Welcome, everybody. My name is Scott Schoenhaus. I'm the healthcare IT analyst at KeyBanc. Pleasure to have Zach Jonasson, CFO of Absci. We launched on AI to drug discovery or tech-enabled drug discovery players in December. And Absci has been the leading performer from a stock performance perspective, with some nice updates on partnerships and pipeline. So really a pleasure to have Zach, you here representing Absci. Maybe sort of give your background to the audience and maybe just a high pitch, elevator pitch of what Absci does, so anyone that's new to the story. Sure, and, thanks for having me here, Scott. It's a pleasure to be here. Just a brief bio, I guess. I started my career doing economic research at the Central Bank, of all places, and then I did a PhD in cognitive neuroscience at Harvard, and then did a couple startups from there, where I was the founding CEO or founding CBO. And then for the last 15 years before Absci, I built a venture capital firm, so investing, in drug discovery in particular, in the last five years of that, heavily focused on the intersection of AI with drug discovery. And that's actually how I initially intersected with Absci. I led the Series A round back in 2016, and so was a chairman of the board for a number of years, worked closely with management, going back to when it was only an eight-person company. It's been a really remarkable journey. This last year I decided to join as the CFO and CBO and roll up my sleeves and work with the team. Great, and, sort of give us a pitch on Absci, the technology, you know, how it's differentiated from other AI or tech-enabled drug discovery platforms that are in the market. Maybe just touch upon all those. Yeah, I think- Differentiation ... right up front, I would say, we focus on biologics, in particular, antibodies or antibody-like drugs. So most of the other AI discovery platforms are focused in the small molecule area. So right up front, that's, I think, a big differentiator. We also refer to ourselves as a data-first AI company, and that's sort of really, that really sort of signifies to us the importance of data, and that's a key differentiator about Absci. The company started its history as a synthetic biology company and built up a technology and an IP base around how to produce antibodies at really high throughput in heavily engineered E. coli system. And we've since taken that, that system and made it into a high-throughput assay system, where we can interrogate millions of antibodies against a given antigen and look at the binding properties at scale. So in a one-week process, we can look at approximately three million antibody-antigen interactions and measure those. Absci—one of Absci's really primary differentiators is its ability to generate large training data sets and then to use those same assay systems in the wet lab to then validate the AI predictions. We create this virtuous cycle, where we create training data to build AI models that predict designs, antibody designs for a given target, and then we can validate those in the wet lab. We take that information from the validation, and we use that to tweak the model and improve the model. Each time we iterate, we're creating more data and improving the model. Just to dig in a little bit on the data, too, you're starting to see more companies talk about data. The way we think about data is really in three dimensions. One is, is it usable? It's one thing to generate tons of imaging data or other data. The question is, is it really usable for the task at hand, designing a novel therapeutic? Secondly, is it quality? Does it actually give you precision and accuracy that you need? And third, do you have enough data? Is it scalable? And so I think at Absci, our proprietary system for generating data meets all of those criteria. We're really looking at that protein-protein interaction, so the functionality of the antibody at scale, and that's data that's not available in the public domain. So we generate that in-house, and we generate it at scale, and it's high quality. When we look at our ACE assay, which is this E. coli-based flow assay that I mentioned. It correlates very highly with SPR, for example. And so we can use that as a really high-throughput way to generate that training data and to do the validation work. Just to follow up on your point, you made, you know, Recursion, which probably many investors are also familiar with, focuses on small molecule. There's a lot of public data available. I think that's a key differentiator. You're in biologics, where the data sets are much smaller, public data sets are much smaller. So generating your own data is really a high-value differentiator in biologics and your, and your, through, and your technology, right? I think it's true for even the small molecule space, but it's definitely true for what we're doing. And if you don't generate the data, the functional data, I don't—there's no way to really build a foundation model that's gonna give you the results that you're looking for. Zach, my next question is, you know, I think of you guys as really focusing on the AI portion of AI to drug discovery among the coverage list that I currently cover. Maybe can you talk about your zero-shot generative AI? So you're basically designing antibodies without using any training data, and it's kind of breakthrough technology. Maybe we can touch upon that, and kind of break it down to an easy, digestible- ... tech investor, what it- Yeah ... what it does. Yeah, and so we use training data for sure. I mean, we built a foundation model to do de novo zero-shot design of an antibody. So the way our model works is we fit in an epitope of interest, and this, I think, very important. We can select the epitope on a given antigen, and we feed the structure in, or if we don't have the structure, we feed the sequence in, and we predict the structure. And