🔬The BioAI Phase Shift - Matthew McPartlon & Neil Patil, Chai Discovery

Welcome to Latent Space

0:00It looks a lot less like a, you know, a ChatGPT and a lot more like Autodesk or SolidWorks or Figma, you know, if you've used those things where you can kind of load up your molecule. There's this almost like Photoshop-esque, like, design suite. You have this equivalent of a paint tool to kind of paint your epitope. You have this equivalent of a content-aware fill tool to kind of get your binders generated from Chai. And I think to add to that, right, you have this notion of target discovery and hit discovery and optimization where each of these has a gate and takes a few months to a few years.

0:30It's this very, like, waterfall model, right, where the cost of trying things and getting things early is very expensive. But I think to what Matt's saying, right, if you start to get in a regime where you can have models give you really promising candidates, you can start to make that look a lot more like a loop, right? It's akin to, like, becoming more agile in software development. But now the next problem is, like, agonists, right? Like, how do you reliably one-shot hitting a switch, like, on a cell, right? Or bispecifics or ADCs, right? And I think this levels of abstraction that we're going to have to climb with the product as, like, the models get better.

1:04If you have, like, these really good primitives for structure prediction and binding and design, and you can kind of compose them, then you can start to just, like, grow into, like, the outer loop of science.

1:15Welcome to Latent Space, AI for Science. I'm Brandon. I build RA therapeutics at Atomic AI. I'm joined by my co-host, RJ Haneke, CTO and co-founder at Mirror Omics. It's a pleasure to have with us in the studio today, Matt McPartland and Neil Patil of Chai Discovery. Chai is a protein design startup, which is about two and a half years old and has made quite a splash in those few years. They have several very exciting announcements that I think they'll tell us about today. But, yeah, to get started, could you two give us a bit about your background and what you do at Chai?

Backgrounds at Chai Discovery

1:46Yeah, thank you very much for having us. We're super excited to talk about Chai today. I'm Matt McPartland. I'm one of the co-founders of Chai. My background is in, like, AI, biology-related stuff during my PhD. I actually started my PhD in, like, theoretical computer science and then transitioned to this later. Yeah, I've been doing this stuff now for, like, about eight years, and I kind of came into the field at an interesting time where protein structure prediction was, like, just starting to see signs of life. So this is, like, AlphaFold 1 days and was in the field during AlphaFold 2 and, like, got to see a lot of the interesting developments at that time.

2:23So, yeah, I'd always been pretty interested in, like, applying this stuff in the real world, and Chai was just a perfect opportunity to do that. And I'm Neil Patil. I help lead a platform and product here at Chai. So a lot of the stuff around infrastructure to train models, serve them, and then the productization piece, you know, the design suite that lets you use the models. I kind of have a more meandering path, so I kind of got into programming, like, 15 years ago, making apps in the App Store, got really addicted to the dopamine hits you get from that, and then actually got nerd sniped by robotics and, like, worked on that for a bit, self-driving cars in, like, 2018, 2019, got really jaded and was like, I don't want to touch hardware for a while.

3:00I ended up switching and joining a SaaS company called Vanta as one of the first employees there and kind of grew with it. I started my own security company afterwards, got a few years into that, and I was like, you know what? Adams are kind of cool. Like, I want to work on something a little more meaningful. And so I joined Chai about a year ago, right after Chai 2 was announced to help with a lot of the platform and commercialization pieces. Awesome. It's like the five stages of grief or something. Yeah. Yeah, we're at acceptance. Awesome.

Partnerships and Business Model

3:28You have these, I think, four now big partnerships and raised a whole bunch of money. Can you tell us a little bit about those partnerships? And then what I really want to know is what are you telling investors and customers that is so compelling that they're willing to do these big deals? Yeah. So, like, we've been very fortunate to partner first with Eli Lilly and then with Pfizer, Novartis, and Argenx. Yeah, I think it's been, like, a really interesting ride, and I think our business model is also very compelling to a lot of people.

3:59Like, we really like to, we care about the partners succeeding. Like, Chai as a company really depends on how the partners succeed. I think Neil probably has some interesting takes on, like, you know, what we actually offer and what makes that so compelling. So, I'll hand it over to you. Yeah. I mean, as you all know, drug discovery is a very lengthy process, right? And a lot of these pharma companies are spending lots of time, you know, years and years and billions of dollars trying to find initial therapeutic candidates. And so, at Chai, you know, we train models that can help accelerate that process and kind of find those initial binders and then some.

