Simulating Life & Accelerating Discovery: Deep Origin’s Biotech Vision | Michael Antonov (Part 4/4)

Intro - 00:00:06: Welcome to The Biotech Startups Podcast by Excedr. Join us as we speak with first-time founders, serial entrepreneurs, and experienced investors about the challenges and triumphs of running a biotech startup, from pre-seed to IPO, with your host, Jon Chee. In our last episode, Michael Antonov shared the early vision behind Formic Labs, how it evolved into Deep Origin, and why building better biological simulations could unlock new frontiers in drug discovery. If you missed it, be sure to check out part three. In Part Four, Michael dives deeper into Deep Origin's current product suite, how they're helping biotech and Pharma teams accelerate R&D, and why making advanced tools more accessible is central to their mission. He also shares his philosophy on company building, how his past experiences have shaped Deep Origin's culture, and what's next as they continue scaling their impact. 

Jon - 00:01:14: And so, you know, you're starting to invest out of Formic. And I know, you know, can you talk a little bit about Formic Labs and like eventually Deep Origin? Like how did that come to be?  

Gemma - 00:01:24: So Formic VC was the kind of name of our fund. And we started in basically 2019 and operated for a few years. We're still like legal entity exists and I'll still occasionally invest. But now I'm very just focused on the company on Deep Origin. So Formic Labs kind of came out with this idea of how do we build simulations for life? So as I was starting to invest, I looked at all these tools. And it seemed to me, I actually wrote, I think, one blog about this. And I wanted to keep like a whole bunch, but it ended up just being one. But around how you can deconstruct biology piece by piece to actually understand it, which is deconstruction in our approach. I mean, arguably that's simplistic because there's so many unknowns around. But, you know, we do have instruments which look like at every level. But how do you build a more holistic model? And to me, it felt like if you're really going to understand biology, we want to be able to represent it in the computer. So in order to understand something, you want to be able to simulate it. If you can't model it in some kind of way, then you can't really predict the behavior. You should have guessed it. Or you may be predicting a little part of a much bigger, more complex system. So our brains like... A good example I like to give is we have like all these, you know, say 10,000 genes which might be expressed in the cell. And clearly different cells have different pools. But just to pick on one cell, there's all these pathways and ways they interact. We can know some feedback cycles and loops. But unless it's in a computer, you cannot actually in your head understand these interactions. So if we have a chance, it has to be digital. And then, of course, you want to be able to reconstruct that. So to me, I felt like the industry, and it's getting a little bit more there now, and it kind of is not there. But especially five years ago, that there's not enough infrastructure to build these kind of models. There's kind of more special case pipelines for inquiry for a particular thing. And I think genomics is probably the most developed one. And then there's a range of like, epigenetics and other things, single-cell stuff, which became more developed. Holistically, there's still gaps. And you can't really bring these multiomic things to fully reconstruct like a cell or a part of tissue. So the early idea of Formic Labs was that if we were to set a task of can we simulate a cell in such a way that it would actually have... Predictive phenotypic drug discovery value and do it on many levels of complexity from say atomic to the going up to like protein structure organelle and then you have the concentration there it's a built multi-level model which could then represent a real cell of maybe like an immune cell or something you and have that, I would actually respond computationally and in structure in the same way, like as a real thing would. So that was sort of the high level idea for Formic Labs. And I reached out to a number of people. One was actually a professor. Well, someone who did full cell models and pathways. In Stanford, Jonathan Karr, and he was a professor at Mount Sinai. So he decided to join as an early employee and we worked together for a while. And then Garegin Papoian, who did fourth grade modeling and versions of physics, chemistry, dynamics, all his life, professor in the University of Maryland, who actually started his own company, BioSim. So I met with him and a number of other people trying, how can I put together a team which can do this? And I think we certainly were under-resourced for the ambitions we had, but I also didn't have a full sense of depth and complexity. And it's sort of, as I thought about it, I realized there's really two kind of domain areas where there's shortage of kind of tools. One is actual simulations that tend to be not connected. We have, let's say, molecular dynamics. There's a very particular drug discovery role that it plays. There is maybe pathway simulation, which might play a role for target discovery. There's very particular tools like docking and screening for very small molecules, particular tools for antibodies. But to be fair, there's really no good antibody tool out there yet. So there's like various pieces in software package, but it doesn't come to the whole, like you can represent biological model and they feed into each other. It just doesn't exist. And so I felt like, okay, that needs to exist. But then there is this whole second part of real instrumental. Data processing of real lab data, because you have the instruments and the infrastructure, but the way the data is collected, there's not always standard databases or ways to arrange or integrate different multi-omic data into something which is, you know, actually representative of the biological entity you study. So somehow all of these different experimental modalities of inputs need to be brought together. So that means you want to have a kind of a data software data stack. So that sort of got shaped into the idea of making more general R&D OS, which on one hand can work with all this lab instrument data and put it into some kind of biologically relevant digital representation and use that to actually feed and calibrate the simulation stack. So that was sort of the meta structure idea of Formic Labs. Admittedly, as you imagine, this actually spans its own sub industries. And there's like 100 companies in this broad domain. So it's sort of maybe naive to think you could do something like that unless you have like, I don't know, $5 billion to spend on a project. But that, that was a scope of things we were you you exploring, and trying to build toward imagining a unified system. So that was a vision of Formic Labs as we started. And then we started building tools and integrating things. And ultimately, it evolved into more what is Deep Origin now. And we ended up combining with BioSim and really going on this more complete drug discovery simulation approach to where we are heading now.  