then the model can design antibodies to that epitope.... and that's done in a way where we don't need anything in the training data set that's actually on that target or an example of an antibody that binds to the target. It's based on just the general training data that we've created. So about a year ago, we put out a manuscript where we showed that we could in a zero-shot fashion design a novel antibody to the HER2 antigen. And we cleansed all the training data in this experiment to not include anything with any appreciable homology to trastuzumab, which is the leading antibody that does target that antigen commercially. And the model was able to find antibodies with higher binding affinity, and we zeroed in on one just as a proof of concept and showed that we actually found one that was higher affinity, or sorry, higher potency as well. We've since taken that same approach and applied it to generating internal assets in our pipeline. So we talked about at JPM this year, we released some data around our TL1A program, ABS-101, and we used the same approach there to do a de novo design of the CDRs to create a novel antibody that targeted the TL1A receptor. We're able to engineer higher potency versus competitors, a longer half-life, and some other developability parameters that look highly favorable for that. That was all done in a first step by de novo design in that zero-shot foundation model, and then we have a second model that looks at AI optimization, where we really dial in the developability parameters. So you just brought it up, and that was a question I had down below, but we'll talk about it now. Let's talk about your TL1A molecule. You just entered into IND preclinical studies. My understanding is that there's other clinical-stage assets being developed by Merck, Roche, Sanofi in this area. So how does your... It's ABS- 101. How does your ABS- 101 molecule compare from early readouts? And then maybe talk about what the monetization might look like, given what we have seen in the public markets or given what we've seen with these competitor assets. Yeah, I mean, I would start out and say IBD is a very large indication. It's over $20 billion a year revenue potential, so it's a big market. There's room for more than one molecule. But I think what we did with the TL1A program is really exemplary of what you can do with AI. We took a look at the molecules that were ahead of us in the clinic, in particular, the Prometheus molecule, which is now owned by Merck, and then the Roche molecule, which used to be the Roivant, was developed by Roivant, as well as Teva. We looked at the weaknesses of those molecules, and we were able to design what we believe is a best-in-class antibody that addresses all those weaknesses. So in, you know, sort of in order of importance, we've designed in a longer half-life, which should enable more flexible dosing for patients, potentially once quarterly. The other molecules look to be months, monthly, and the Teva could be even less than that. It seems to have a very poor half-life. Secondly, we engineered higher potency, and we've demonstrated that in a number of in vivo assays, as well as cell-based, and we've also looked in sort of some mouse studies to demonstrate PK and some other parameters. And then thirdly, we've designed this molecule so that it has very good developability parameters, easy to formulate in a high concentration, so should be suitable for sub-Q administration. So we did all of that using the AI platform and looking at, kind of... We've done the comparative studies to validate this, just looking at what the weaknesses were of the leading molecules in clinical development. We'll plan to be in, finish our IND-enabling studies this year and be in, phase I trial early next year. Great. I want to move on to your partnerships. So let's talk about your partner programs, which pharma companies you are currently partnering with, who you'd like to partner with in the future, what targets or therapeutic areas are those partners interested in deploying your technology? Yeah, I would say, you know, our business model is really focused on partnership, and we do it both at the drug creation phase, as well as we will expect to partner our internal programs at later stages of development for higher economics, of course. But in the drug creation phase, we have a number of marquee partnerships that include Merck, AstraZeneca, and Almirall. And I think in all of those partnerships, it's a pretty similar deal structure, where there are upfront payments, research fees, option election, and then there are milestone payments for clinical milestones and commercial milestones. And each of those agreements also includes a royalty, which typically we're not allowed to comment on publicly. So what we're doing is building a portfolio of those partnerships at the drug creation phase. And to your point, we like doing that. They give us non-dilutive capital, of course, but they also give us access to real synergy. So I, I'll just point out, our AstraZeneca deal is working with AZ on an oncology target. It's very hard for them to drug with traditional means, and AZ has a very deep experience in oncology, great ability to execute clinical trials, so it's a really nice synergistic partnership. Our partnership with Almirall that we announced last year, that's about a $660 million deal. The AZ one was one target, $147. Almirall was a two-target deal. Those targets are based in dermatology indications, and that Almirall is a pure play pharma that's focused in derm, so great synergies with them as well. And shifting back to your internal pipeline, maybe talk about the strategies, near and longer term. My sense is that you are trying to progress these assets further along the clinical trial side to create better economics, potentially in the long term, or value for these assets. Is that the