4:31And, you know, we, you know, there's a lot of bio companies, AI for bio companies, that are, like, making their own drugs. We really don't see ourselves that way, right? We see ourselves as almost a neutral software factory for making medicines. And so, that's what, you know, lets us go then work with and support all of these other farmers in their kind of drug discovery journey. And so, yeah, I mean, a lot of this capital is just another proof point that we can sort of start to really accelerate that software factory, right? Go after harder modalities, train bigger models, and ultimately just build what our partners and customers ask us for.

5:04But what is it that, why you and not other structural companies, why are they compelled to buy from you? The thesis of Chai has always been to, like, be the software and modeling layer, which was, I think, like, very controversial at the time. Like, everyone, you know, this play has definitely been tried. Only two years ago, and it's already, like, a completely different world. Yeah, it's pretty crazy. Like, people tried this play for a while. And I think, like, the models just really weren't there yet. And even, like, for us, we were taking a risk in the very beginning.

5:35Like, we were kind of banking on the models getting there. And, like, I had seen early signs of life in my work and our CEO, Josh. Like, he was on the original ESM papers on that team in Meta. And he was seeing, like, pretty early signs of life that, like, you know, there might be scaling laws here. They, like, I think we'll actually be able to start, like, designing things. Structure prediction is getting really good. Like, one, like, crazy thought is, like, we didn't have a multimer structure prediction model until, like, 2021. That was five years ago when we could, like, start with deep learning to, like, actually predict the shape of two proteins at once.

6:06Like, it was a, outfold one was, like, and outfold two was, like, this huge breakthrough. But then, like, outfold two multimer came out, like, a year later. So, like, you really kind of needed that to unlock design in the first place anyway. Like, we weren't even trying to predict multiple proteins at once. And then, really, like, around that time, inverse folding kind of started working. And it was, like, oh, protein MPNN. This actually works in the lab. Like, credit to the Baker Lab for doing all of this really excellent lab validation on all of their models. But I think, like, we're starting to see them do interesting things and, like, actually work on, like, real-world experiments.

6:38And now is probably the time to start betting on this. I think, like, before then, maybe you could take, like, some experimental data from a campaign on, like, this one target that you had and you care about. And you might be able to, like, make some progress on that and, like, keep hill climbing in this, like, one very specific case. General models weren't really a thing back then. So, I think, like, yeah, we took that bet pretty seriously. And, like, we decided to just, like, push as hard as possible and to really, like, shoot for generality in our approach. And then when CHI 2 came out, our second paper after CHI 1, we kind of, like, showed the world, like, this is actually possible and it's possible at scale.

7:14We didn't show this for, like, one or two targets. Like, it kind of works. Like, we were like, let's just go all in. I think Josh likes to say we set up bold company-wide challenge to design antibodies to 50 targets. And actually, like, we saw some signs of life. We're like, all right, let's, like, let's do this with real statistics and see if this actually works. It's an interesting story of how we chose these targets. So, we were like, all right, what targets are we going to choose? We should choose, like, some interesting targets, whatever. And at that point, we were, like, kind of ramping up with CROs and figuring out, like, what does our wet lab process look like?

7:45And we decided after trying some stuff with, like, mini proteins, whatever, we're like, here are the interesting targets. This is what we should look at. And, like, half the time the targets just, like, kind of didn't work. We were still learning, whatever. And we're like, all right, maybe we should just go with, like, targets that the CROs have actually validated. So, let's get the CRO catalog, see what they've already worked on, restrict that to, like, an interesting set. So, from that, we chose 50 targets, designed antibodies against them, got hits to half. And at that point, I think pharma started to realize, like, okay, there are actually signs of life here, and this might actually work in some of our programs.

8:17And so, antibodies is maybe a more challenging domain than other structural prediction problems. So, why tackle antibodies?

Why Tackle Antibodies

8:28So, maybe back up, what is an antibody? Yeah. And what do you do with it, and why is it an attractive target? The analogy that everyone gives is, like, this lock and key kind of problem, where, like, your target, this protein that you're trying to bind to, it might be some, like, disease protein. That's kind of, like, your lock, and then you want to design this key that fits into it, and, like, in our case, just, like, sticks there. The interesting thing with antibodies is, like, these, like, really flexible, general proteins. Like, in a lot of ways, they're very general. In a lot of ways, they're actually, like, pretty uniform.