Jon - 00:08:33: Very cool. And there's a couple, like a bunch of different directions I want to go, but the two things that just stood out to me immediately... It was like your conversation with your father about the complexity of a cell. And it's like full circle. And you're like, I'm going to do computer things. Whatever, dad. But now you're like marrying the two. And I love when that happens. You're like, okay, we can use the computer to kind of like make sense of the complexity within the cell dad. Other part that stood out to me, it was kind of, as you kind of described the Formic Labs kind of inception, it kind of reminds me of like Scaleform where you're like, how hard can it be? Whatever. Like we're just going to, we're just going to like jump straight into it. And I sometimes think that is, that is like a prerequisite for like really impactful stuff. Because if you knew, it's kind of like the Jensen Huang thing. Like if he knew how hard it was to start NVIDIA, you probably wouldn't have done it. He just, you just have to kind of go and kind of leap of faith it or else you're just going to get like discouraged. You're like, holy shit. Like that is a lot. Formic Labs was kind of like this, kind of the dream. And then Deep Origin is like actually starting to execute on this and starting to like pull it all together. Can you talk a little bit about the one, you kind of alluded to the early teammates that you kind of brought into Deep Origin. Can you talk a little bit about those like early days of Deep Origin and also maybe just set the table for us on what is the status quo and how does Deep Origin? Disrupt the status quo.  