right way to think about your strategy, Zach? Absolutely. And, And, you know, I think the reason to do this is if you look at what we've demonstrated with the TL1A program, we were able to reach a drug candidate in 14 months and about $3.5 million of spend. We'll finish the IND-enabling studies in a roughly two-year timeframe and about a $12-$13 million spend. If you look at what pharma does to do the same thing, to advance from a target all the way to an IND, that's typically a five-year project and can be $50+ million. So we can do three-four times the programs for the same capital allocation. So our goal is to sort of feed that AI engine, and keep in mind, we're constantly improving that AI capability as well. But the idea is to really capitalize that. It's very efficient, develop assets, and then to monetize those in partnerships once we've got validation in the clinic, whether that's Phase I validation or phase II validation. We would not be looking to go beyond phase II. I guess, the last question I'll have on the internal pipeline is: can you just remind us what assets you have in the internal pipeline and what therapies they address? Yeah. Yeah. So the lead program obviously is TL1A, and that's in IBD. We are looking at other indications, fibrotic indications, as potential opportunities as well. Right behind that is ABS-201, which is a dermatology program. We have not announced the target there, but we would expect to announce that late this year, probably in the second half, and announce a drug candidate. And then the third indication, or the third program, ABS-301, is an immuno-oncology program, and that's a very interesting one. We have a target discovery platform as well that's based on taking patient samples and looking at TLS biology. We basically take a patient, and we can have a longer discussion around tertiary lymphoid structures, but it's essentially patients in autoimmune disease and oncology will develop a localized immune response, and that's through a structure called the TLS. And so we take those TLS samples, and we mine those to understand the antibody repertoire. And so in a specific high-responding patient, we discovered this novel human antibody that has some very interesting pathway modulation for immuno-oncology, particularly modifying or modulating the innate immune system. So we would look to have some in vivo data on that program that we would share later this year as well. And then, you know, as you might imagine, we will continually be developing new internal targets to bring forward towards drug candidates through the course of this year and next year. Just out of curiosity, when you target new programs, is it gonna be of the same kind of origins of TL1A, where you're looking at other competitive molecules that you feel like are not adequate enough, and your technology could help discover better molecules? Is that how you... Is that how you kind of think about it in terms of, like, the monetization you're seeing at the end, and then you're targeting those to develop in your pipeline? Yeah, you know, the short answer, it's a mix. We wanna have a portfolio because the fast follower approach is one where we think we can design the better molecule- Yeah ... but there are already some other molecules in development ahead of us. And we can leverage, you know, their weaknesses. We can leverage what they've done in the clinic. So that's a lower-risk strategy. So we wanna do some programs that fit that model, and then we wanna do some other programs that are first-in-class or novel targets, such as the 301. And that's why we have this target discovery platform that's integrated into Absci. And I would say, too, you know, we were very careful about sort of embarking on the journey to develop our own programs, and we brought in Andreas Busch. He used to be the head of R&D at Shire and then before that at Bayer. He has 10 approved drugs. He brought in a team that he had worked with before at Shire and Bayer, and they really lead our drug discovery effort in terms of, you know, deciding on which targets, which indications. And so we're leveraging a, I'd say, a very differentiated set of expertise there to make those decisions. So recently, you did a public offering. You initially targeted raising, I think, $75 million. You were oversubscribed, and I believe a little over $80 million proceeds? Yes. So actually, we initially on the book was $50 million. Okay. We ended up being about 7x oversubscribed, and we're very happy with the financing. We upsized it. With the greenshoe, we ended up raising gross $86.4 million last week, so we're very happy to add that to the balance sheet. And- I was just gonna say, what are you gonna do with that cash infusion? What, what investments are you gonna make? We're gonna invest it very wisely. No, I mean, that's gonna allow us to have runway into first half of 2027, but really, more importantly, achieve a number of key milestones. So we'll be able to bring ABS-101, our TL1A asset, all the way through phase I. The phase I should complete late Q2 or Q3 of 2026. We should have an interim data readout on the phase I in the second half of 2025, so it'll enable us to advance that program. It will also enable us to advance 201 and 301, the other two pipeline programs to IND, as well as to advance some other drug candidates. On top of that, we also plan to continue to do partnering, and so we would expect to do more drug creation partnerships with pharma and biotech over the course of the next couple of years. Great. My last question is kind of a philosophical, higher, broader question about the industry over the next 12 months, maybe next five years. What, where... you know, we talked earlier before the fireside chat about being kind of the golden age, with both AI and kind of biotech, assets being deployed.... Where do you see these tech-enabled AI drug