8:59But at least, like, how they bind to a target is very general. So, like, you have a lot of optionality in how you design this kind of binding interface. The structure prediction problem for antibodies, like, predict how this antibody actually binds to the target, how the key fits into the lock. That's been a notoriously difficult problem. The nice thing is, like, so, we've made a lot of progress in structure prediction. Kind of the field as a whole has come a long way along, like, in getting structure prediction to where it is. But in the design setting, you can be a lot more selective about the types of designs you want to make and the types of structures you actually want to focus on.

9:35And in some cases, it might actually be even easier to design a protein binder that is an antibody than to actually predict how it might bind that target in general. So, like, it's kind of like if you have the freedom to choose, you can kind of just pick the easy cases, if that makes sense. So, the antibody is, like, there's a whole machinery in the body that works with antibodies. What does the body do with it naturally and what can you do with them that is sort of not natural but is useful for therapeutics? This is coming from a non-biologist here.

10:06But I think of antibodies, like, there are these kind of, like, Y-shaped proteins. So, like, kind of looks like a P sign with your fingers. Each of these fingers is kind of like an arm of the antibody. And, like, it's really actually only the tips of your fingers, the tips of the antibody, that engage in binding. So, this makes these really, like, nice therapeutic design targets for that particular reason. The nice part is that, like, the rest, apart from the tips, is, like, actually relatively constant. So, this is called, like, the framework region of an antibody. And the design problem, you're typically just designing, like, the very fingertips.

10:40And you can actually choose, for the most part, like, these kind of framework regions that your immune system already recognizes. So, antibodies, kind of like these Y-shaped proteins that your immune system, like, recognizes and knows really well. It's kind of like your body's, it's one of the lines in defense against pathogens and other types of diseases. So, I guess antibodies can, on the one end, like, connects to proteins on the surface of a cell typically or other things, but typically on the surface of a cell. And then the other end helps the immune system identify a pathogen typically.

11:14But you can also do things like, you mentioned, ADCs, anti-antibody drug conjugates. So, that means putting a drug on the other side or something like that. And that causes the, when you bind to something, that it releases the drug into the cell. Right. They're like this very general framework, right, where kind of on the ends you have these CDR loops. And you can design them to kind of bind to arbitrary things, where maybe one end you bind to a cancer cell, the other end you bind to a toxic molecule. You're now precision delivering that toxic molecule to a cancer cell, right?

11:45Or you just have two ends bind to things and kind of force, like, induced proximity to have some effect in the body. Or, you know, a lot of drugs historically are really just, like, about, like, blocking things, right? Like, anti-agonist behavior, right? But maybe you can have agonist behavior where you actually, like, really precisely, like, press a switch. Like, there's a GPCR protein, which are these, like, doorbell proteins that sit in your cell membrane. You have an antibody, like, very precisely engineered to poke it in a certain way that causes a downstream chain reaction.

12:16And I think, like, one of the things that's really exciting about where we're getting to with some of these models is we can start to get that precise, right? We can really target a very specific epitope, right, meaning, like, binding spot, right? A very specific set of atoms to have the antibody go after, which, you know, historically, you're, with a lot of drugs, you're just kind of brute forcing, you know, a lot of antibodies and just trying to come up with a bunch of things and see what sticks. But maybe that gets you a binder to some spot of your target molecule, but that doesn't let you precisely engineer where you're poking after. I know you're not biologists, but do you have any, like, idea about how they used to design these before, you know, these models came up?

12:52Like, what would you, what was the grueling process you would do to find? Or is the grueling process? Or what is, which is actually still, yeah, what still is the state of the art in terms of drugs which have made it to the clinic? Yeah, Josh, our CEO, likes to say that our biggest competitor is the mouse. So, like, or nature in certain ways. So, like, traditionally, these, these types of, like, drug-like molecules were either discovered in, like, these immunization campaigns. So, like, you literally will just, like, infect a mouse with a disease and see what antibodies it makes to try to, like, combat that.

13:23Other ways of doing this is, like, super large yeast displays, so on. So, you might, like, start with, hey, I really like this framework. And how am I going to, like, figure out the right loops to design to bind this target? I'm just going to try as much as I possibly can and just, like, literally search for a needle in a haystack. And this would be, like, on the order of, like, at least billions of potential molecules that you're screening against this one target. And in that case, you might, like, end up with, you know, one, two, maybe, like, a dozen potential hits to this target. You actually, you don't know much about those hits. All you know is that they kind of, like, stick to the target.