Gemma - 00:10:12: So we covered a lot of space, like in just this broad vision. And there is kind of the ambition of wanting to have this kind of platform to enable people and also use really good engineering principles. So it's like pluginable and accessible and open. I will say there's, I observe there is a number of challenges in this space and they come from two different sides. There is some I would say is more on tool ecosystem side. And then there is challenges more on the simulation drug discovery side, which I'll talk more about. So maybe I'll mention the ecosystem first because I care about that a lot too, and we're doing work there too. And even though that's not like the main direction of this moment. But I think if you look at about the scientific tool ecosystem and software, I mean, you get like the genomics tools, you have various like packages, often it's software, which sometimes it can run in a compute cluster, but often is it like Python scripts, academics really in the university it releases analysis software. So that's a kind of a common thing and others pick it up. Um, the, and then there is devices usually which are expensive. So like a sequencer or like, hi, Michael, you're more familiar with it than I am. And the challenge with that is, on some level, it's not as accessible as one would like. So part of the reason that I think the costs are high, and then people want to have consumables, which companies can make money, part of the reason is because there's not enough scale. So if you have a PC, you can sell it to millions of people, but millions of people don't yet need a DNA sequencer or a microscope. So you're limited to your number of people, and if you're limited, you have to raise the price. But if you raise a price, it's harder for someone like a small startup to have accessibility to it. So that, I'm sure, speaks to what you do. So that is itself a barrier. And then technologically... There is, a number of these packages which work with it on the software side, they're usually released by research group. And sometimes PAGs move on. So a tool is not very well maintained. So when others come pick up, they may have bugs. So there's a handful of popular tools which are mal-maintained or may be done by the industry. And then there is a universe of things which are not very easy to reproduce, pick up, or reuse. Because it's not published, it's not documented in such a way that somebody can pick it up. So that's a challenge. And then the Python ecosystem, for example, is actually not very robust from versioning an API perspective. Then if you're going to run it, you have to have a certain environment. You have to install certain packages. You have to have certain compute. So maybe you don't have a compute because your laptop or computer is not big enough. Maybe you're on the wrong OS, right? Maybe you are, et cetera. Maybe you need a GPU. You don't have a GPU, right? So there's not a very easy universal way to write a scientific application. So it can run by anyone, anywhere. And it will keep running if somebody just accesses it. So that's part of a reason. And then there is a number of companies which are coming in, which are trying to solve this for businesses. And so you can look like there is good work, let's say, Benchling has done on infrastructure. There's a number of other startups. The challenge I find is whenever it comes in, somebody wants to build an ecosystem. Well, how do they bring customers? Well, they usually want to lock in people in their own cloud platform because that's how they're going to keep customers. So on one hand, they're solving and improving in there. And others, they're kind of trying to control it. So there's not like an open ecosystem where you can write an application which runs anywhere, which maybe you can even make money for, which does scientific stuff. That is part of a ecosystem challenge. Another challenge is that if you are in investing, and you're investing in biotech startups, It's actually hard to raise money for tools. So it's more, there's a typical way, if you have an idea for something and you have one therapeutic, if you have some platform, you build a platform to find a drug so you can sell a drug to make money. And investors, the majority of money will go into the drug, like the wet lab and clinical stages. So at some point, once you have one or two assets, which are going up into clinic and you raise that. $50 million, multi-hundred million dollar round, your investors don't really care about your technology anymore because they're betting on this one or two assets. And that's what's going to come. They may care a little bit about technology. I'm exaggerating. But once 90% of your capital is going into something, the other 10% is not going to be as... So therefore, you have these companies trying to make platforms, but they have no reason to come together and build a unified tool framework that works together. So, and if not enough, people who want to invest money in the tools because it's hard to sell. And we discussed that before. So those are the challenges, I think, in the ecosystem. And there's a question of how do you address that? And I think there's room for like a unified platform which will run scientific tools, which would open and doesn't lock you into the cloud, but can run on clouds. And I think there's a number of companies which do interesting stuff in this space, but it hasn't evolved yet. So we want to contribute to this space. But I also realized it's a very large surface area. And we were very interested in simulation. We talked about doing cells. We can't do it all. So we ended up working a lot more closely with small molecule drug discovery and simulations of like proteins and pathways. So kind of this lower level things. So in the end, we decided to focus our attention on tools that really simulate life on that early kind of biochemical. Our level and then integrate with enough infrastructure and tools to support it. To really help the early stage drug discovery and understanding of biology. So that's kind of we, I looked at it and how much resource it would take to work on several of these areas that we narrowed in, in kind of this area. And this area has its own set of challenges I can go into. But, if I was to step back a little bit. Based on all of this, the goal of Deep Origin, our mission is to really organize, model, and simulate biology. That's kind of the premise in order to enable understanding to make cures. That's really kind of the basic mission. And we are building tools around that understanding. So that's... Of a long-winded impetus of the areas we worked on and Now we focus on simulations.  

