilot projects with insurers or payers if you're trying to develop something that they need to prove the ROI out for themselves. And that gets really interesting because then you can start with the end in mind as you sort of laid out. Yeah. Great. Well, you know, so we covered therapeutics and diagnostics. Because maybe one last thing we could talk about and then we're maybe running close to time is sort of biology more broadly.

Like, so maybe you're not doing something that's going to sort of connect with human health such you'd have to get it done. But like you're, let's say a company designing bacteria to do something new. The bacteria is not going to be ingested by human or anything like that. Now there's sort of different challenges. I mean, there's things that may be tempting to look like a temp company because you may have tech behind it. But, and the deals could be really large. I think this maybe just takes us back full circle to what we talked about on the therapeutic side, that the POC still become a problem.

Yeah. And I think it's, I think those are big, I mean, that is a big challenge on the therapeutic side. It's a big challenge on sort of the broader platform, you know, biology side. And to us, or at least, you know, as we've talked about it as a group collectively, I think what's really interesting is what do you do with that challenge, right? So one way to do it is to say, now to go all the way back to the beginning of this conversation, is to say, well, maybe you do start off as a service model. Or, you know, in the case of, you know, I don't know what a box, a magic box that could make

engineer bacteria would look like, but let's assume that this is something you could productize as an instrument, kind of like what Illumina has done, where you sort of now go into this box and sort of consumables model. And what you do there is you basically start and you sell to the high end of the market first, right? So what we saw what happened with sequencing was that obviously in the very early day, days when the throughput of the sequencer was low where the cost was very high, the only natural buyer for that was, you know, large research institutions and those sort of became

the initial sites. And as the, you know, sort of performance got better, the market opened up and opened up and opened up. And of course, you know, today, if you can get a bench top sequencer for relatively low amount of money and basically any lab can afford that. And so I think the same is true when you have these other technology platforms. So the big question becomes how do you get to fundamental POC in the case of sequencing it was straightforward you could sequence known DNA and basically show that you can generate that same result in the case of sort of let's take your you know engineered bacteria um is you would want to design a simple set of experiments and i'll go back to your apollo mission sort of analogy right where you a set of

experiments that say i can design basic functionality it's going to have very high predictability and then over time, the level of complexity that you can design into a system goes up or the throughput goes up or the quality goes up or the cost comes down. And so there are sort of these various variables that folks, I think, feel comfortable around. You can start to sort of march towards, you know, increased proof points and eventually get into what, you know, POC. And the other thing I'd mention is POC is not the same for all parties. That's right.

Right. So you'll have early adopter POC is obviously a near term target than your late adopter POC. And so you should always design for the early adopter POC with a clear path to how you can actually eventually engineer in all the way to the end of the market. Now, the one thing I would say, I think this is true for all entrepreneurs in this space, and this has been my humble experience, is, you know, if you're developing technology in the biology space, you should know what your near-term killer experiment is. And the experiment that if it does not work, it's, it's, you know, kill mode for whatever you're developing. And you should know what that is, you should have in

your mind, what could I show in the next six to 12 months that would essentially cause me to kill this idea? So, A, you should know what it is. And B, you should be doing that experiment. The number of times I've talked to entrepreneurs. And that's scary because it's a very existential thing, but that's exactly why you've got to do it. Exactly. And the number of times that I've talked to entrepreneurs that have said, well, yes, we think that's a big experiment, but we have all this other stuff going on. So we're going to get to it. And that's, I think, you know, for everyone's sake, it's important to do the kill, you know, the kill test early. Yeah, yeah.

So that's always been my view on. There's POC and then where there's the equivalent of proof to, you know, it's proof of concept and then, you know, the proof to me that this hasn't failed, you know, in a fundamental mode. And so you should know what that looks like and do it early. So maybe one last pitfall we could talk about is just pitfall for getting yourself funded. So if you think about like three poles, we've got consumer enterprise and biotech. And the space that we're talking about sometimes looks more or it's like some combination of those.

You might have a company that's using technology and health care but direct to consumer, something that's more enterprise focus, something that might be more therapeutic or diagnostic focused. And each one of those has fairly different ways of proving to investors that you're making progress. You know, consumer, maybe it's about the graph, you know, DAU versus MAU and so on. For enterprise, it might be revenue. For biotech might be hitting milestones. And so a lot of the challenges here, I think, is, you know, trying to figure out where, you know, the pitfalls are figuring out, How can you prove that it's really working when it's some complicated mix of all these things?

And then which is the best investor, actually, at each stage of the business? So, I mean, there could be periods of time where it's, you know, much more tech-like building a platform. And there are periods of time where there might be a lot of science risk. And, you know, tech investors, I mean, ultimately, they're just not set up to take on science risk in the way that biotech investors, you know, have arranged themselves. So, you know, how do these companies go and pick which thing it is that they should be doing it and who they should get money? money from? Yeah, I would say a couple of things. One is, you know, in some cases it might make sense to actually have a hybrid, right?

Because these companies are hybrid. I could think of one firm like that. Yeah, exactly. Like a one firm who spends their time thinking at both levels, right, from the bio side and from the sort of the more tech side. And it also makes sense to think about how you would build up your investor syndicate, right? And so I think one thing that's a truism for all entrepreneurs is when you're raising this round, you should always be thinking about what the next round is going to look like.

And that's not just in terms of what, you know, valuation metrics you hope to hit, what milestones you hope to clear to make sure you hit the inflection points to allow you to hit the valuation metrics, but also who's involved today and who might be involved tomorrow. And, you know, one thing that we've seen work before is, you know, you see a traditional bio investor, connect with traditional tech investor. So they can, each side can help the other know what they don't know and sort of get comfortable. and then as the sort of company gets more mature, that mix of the syndicate might change.

Either becomes more traditional bio or more traditional tech. And then eventually you get to, of course, late stage type investors. So that's one way to approach it. The other one is to go back to where you started, Vijay, which is to be very clear on what metrics do you think you can prove

over the course of whatever the funding period is, the runway, and then base your investors on what you're going to prove. Because, again, different audiences, care about different POCs, whether it's the graph or the milestone or what have you. And I think being very clear to yourself about that up front should help you define how you seek an investor base. And then the final point I would make is I think the initial comment that Jeffrey made,

which is sometimes we speak two different languages. And so, you know, you need to be, if you're not fluent in two languages, you should definitely be fluent in one and, you know, functional in the other. And I think it's important to understand what the two languages are so that you can ensure that if you intend to bring bioinvestors online in the future, that you've set up a company and set up a sort of a common set of languages and ideas that they can understand.