13:56You don't know necessarily where or, like, if they're even necessarily drug-like. Like, I think, like, one big separator of chai and, like, a thing that definitely our partners like to see is, like, you can be really intentional with how you want to do this design process. You can say, I want to bind this target in this particular area. You can even go back and look to the designs after. Like, we validated that are designs. So, you can go back and look and say, like, is this antibody engaging the target in the way that I expect? Do I think this will actually have the therapeutic effect that I'm going after? One of the cool things about knowing that you have the right binding pose is that you can now also design selectivity into that.

14:32Does your platform have some technique for doing selectivity? Yeah, there's a nice mix of ideas that went both into the modeling side and especially on the product side for dealing with selectivity and cross-reactivity. So, in some cases, you want your molecule to bind one target and avoid another one. So, you might have, like, healthy variants of protein and, like, disease variant of protein. You want to avoid this disease variant or you might have some other similar protein that's, like, not actually harmful in your body that you don't want to just, like, artificially block.

15:03So, I think, like, on the modeling side, yeah, we've come up with ways of doing that. But I think it's even more interesting on the product side to, like, how do you enable customers go through or partners to go through and, like, actually intentionally design for these things? Yeah, and maybe to, like, back up and define cross-reactivity, right? Like, it turns out when you're developing a drug, you're not necessarily going straight to injecting that into a human, right? Like, you might want to put it in monkeys first, for example. And the monkey might have a maybe mostly similar but slightly different variant of it. And so, your drug, you know, not only needs to bind to the human variants but also the monkey variant, right?

15:36And so, you know, the way we've tried to model the models and the product is to kind of let you account for those very general cases where you say, hey, I'm trying to design something that can bind to both of these things so that I can actually go and develop the drug. Let me actually identify maybe the region that's conserved and then target conserved means, you know, doesn't change much between the two and target that exact region. And then, you know, similarly with selectivity, right? Maybe you might want to – there's a very similar protein in the human that if you accidentally bind that one, that's very bad.

16:07And you only want to bind the target protein. And, you know, that's why a lot of drugs, right, you know, fail or are toxic or have, you know, really bad side effects, right? And so, it's kind of – you're kind of having this, like, combinatorial problem of, like, you know, bind only these things and avoid only these. And I think what's been really exciting with some of the progress recently has been, like, a lot of the improvements we've been able to make on the level of specificity we can get to with those models. So, you're not only designing the bind here but you're also making sure that it doesn't bind to another thing.

16:39Exactly. And so, other ways – like, CAR-Ts have tried to tackle this by having some molecular or some sort of signaling pathway that says if I bind – I only fire if I bind, this one binds and this one doesn't bind. But you're saying you just design an antibody that actually only will bind to the thing that you care about. We're getting to the point where in some cases – I mean, it's nuanced, right? But in some cases, you can actually try that. Okay, that's amazing, yeah. So, you're saying you essentially call it counter screen or you have – in part of your platform, you can know reliably counter screen against, like, a large, diverse set of proteins, which might be issues for downstream.

17:16I would say the framing is more you can be very specific about what you care about binding versus what you care about avoiding. But I think, you know, for example, like, a lot of the money that we're raising now will let us train bigger models that can maybe be even more general and start to account for even more things at the same time, right?

The History of Chai Models

17:31Maybe we should back up. Let's talk about – so, the history of the CHI, you know, series of models. Well, why don't you tell the story? We started CHI around two and a half years ago. The first couple months, we're like, all right, we're going to work on protein design. And we're working on this. We're making some progress. We're like, oh, it's pretty interesting. Like, we had some ideas and models. And then kind of, like, that was right when ALFOLD3 came out. And we were – we'd, like, been talking about, like, man, we really need, like, an MSA pipeline.

18:01We need, like, all of this infrastructure built up. MSA is – Multiple sequence alignment pipeline. Why is this just – we've covered this before, but what is a MSA, like, in two sentences, and why is it important? So, if you want to predict the structure of a protein, it might be really useful to see a bunch of very similar protein sequences. And what those protein sequences that are really similar tell you is, like, kind of what positions, like, which amino acids end up being conserved across many variants of this protein. And if you see, like, high levels of conservation or, like, kind of high levels of mutation, like correlated mutations, it typically gives you some indication that these amino acids are close in 3D space.

18:36So, you kind of have this, like, 2D view of a protein, which can then be used to help you predict this 3D structure. So, you're learning from evolution what was conserved because the things that weren't conserved probably broke the protein and something died or didn't make it. Exactly right. Yeah. Yeah. It's pretty remarkable that this works, honestly. Yeah. One of my favorite, like, bio facts here. Yeah. So, we were, like, kind of thinking, like, oh, man, it'd be nice to have, like, a lot of infra and whatever. So, ALFOLD3 came out. We're like, hey, we should, like, open source this model. We should just, like, you know, bunker down, build all the infra that we need.