Jon - 00:17:56: Very cool. And, you know, I think exactly what you said. There's like, it reminded me of like all the way back to, you know, Kickstarter. It's like, you talk about the tooling. It's like you raise the price. Not that many people can afford it. So you don't actually get to scale. If you don't price it enough, the margins are too slim. And then you don't, you might not be around if you're just like, you know, hand them out. And I think I've always thought about that ecosystem problem too, that, man, it, what, how, how much more efficient we could all be if we were just all kind of like, had this like common foundational layer. Versus kind of these things that kind of like live in isolation. And then, like you said, that there's not like, there's sometimes there's not maintained. So you just have to like, you scrap it. You're like, you crap, we can't even use this. So it's like this big, it's a don't have the solution, you know, can't pretend to have the solution, but it's definitely something I've observed as well. But on the simulation side, you know, where you started to focus, can you talk more about Deep Origin's product? And you talked about a little bit like small molecules. Who are the folks that? You guys seek to work with and find the most value from your product suite. And tell us about the product suite.  

Gemma - 00:19:06: So basically, the vision of a company is to build these multiple levels of simulations from kind of like atomic interaction up into like protein structures and so on to enable drug discovery. So what we actually have today is a range of tools which cover things like docking, which is a way you get the molecule to plug into a protein to see if it's doxodont. We have our own version of molecular dynamics. We have... The something called EVP, free energy perturbation, which is a way you compute the energy and better understand how well does your molecule bind. Virtual screening around taking tens of billions of compounds electronically and seeing which ones might actually fit your drug. So we have a... Full suite of early stage drug discovery for small molecules. So that's one level of biology. We also have kind of these coarse-grained models around protein dynamics. So AlphaFold became very popular for protein folding. And the thing with protein folding, it kind of tells you one state, one likely state. And it's actually learned based on a lot of similar states. But often you may not know how does something get there. Or if you have a disordered region, there is things you don't know about movement of these, which makes it hard to drug. So that is a separate kind of domain. So we actually have a system, a model, which allows you to do that type of dynamics and actually understand the folding and how it interacts. So that could be useful for protein-protein interaction, antibody discovery, like a big focus on Protex, which is these different like protein binding drugs. So we have these coarse-grained models, and then we have pathway simulation models around target discovery. So that's a full tech stack. And in addition to that, we want to make it very accessible. So we have an AI system called Balto. So as a drug discovery user, you can come in and you can ask it, I want to learn about this protein. Can you download it? Which properties does it have? Does it have a pocket? Can you try to dock these things to a pocket? So it will do that to you in a few commands without having to install complex tools. So we are trying to both have world-leading tools around the type of technology I'm describing and the kind of infrastructure and data and convenience to actually make it easy and accessible. So that's really the core tool offering. So then the business model we provide is really two-pronged. We have SaaS, software as a service. So users can come in and sign up. We actually give free access to students or people who are trying it out to just try out Balto AI Assistant. People can come to our Deep Origin site and do it. And then we have kind of these high-end enterprise tools and virtual screening. And then we have partnerships. So a partnership model is when someone wants to have a target they want a drug, or maybe they screened it a bit, maybe they run a Del screen. But if they don't quite have the hits they want, we can run... Like 10 billion on like, 20- All eMolecules or Enamine or some other databases against your drug and be able to tell you the best hits in a way that we believe we're likely to be the best in the world. So that's our product suite. And then we have extra infrastructure compute tools to support the development process and kind of data management around that. So that's a full suite of our software. So we'll either help you if you have your target and we'll work with your company to actually give you answers or we'll give you software to... Actually where you can do it yourself. Or you can learn how to start playing with it and then decide to use it for a more complex project later.  