discovery players, like yourself in the future? Is it a consolidation effort? Is it, a takeout from large pharma? Or is it just, you know, each of you has your own sort of unique technology, which is, you know, valuable to be part-- through partner programs and internal pipelines that can continue to be developed. How do you view the space in the next five years? Yeah, it's, it's a, you know, I think about this a lot. Yeah. I think, you know, just in the short term, you're gonna see companies like Absci, which I think is a clear leader in the antibody space, but we're gonna see advancements in different modalities. So there'll be companies that are gonna make great advancements in small molecule. I think we're gonna lead the way in antibodies. There are other companies that are just starting to apply AI to, we were talking about this earlier, to engineering novel or designing novel genetic enzymes. So I think we're gonna see it applied to lots of different drug modalities. And we're now seeing, and so just sort of transition to where pharma is gonna go. You know, a couple of years ago, when you talked to pharma executives, there was sort of this question of: Well, is AI real? Is it going to transform drug discovery? That's no longer the question. The question pharma's asking itself now is when? When is it gonna transform drug discovery? And so we've seen pharma companies investing more and more in trying to build their own internal capabilities, and this is traditionally what you see. I don't think they will be successful, by and large, but they'll have some success, but I don't think they'll be as successful as the more nimble companies that really build the whole company around how you integrate data and data creation with AI, and that's the, that's really how the, the companies are built. So my expectation is, as we move forward in time, is that pharma will initially start buying and partnering assets, which is what they're doing today and doing partnerships. But ultimately, it seems to me like pharma will come to a point where there's a tipping point, and it moves into acquisition. We don't have a strategy to be acquired. I don't believe acquisition is ever a strategy. We certainly can respond to those kinds of interests. But if you look at the economics, the unit economics that I mentioned before, where we're able to get to an IND in, let's call it $12-14 million versus $50-60 million, with really high-quality assets, designing in the features that you want. It's really hard to ignore that when you look at the current ROI on R&D spend at large pharma. So I think it, eventually, that's probably where this heads. But in the meantime, I think, we're in a good position to just keep generating super high-quality assets and partnering with pharma. Great. I'm gonna open up the floor Q&A. So, for the partnership programs, do they share the data that you can, that you can train on it, or how does that work? Yeah. So this is a really important point. In all of our partnership programs, we do data generation using our own assay system around the target of interest, and we do keep that for improving our model. If the partner contributes its own data, then they would keep that, but we're using our own. As I mentioned before, our assay systems are designed to generate the kind of data we need for training, so we keep that data. Yeah. Another question, just, you know, when you think about, like you said, maybe it's $15 million-$18 million to take it to IND for the full program. How do you think about optimizing, like, how many different kinds of programs do you really ultimately want to go? I mean, obviously, you just have money that's gonna dictate to a degree how much, how many programs, but how do you just sort of... It's a giant universe of opportunities. So how do you think about narrowing down to optimize on the right programs? Yeah, and a couple of comments there. One is, you know, in some sense, you're capital constrained, but on the other hand, the AI platform allows us to scale that effort in a way that's, there are definite economies of scale, right? It's not a linear scaling. We can do a lot more with AI. We need more program managers and people like that, but we can really scale up that effort with capital. But that's part of the answer. But I think the more direct answer is, you know, we have a team of deep experience that's led by Andreas Busch, and we have a target selection committee that's managed by him and his team. And so we evaluate targets for fast follower, and we also evaluate first-in-class targets that either come out of our program or in some of the co-development partnerships that, that we structure. So we have a co-development partnership with a company called PrecisionLife, for example, and that's a place where we're looking through their platform to see if there's a first-in-class or a novel target that we'd want to pursue. But that effort and that evaluation is led by a very experienced team that's headed by Andreas Busch. That last point on PrecisionLife, maybe that's my last question here, is when we think about uses of cash outside of deploying for your internal assets, is there other optionality that you could see on acquiring little sort of bolt-on or partnering with other companies to enrich your data set, basically? Yeah, and look, I have the CBO title, and I'd say it, this is. It's a lot of fun because having the platform here opens up lots of creative possibilities. So we do have initiatives to explore, potentially spinning out some assets with other parties. We have some initiatives looking at R&D financing structures to enable more of this asset creation. So we are trying to look very creatively at how do we leverage the platform in a very capital-efficient way and really maximize the output. Great. Well, thanks, Zach, so much for coming, and thank you, everyone, for attending.