Thank you. Welcome, everybody. My name is Scott Schoenhaus. I am the healthcare IT, healthcare technology analyst here at KeyBanc. We have the pleasure to have a fireside chat today with Exscientia. David Hallett, Interim CEO and CSO, and Ben Taylor, CFO, joining us today. Gentlemen, I'd like to pass the floor to you to provide a brief introduction on your background and maybe a little bit of brief introduction on Exscientia to investors that are new to the story here on our fireside chat, and thank you both for participating. Sure. Good morning, good afternoon. I guess we're a virtual setting. I guess people can be in various locations. Yeah, thanks for introduction to the platform. My name is Dave Hallett. As Exscientia's Interim Chief Exec and Chief Science Officer. By background, I've been in the small molecule therapeutics kind of arena for nearly 30 years. A chemist originally by education. I've worked at a variety of companies, both kind of biotech and pharma, and joined Exscientia in January of 2020. I guess just how would I describe Exscientia to your audience? So, we pursue transformational small molecule therapeutics. The way we go about that is actually kind of to operate at the interface, so leveraging artificial intelligence, but also incorporating human ingenuity, automation and physical engineering. We kind of operate a kind of business kind of strategy where our internal kind of pipeline focuses on oncology, but we also engage in kind of significant partnerships with pharma and biotech to also advance kind of their pipelines too. Great. Ben? Yeah. Ben Taylor, CFO, Chief Strategy Officer. Spent 15 years in healthcare investment banking and now 10 years in executive management. Have run the day-to-day operations at an oncology biotech for about four years, and then now I've spent about six years in tech and diagnostics. Great. Thank you for that, introduction, guys. So I just wanna start broadly. You... I know there's, several players in this tech-enabled drug discovery or AI to drug discovery space. Maybe how, is your technology differentiated? You mentioned you're in small molecules. I think you've expanded since, David. Maybe just talk about your place in the market. There's several players, and it's not like you're competing against-- you're all competing against each other. And so just kinda wanted, for the investors that are new to your story, to hear the high level, you know, broad-based, you know, marketplace position- Sure ... of Exscientia. No, I think it's a fair question 'cause it is a large ecosystem that we sit within, from a drug design all the way through, obviously, to development. I guess just to set the scene for everyone is that, historically, most potential drugs never make it to market. As an industry, we have a poor track record of success. Actually, I think fewer than 4% of actual drugs that make it into the clinic ever make it onto the market. And why is that? I think there are kind of three core reasons for that. The first one being that the data that looked like it supported a link that said, "Look, if you modulate this target, it might actually help the disease," actually turns out to be fundamentally flawed. An area that's kind of very central to us is that issues that were actually baked into the molecule on the day the molecule was designed, and ultimately they reveal themselves through the R&D process, sometimes quite late and after a lot of money's been invested. And that even includes, like, you design a good drug, but actually you can never test the mechanism. And thirdly, the conduct of the clinical trial. Often incorrect choice of patients in a clinical trial means that statistically, the trial actually fails to meet it... At this endpoint, even if sometimes a subset of patients did respond. So the way that we think about that challenge is that, and it's something that we have to address because the attrition rate of cost of bringing kind of transformational small molecule therapeutics, or actually any kind of therapeutic to market, is almost economically unsustainable at these days. I think current estimates would suggest that's anywhere between $2 billion-$6 billion. It's like that's a significant investment over a long period of time. I think where we start to differentiate is kind of at almost at a philosophical level. So we believe that drug discovery is a learning problem that's based on sparse data. It's not a screening problem or something that you can just simply unpick by kind of leveraging big data and high-performance compute. It's far too complex for that. So as a result of that, that causes us kind of sort of pursue kind of two strategies that that speed of learning, or more specifically in our case, the speed with which we can design, make, test, and then learn, is central and paramount to our strategy. The faster and more effectively you can learn and the systems and the processes you put in place, that makes the predictive model, so kind of re-referencing back to kind of artificial intelligence, makes the models better. And then ultimately, the journey to kind of quality solutions, either drugs, is much more resource and capital efficient.... I think the other thing is that it's important to realize that you can't just use artificial intelligence to get to a drug candidate. That might sound a bit like heresy, but it, it's not. As I said before, that there's a make and test. You have to synthesize molecules, you've got to test them, and that's actually where the virtual world meets the real world. And that's kind of one of the reasons why we've kind of, I think we stand out. We've made recent significant kind of investments into predictive synthesis, into automation, and the orchestration of physical engineering and the logistics. The idea behind that is to basically push the boundaries about how fast we can design and make and test and learn. I think last but by no means least is that we've applied this belief system. We were founded as a company, Exscientia, back in 2012 as a spinout of University of Dundee in the U.K. That philosophy and our