Jon - 00:23:26: That's Awesome. I was going to say, like, I'm using a different software as an example, but like usually, you know, there will be a bit like HubSpot, very SMB focused and almost kind of like a freemium. But and then, you know, you have Salesforce, which is like the enterprise, like really big. But what I'm kind of at least what I'm hearing is that you guys kind of like cover the gamut. You can be you can be a student, self-serve, test it out, give it a test run. You can also be an enterprise, right, where you can like, you can be a larger organization, like a large Pharma or, you know, in a medium size. And then you have like these partnerships, which is like we have a very specific goal and like target that we're like or one asset that we're trying to progress here. And you kind of like cover the whole spectrum, which is really cool, you know, really cool to see. Because like, again, it's like sometimes I feel like, you know, especially with tooling, you kind of like you're either in the enterprise or you're in, you know, SMB. But in this case, it's kind of like you're you're democratizing the tool and giving it access to everybody, which is really rad to see.  

Gemma - 00:24:31: I mean, there is as kind of some separation of capabilities there. So, if you think about the organizations we target, they could be a range, it could be biotech and Pharma. So we really don't discriminate too much based on size. At this point, it's probably easier for earlier biotechs to start with our tools. But also a lot of Pharma have expressed interest in Balto and just the ease of use for their scientists. So it's those relationships take a while to build. There's like trust and, you know, enterprise BD is non-trivial. Let's just put it there. It takes a bit of time. But like we have a very strong AI team around those tools. So it's quite smart in terms of being able to take you through the different drug discovery steps. So the customers can be small biotech research people. And then specifically within those groups, you would be helping medicinal chemists who normally don't use computational tools, but now they can use it for research because it's so easy. So that's one example. Or if you have computational tools where who wants state-of-the-art tools, you want to have a distinction. Like if you use the broadly available software, then you might not be able to get as much as likely or a strong drug for your target. So our virtual screening is novel. And we have invested a lot over the years to actually make sure on the benchmarks and things where they're like... In some cases, we believe we're like 10x better than, say, something you can find in open source. So like docking or certain energy, which means you can literally do less experiments. So let's say you ran a more conventional screening, and then you need to take maybe 500 compounds, and you need to do that level on them to maybe have got a hit. But if you have a better solution, maybe you can only do 100 compounds, which means you get savings and you can get there faster. So that's kind of the offering message for computational chemists and for ultimately business people in those companies who we will help them to get their lead compound quicker. So they want to get from target to lead in a shorter period of time, more cost effectively. That's where we can help them with the software suite.  

Jon - 00:26:56: That's really rad because like, I think, I don't know, I've seen it just, and you know, this is within Excedr, just like having the proper tooling and like getting that 10X, like, and I'm starting to see it, like you talked about just like, try out the AI, like, and you can just speak to it in natural language. And like, how much that type of tooling has unlocked us internally at Excedr. So like, I'm really pumped up, you know, I hope I encourage anyone out there. Give it a shot, like test it out, give it a run and see how you can get unlocked. Cause like I was a skeptic at, you know, at first with all like the AI tooling, you know, in the very beginning. And then I was like, holy crap, like this is actually game changing. And it kind of re reoriented me about like what a given person can do. Like you talked about the 10X Multiplier, like I'm starting to see it actually manifest into like business outcomes for us, which is Awesome. And so, you know, as you're, you know, kind of looking forward and you're continuing to introduce your tool to the, you know, the startup community, the larger enterprises too, I'm going to imagine the, that there's going to be a company building exercise for Deep Origin and also a kind of, you know, a fundraising kind of motion as well. How are you doing, you know, how are you seeking to build Deep Origin from an organizational perspective? And has your philosophy on fundraising changed, you know, since in your Scaleform days? Talk a little bit about just like building Deep Origin as a company.  