capabilities has enabled us so far to deliver eight novel, so these are not repurposed, drugs. It's an important point to note. Four of these are currently recruiting into clinical trials, and we've created substrate for a pipeline to have at least kind of up to two more by 2025. So I think, I guess in summary, it's the kind of—it's that, it's that that true interface between, yeah, using artificial intelligence, but accepting the kind of the there are some fundamental things you have to resolve when you enter the real world as well. And so that investment to automation is something that is kind of central to kind of our processes and something that we will continue to push forward over the coming years. Great. You know, I wanna really focus in on your your precision design. From my experience, it's really a differentiator as well. You design with a purpose, and you specifically design drugs or programs to overcome challenges or issues that have already been identified in competitor compounds, which I think is unique, amongst these AI to drug discovery players, peers. Can we talk a little bit more about your your precision drug design and how it leads to better efficacy versus competitor compounds? Sure. If I may, I'll start that, and I'll kind of... I'll pass over to Ben for some specifics. But I think, first of all, it's like at the very start of the journey, there's a few things that are kind of essential. One is that you spend time to really deeply understand the biological mechanism. You also have to think with the end in mind, think with the patient in mind. How might that drug be used and deployed in the clinic? And also, on those occasions when they do exist, what do we know about early innovators in that space? Kind of what do we know about their compounds if they're known? So that combination of understanding allows us to write what we call a target candidate profile or a target product profile. It's a detailed and written set of kind of sort of characteristics or design parameters. This is what the drug needs to look like. This is what it needs to do. Does it need to work in combination? Does it need to work on its own? Kind of what's known about the mechanism in terms of kind of potential risk for both for safety. So I think our precision design platforms, that means, I think, is we've. With the investments we've made and the times and the learnings over the last 12 years, is that we have a specific advantage for targets that have either been historically kind of challenging or where there are kind of known design flaws with competition. And so we're able to kind of to kind of design against that target product profile and then leverage what we call multiparameter optimization. As I said before, drug discovery is highly complex. There are a lot of variables that you have to get right in terms of from the time that kind of, if you like, that the tablet is taken by mouth, to its journey through the body to the tissue of origin. That's. There's a kind of thousands of kind of little parameters you need to optimize. And so we use that. We leverage the real benefits of AI because that is just something, the multiparameter optimization is something that kind of computational techniques do far better and far more efficiently than a human brain can. I mean, specifically before I hand over to Ben, in oncology, this comes back to kind of probability of success. We're using our human tissue platform to create models, and these models are there to kind of strengthen or sometimes to refute either the link between the target and the disease, if it's early on, but also help guide patient selection strategies. And I'll pass over to Ben at this time to maybe take you through some examples of our current pipeline. Yeah, and I think Dave really hit on an important point there with the multiparameter optimization. I mean, this is just something that we are doing fundamentally better than traditional methods, where you solve one problem at a time, traditionally, and you sort of chemically, you're painting yourself into a corner by your choices early in the process. By the time you get to the end, you don't have a lot of flexibility to change compound properties. What we can do is say, "I want this dozen, two dozen, three dozen," however many properties you want, "I'm gonna layer them up side by side and design for them from the beginning." And so what that means is, we can be more complex in how we think about a drug. And what we do, every single one of our drugs has a story. It has a story of why we think the existing drugs in the class have predictable failure points. And so, what we are doing is not, we don't even necessarily have to look into the biology. Sometimes we do to identify these. Often, it's just in the drug design, where you can look at it and say … with this drug, with this patient population, with this mechanism, that drug is likely to fail. So I'll give you an example with CDK7. So this is the drug that we currently have going through Phase 1, 2 trials. We have data coming out later on this year that'll give a clear indication of did we design this to do what we set it to do? And when we looked at the CDK7 space, we said, "All right, CDK4/6 is a great example class." It's about $9 billion in current sales, but they have a significant toxicity risk, and also the tumors often adapt to the CDK4/6 inhibitors, and then you have progression. CDK7 is a very interesting target because one of the ways that the tumors mutate, CDK7 naturally inhibits. And so if we can get a safe CDK7 drug, there is a high likelihood that that mechanism should be successful because we've seen the CDK4/6s. So we looked at that class of drugs. Most of the drugs in there, the way that they bound, it's called irreversible or covalent binding, means that you're gonna have a high degree of toxicity. Because what you actually wanna do is get in, hit the cancer, and get out of the system. And so we could specifically design our drug not only to hit that CDK7 target really well and not hit any of the other targets that we wanted to in a significant way, but also to get into the body quickly, affect efficiently, and get out of the body quickly and efficiently, so that we were hitting that optimal window for when we're dosing to hit the cancer and to spare the healthy tissue. So that was one big step because we can do multiparameter optimization, we can actually try and do that precision design aspect to it. Another thing that we looked at, and this is something that you'd never talk about in the biotech world. I've sat through thousands of biotech presentations, and not once has efflux ever come up as a core topic of discussion. But this is something really important. So what we identified is these are all transporter substrate. We don't have to get into all of the biology of that, but what it means is that drug's gonna be recycled in your gut, you're gonna have variable absorption, and then the tumor microenvironment is going to try and push it out. And so that one design property, that chemical property that nobody ever talks about, is actually a very likely reason that... Those drugs would fail in clinical trials because you're gonna have a high degree of toxicity in the GI tract, you're gonna have not enough drug getting absorbed, and then the drug's gonna be difficult to stay on target. And so we can look at that, that really complex profile, and then design a drug that meets all of those requirements, and that's, that's what we do. That's our, that's our bread and butter. We've now done it, as Dave was saying, with eight development candidates. We've got five drugs that are in or approaching clinical trials. And each one of them has a story like that. Great, yeah. So wanted to, since we talked about the CDK- CD- CDK7 molecule, just wanna talk about kind of the next steps here. It's already in Phase 1 trials. Do we know... Can we give any color on the timing for key milestones and dose escalation to dose expansion for that molecule? Sure. So that's currently in dose escalation. So we've got two stages of the Phase 1/2 trial. First is in monotherapy, and second is combination. So this is a mechanism where you should see some monotherapy efficacy signal. In the clinic, it would... or in commercial use, it would most likely be used in combination. But you should be able to see, is the drug working or not from the monotherapy. So what we're having is an escalation period that's ongoing right now that started in July of last year. And we expect it to transition to a Phase 2 efficacy portion in the second half of this year. Also, shortly thereafter, we would also start the combination dose escalation, and then move that into its own dose expansion. But I wanna highlight a really important point. A lot of times people look at phase 1 data, and they say: "That's not as important here as the Phase 2." But actually, because of everything that I just said, that Phase 1 data is really important because you'll know from the safety data, from the PK data, from the PD data, did we do what we said we wanted to do with that drug in the design phase? You'll know, does it have a therapeutic window that is better than the other drugs? Is it actually getting in and out of the body? And so not only will that be really important for CDK7, because it is a story of how do you manage the tox, but also it's a, it's a huge point of platform validation for us. Great. I kinda wanna shift over to the first-in-human study for Bristol Myers, Bristol Myers Squibb in-licensed PKC theta inhibitor. I know we can't talk about the timing or the design of the program and actually, can we talk more about the design of the program and less about the timing? And I believe there are several other competitor programs based, focused on that target. Yeah, absolutely. I think, like, again, this I think it's a good example of a challenging but highly interesting immunomodulatory target. So you are correct in that, historically, somewhere between sort of 10-20 companies have tried to deliver an orally bioavailable, selective PKC theta inhibitor into the clinic. Kind of a mixture of biotechs and big pharma. To the best of our knowledge, only two ever made it. The first one, kind of made it as far as phase two clinical trials. Kind of showed some hints of efficacy, was highly potent against the target, but was kind of not selective against other related kind of kinases, 'cause this is a kinase biological target. As a result, it... There was no therapeutic index to kind of the efficacy, kind of adverse event ratio was, like, not good enough. Another kind of major biopharma managed to get their compound into Phase 1. But what they found is that while they'd addressed the kind of selectivity issues, that they just simply weren't potent enough at the target. So nobody's really ever kind of brought forward a kind of molecule that's kind of good enough. And we were able to do that. So, with a collaboration with originally Celgene, and then kind of obviously through the acquisition, BMS. BMS in-licensed a development candidate that we designed specifically to be highly potent on target, but also selective, so that we can actually ask the question in the clinic about this particular target. Bristol Myers Squibb began their healthy volunteer study in the first quarter of 2023. And we're expecting kind of an update from them, if you look at normal timelines for healthy volunteer studies, shortly. But again, just to finish off and come back to Ben's point, yeah, people tend not to get excited about healthy volunteer studies, and I understand why. We're always looking for efficacy. But again, this is another example of that, that the importance of that healthy volunteer study, I think will tell kind of Bristol Myers that, have they got a molecule that's actually kind of good enough at the right doses to test the mechanism? But it has a huge potential, that one, and ultimately, it is a has a high potential for being a high-value, first-in-class drug. Great. Maybe let's move on to your Sanofi program. How are the milestones progressing there, David and Ben? Sure. So I think kind of made fantastic recent kind of