Gemma - 00:28:24: I think we are very much an impact-driven organization. In a sense that, for me, I got into this not because of any specific financial outcome, but because I really wanted to understand biology, help other people there, and ultimately solve aging and disease. And my view on that was Turing. That's partially in the culture, but we want to build impactful things. That said, we definitely need to have business outcomes and support our users through the business outcomes. So we have a team of... Like PM and BG, who is really focused on that. I think... In terms of... The mission, we also want to be building really novel things. So we talked about how we started working together with the BioSim team, which was actually another startup. And I was an investor in their startup. In the beginning, they were independent. And ultimately, we decided we want to build a common suite of tools. And it kind of converged that I invested in them. And ultimately, we decided to merge. So that was kind of an atypical. But the good thing is around BioSim team is they have a very strong team in the big part of it in Armenia, which is a country of Armenia, which is because the CEO has roots there. But also, we have a global team and some core in San Francisco. They are very strong at really hardcore research. So when we say we have molecular dynamics, it doesn't mean we took an open source package and shipped it. So there's other tools out there, like AI Systems, who will take open docking and ship it. No, we actually implemented our docking from scratch. And we ran the benchmarks to make sure that we are meaningfully better. We have our own AI, ML, and virtual screen pipelines. So everything we present, we want to take level each and take it to the next level with a combination of physics and AI. So that's actually our strong point is we don't just do a foundation model. We do actually have some foundation models around certain drug discovery things. But we actually try to do this hybrid thing where we use a lot of these knowledge, which has been collected for decades in physics, which has some advantages, but it has shortcomings in terms of too much compute cost, et cetera. And combine it with AI in other ways. So when we build something, we want to build it top notch, and then build it in our tool suite. So part of our culture is to have this hard core research angle to actually deliver something which maybe other companies don't have. So that's, I think, is part of our culture. So we try to do things out as can and build toward this more unified solution as it's simulating biology. So those are like the principles onto which the hardcore research, open collaborative team, try to make things accessible, which is actually on the business side, I will tell you, it's generally a hard conversation because, the first of all, there's a trade-off and we don't make everything accessible. But, if you make something accessible, it has to be free or low cost. And then what do you charge on enterprise? How do you make money? How many users there are? So I strike in that balance as hard. And I think part of the reason, because I've had previous exits and enough funding, we have a few external investors, but we have predominantly self-funded this. We have a bit more leeway to do and open things. Based on mission than some other companies would. That said, we do need to make those trade-offs like any startup. So I think my philosophy on fundraising is probably different. I want to have... Partners who are very mission aligned. And invest inside in building this broader ecosystem of tools and really simulating this part of life. In an accessible way. And then, of course, who can realize that this is actually a non-trivial many years journey. So, yeah.  

Jon - 00:33:01: That's super cool. And I think I love that, like, with the mission to do really novel work and, like, sometimes novelty takes time. Like, and if you're self-funding this thing, andobviously that might not be an option for every startup out there.  

Gemma - 00:33:20:

And it's not going to keep going. And at some point, we're going to be, like, next step is to have extra-to-funding.  

Jon - 00:33:26: But at least in the nascent stages, like, I'm going to imagine you could go down different paths that maybe you couldn't have explored otherwise. And that's super cool because I think, you know, especially in the life sciences, timelines are really long and problems are really hard. And sometimes things just, like, frankly, just take time. Like, and sometimes it is not the kind of the trade-off that you're making, right? Like, do we, like, it's kind of like, do we ship this thing, like, right now in this current form? Or do we, like, let it bake for a little bit longer and try to get it to a place that we can actually say this is truly novel and moves the needle? So, I love that philosophy and kind of that North Star for Deep Origin. And, you know, it sounds like a super fun place to work, too. Because, like, who doesn't want to be working on, like, hard, novel problems that's, like, collaborative? And you've got to have the ability. You are given the kind of autonomy to do that because, you know, you're reflecting on. And then, no, we're not throwing mud on, like, other organizations. But, like, look, with Facebook, it's not a little bit harder. Like, you're, like, I don't get to work on all the things I want to work on. I have, like, 12 different, I have a committee that I need to convince before I can, like, do this thing. So, it's cool to see that you've created an environment where you're allowing people to work on the cool thing and the hard. So as you're looking out. One year, two years. What's in store for you and what's in store for Deep Origin?  