progress in this important collaboration. So it covers both oncology and inflammation immunology targets. Kind of very recently we achieved our first discovery stage milestone for one program that was publicly released. And then towards the kind of sort of late 2023, early 2024, announced that a really high-value kind of inflammatory target that was actually part of Exscientia's own internal portfolio had been added to the collaboration. Again, this is a target we, we're not allowed to reveal a target, but it is, it is, it is a well-known one. And again, a whole variety of companies have pursued this mechanism, and we've identified a kind of a common but major flaw that exists in that, within their design parameters. So we understand that and we've been able to kinda design against that. And I think with that in mind, Sanofi liked what we were doing. I think that is the best home for this. As I said, we're an oncology-focused company. They have a very strong presence in this particular franchise. So we were delighted when they brought that in. We've actually enhanced economics to reflect the kind of pre-investment that we'd made. Great. Oh, I wanna talk about your new deal with Merck. Broadly, how's the initial phase of the program going, on the Merck, deal? Sure. That's, that's a relatively new one. I think it's the projects have started well, writing the research plans. Yeah, building that relationship with the partner. These are three targets initially, in the, again, in the oncology kind of inflammation space. And yeah, made good initial progress. And we'll but it's, it's still very early days, and so we'll, we'll look to update the kind of the public and the capital markets as we make progression through those, through those discovery stage milestones. Following up on kind of these, your new business development and your partnership strategies, are you thinking about broadening the type of deals you're doing outside of just discovery platform deals? Yeah, definitely. And I think, one of the interesting things for us is, it really is a platform. So we have developed it not around a single source of AI or something along those lines. It's really been 12 years of problem-solving and using whatever is the latest technology, whether it's virtual or experimental, to advance that. And so what we've ended up with is a really, broad base of technologies that are all designed to be integrated together. And so, I think what you'll see from us is a mix of different, types of partnerships, and BD opportunities going forward. We love to figure out ways to, you know-... Add even more money and value to our agreements, like you saw with the Sanofi addition to that collaboration, where we took it a little bit further ourselves, and we got great value out of it. So we're never afraid to license programs. We've got about 20 programs in development right now. I think also we're open to different ideas on the tech side. On the tech side, just to be clear, we are quite different than a traditional AI tech company. We're not a big data company, where, you know, we're gonna generate a lot of data and then screen it to try and ask a question. What we actually do is we know what the end product we want is, so we generate really specific data around that and then use the AI to put it all together. That is actually really exciting. In the tech space, we do a lot of very innovative AI, but it's a little bit different than traditional. I would expect our partnerships in that space to be a little bit different as well. That makes a ton of sense. Lastly, I wanna talk about your new facility that opened last year. How important is this integration of the in silico computational, and then the hardware to execute on the efficiencies and volume of programs? As I think I said in my earlier remarks, I think it's essential. It's kind of critical in the way that we think about drug discovery. So as a reminder, we talked about... Because we truly believe that kind of the complexities of drug discovery are involve learning and iterative kind of cycles of learning, there's a need to kind of make compounds and to test them and to generate high-quality data. And something that people often maybe not realize is that, you've also got to generate that data in a very conscious and a specific format, so it actually enables machine learning. You can't be passive in that sense. The kind of, sort of the fidelity and reproducibility of automation kind of plays very much to the strengths of kind of artificial intelligence and machine learning. That investment and a major investment was kind of critical for us to kind of really drive forward kind of towards kind of economically sustainable drugs. I think we've kind of pushed our current operating model pretty much to its limits, so we've been incorporating kind of biology experimentation in certainly for quite a few years. But still, like a lot of companies, outsource, have an outsourced strategy in this space. And I think in order to kind of to get to the sort of the next level of kind of speed of learning and we needed to make that investment. So I think we are genuinely kind of leveraging and operating at that interface between AI and human ingenuity. We literally encode and automate. That's something you might see in some of our literature. And why are we doing that? The reason we're all here is that, and why I came to Exscientia, I think, and why all my colleagues are here as well, is that patients need better medicines. They need to be designed and developed and brought to market much faster and in an economically sustainable way. And that's kind of what we're about at Exscientia. Great. Well, I think we'll end it there. Thank you so much, David and Ben, for joining us at our virtual healthcare conference. Truly a fast-paced and fast-growing and important industry and company, Exscientia. If anyone has any further questions, feel free to reach out to me, and I can connect you with the management team at Exscientia. Thank you, everyone, for joining. No, thank you very much, Scott. Perfect. Thank you, Scott.