Gemma - 00:34:53: Right now, we are in a stage where we have had a number of early partnerships. We have some traction on Balta product. We've actually had some very excited users. That said, we have a lot of research, such as I talked about this coarse grain research, that we need to turn into a product. We can sometimes do a partnership with it right now, but we want to be able to easily make it accessible to users. So the next two years is really about... Productizing a lot of our research and getting it to a stage where it's really exciting. So for a moment, if you like, I've had like this little video, something which is actually a research tool, I can, let's see if I can share it for just one second. We have this tool called Awesome, which is a funny name. It's abbreviated for associative water-mediated simulation model. But anyway, this is our alternative to say AlphaFold, but it does protein dynamics to actually see how, in this case, there's a pretty long range simulation where a molecule is taking many places and figuring out where it's actually going to be bind. So you can't really run this in a molecular dynamics because it would take too long. But the coarse-grained energy model, it runs about a thousand times faster. So it can do it in much more biologically relevant scales. So it's an example of a type of novel tool we have, which where we can work with partners with right now. And this can be useful for Protex. It's useful for protein interaction, for antibody stuff. It's useful. You can even start with AlphaFold results and then combine them with this to see the dynamics. But, we haven't productized it yet. So I wanted these type of things and actually put it in the product and so people can just run it on their own and also run our tool zone. So that's a big goal for the next year. And really gain core partnerships. If there are speakers here, listeners who want to know, have a target they want to help drug on list with, or even just want to try out our free stuff to see if it helps them without us being involved, we're happy to help. But that's really next two years is really double clicking on partnership, productizing tools and SaaS. And really... Ultimately, also probably getting to the next funding round where we can. Kind of expand some of the lab work we do and expand the offering.  

Jon - 00:37:38: Very cool. That's super exciting. Honestly, I love seeing how it's kind of like, I can see as I've kind of, we've had this conversation, just like how all of the experiences are kind of like accretive and just like adding. It sounds like you, for each experience, you've taken away like various kind of lessons and are bringing it to Deep Origin. And it's cool to see. And I love also the idea of just like, you did mention there are trade-offs, but I love that, like the focus on access. And then eventually you're going to have to, you know, you're going to roll out new products. And I love, thank you for walking through that. The, I can't wait for that to actually, the Awesome to like actually roll out. So everyone stay tuned because that's going to be a, that's going to be super exciting. Now, you know, in tradition and one, first off, like, thank you for spending so much time with me. You've been really, really generous. I'm learning a lot and having a lot of fun. I can keep going on and on and on, but I know you have very important things to be working on. So in traditional closing fashion for the podcast, we have two questions. First, would you like to give any shout outs to anyone who supported you along the way?  

Gemma - 00:38:45: Well, it might be a little bit redundant, but I'd say my business partner, Brendan, for many years. Has been in my life. We're not working closely together anymore, but he's really supportive and helped me get where I am. And we were doing stuff together 15 years. And he's actually one of the investors in Deep Origin. So there we go. It's a very special person for me.  

Jon - 00:39:12: And all the way back from the dorm rooms.  

Gemma - 00:39:14: From the dorm rooms, yes.  

Jon - 00:39:15: Which I love that.  

Gemma - 00:39:17: Involved in each other's projects.  

Jon - 00:39:19: Yeah, I love that. And looking back, if you can give any advice to your 21-year-old self, what would it be?  

Gemma - 00:39:25: So that's actually hard because, you know, when you're 20, you're exploring so much. Like we were sitting in that. Vault programming, that's what I did when I was 19 and 20 and then 21. And it wasn't clear where it's all going to go. So I think there's like a tiny bit of business advice, which sometimes for me to even hard to internalize right now is listen to the market and customers to actually like figure out that you're building what's needed versus what you want. But you have to balance that with listening to your own heart and belief. And I think the most basic thing I would say is probably just... Stay calm, focused, and just steady and keep going, that's life will be okay as long as you are like consistent and persistent in things you want, right? And that's very simple. I don't think it's any very special.  

Jon - 00:40:24: And honestly, like I can't find any other place, like a better place to wrap things up. Because I think to like, especially when you're, especially when I was young, I'll speak for myself, like that doing, doing it consistently and just like pushing through. There were so many times where I just wanted to get like, I have like whiplash and distract, like not distractions, but I was interested in very, like a ton of things. And I could have gone in a million different directions. And the power of compounding and pushing through and being consistent. The thing is like for compounding to really kick in, you have to work at a problem for a really long time and just like chop wood. But if you stop, if you stop chopping wood, the compounding just, you don't actually get to see the hockey stick at the end. 

Gemma - 00:41:08: Yes. So, well, huge thank you to you. I very much appreciate it. A great conversation and the energy you bring to your podcast and to the industry is contagious.  

Jon - 00:41:21: I'm flattered. Thank you for taking the time. I'm, you know, I'm, I'm in the stands rooting for Deep Origin and, you know, We're in the Bay Area. I'd love to grab coffee with you. And, you know, I'm excited for what you guys are about to launch over the next one or two years. So, Mike, thanks again. Really appreciate the time.  

Gemma - 00:41:37: Thank you so much, Jon.  

Outro - 00:41:40: That's all for this episode of The Biotech Startups Podcast. We hope you enjoyed our four-part series with Michael Antonov. If you did, consider subscribing, leaving us a review and sharing it with your friends. Be sure to join us for our next series featuring Johnny Hu, Principal at Menlo Ventures. Johnny invests in biotech and life science companies with a focus on novel therapeutics and new technologies for improving medical outcomes. He has first-hand experience working at the bench to engineer new tools for the clinic. While completing his PhD, Johnny helped develop gene editing technologies that were licensed by companies such as Editas Medicine and Beam Therapeutics. He is passionate about partnering with founders at the earliest stages to translate groundbreaking science into new medicines. Prior to Menlo, Johnny was a vice president at Longitude Capital, where he made several investments in early-stage biotech companies. He served as a board observer at Lectio Therapeutics. Endeavour Biomedicines, Omnibio and Vilya, and was part of the investment team for Amunix Pharmaceuticals. Before Longitude, he was an associate at Omega Funds, where he worked on the firm's investment in Nuvation Bio. In addition to his deep experience as an investor, Johnny has a distinguished academic background. He earned his PhD in Biological and Biomedical Sciences from Harvard University. As an NSF Graduate Research Fellow, and holds an MPhil from the University of Cambridge as a Gates Cambridge Scholar, and graduated summa cum laude with an AB in Chemical and Physical biology from Harvard. With deep experience in biotech investing, gene editing, and life science innovation, Johnny brings a unique perspective to translating groundbreaking science into real-world impact, making this a conversation you won't want to miss. The Biotech Startups Podcast is produced by Excedr. Don't want to miss an episode? Search for the Biotech Startups Podcast wherever you get your podcasts and click subscribe. Excedr provides research labs with equipment leases on founder-friendly terms to support paths to exceptional outcomes. To learn more, visit our website, www.excedr.com. On behalf of the team here at Excedr, thanks for listening. The Biotech Startups Podcast provides general insights into the life science sector through the experiences of its guests. The use of information on this podcast or materials linked from the podcast is at the user's own risk. The views expressed by the participants are their own and are not the views of Excedr or sponsors. No reference to any product, service or company in the podcast is an endorsement by Excedr or its guests.