19:10I think, like, this will pay back, like, in the long term for sure of just, like, as a forcing function to, like, be where we are and also just, like, to contribute to the community as a whole. So, it's interesting that you chose, okay, this, we're actually, what we're doing here, we're building a model, but what we're really doing is learning how to build the infrastructure. Is that kind of what you're saying? Yeah, that's exactly right. And, like, I had built a lot of, like, similar infrastructure in my PhD, but not at a production level for a company. So, like, at that point, I think we were five people. So, there are five of us at China. We're like, all right, this is our forcing function.

19:42We have, like, a clear goal to work towards. It's, like, very direct. Let's get this thing going and see how fast we can do it. You guys were, at this time, sitting in the OpenAI offices? We were sitting in the OpenAI offices, yeah, in the mission. Right. So, like, what's the backstory on that? It's really interesting. Two of our other co-founders, Josh and Jack, had a relationship with some of the OpenAI people. Actually, OpenAI co-led our seed round. So, we were, like, kind of thinking, like, all right, should we get an office while we're only five people? And it turned out, like, that office was mostly vacant. So, we got to sit in on the, like, in the OpenAI offices for a while.

20:15Taiwan, built it, open source, learned about infrastructure. Yeah. So, then after that, like, we really set the sights down on protein design. And worth pointing out, CHI-1 was a structure prediction model, right? So, you have the sequence, what is the structure that it folds to? And then that was it, CHI-2. Yeah. CHI-1's finished. One other crazy story there. Let's see if we can actually share this. But this is a hilarious one. So, like, we were, like, oh, man, we really want to be the first to put this out. And we were, like, okay, we're one week out.

20:46We're, like, the model's, like, almost done training. We're, like, should we build a web server? And then we're, like, oh, yeah, maybe not. And then, like, we ended up spinning up, like, this whole web server. So, like, people can use it. Like, rather than just, like, download the Git repo. It's kind of annoying, especially for biologists. And, like, we actually wanted people to use this. So, like, let's spin up a web server. Let's get the technical report out, all this stuff. So, we ended up, like, we were up for, like, 40 hours straight. Just, like, getting the paper over the line. Getting, like, all the last things done on the web server. And then Josh was interviewing with, like, Bloomberg TV or something that morning.

21:18And we've been up for, like, 48 hours straight. So, Josh, like, runs into a room to do this interview on Bloomberg TV. And, like, I think it was, like, 7 in the morning. Everyone's in the office. Like, we didn't, like, want to be seen or whatever. And, like, the interviewer's like, oh, like, interesting company. Doesn't look like there are any employees here. But, yeah, it was a really fun time. I think, like, the early startup days were just super fun. So, yeah, after that, we kind of set our sights on design. And really what we were thinking is, like, we kind of always had antibodies in mind.

21:49We thought of this as, like, the most tractable problem. The nice thing with proteins is you have this beautiful sequence representation. There's already a lot of research been done in, like, how do you autoregressively generate sequences? How do you, like, the sequence generation problem is well studied. So, we were thinking, like, what's a nice, like, area to apply sequence generation to in the biospace? And it's pretty natural to do, like, linear sequences of amino acids. So, we started working on design. A unique thing about CHI is, like, we're not, like, we're designing antibodies.

22:19Like, we're an antibody company. Like, we don't really, like, pigeonhole ourselves into, like, one therapeutic area. So, we, like, try to really tackle this problem very generally. So, we were thinking, like, can we design many proteins? Can we design antibodies? Can we scaffold regular complexes? So, like, really just take a holistic view on, like, how do you design proteins in general? And that eventually led to the CHI-2 model. So, that was our, like, first flagship design model. And that's where the CHI-2 paper and, like, our bold target discovery project came in.

22:51So, we designed antibodies to 50 targets for that paper. Got binders to about half of them with, I think, on average around a 20% hit rate for binding. And then afterwards started working on CHI-3. So, that's our latest series of model. But I'll break there. Before we talk about CHI-3, can you tell us about, especially for listeners that may not be familiar with structure prediction models, what does the model look like?

Anatomy of Chai 1

23:14How does it work in general?

Anatomy of Chai 1

23:15So, let's take a look at CHI-1. CHI-1 has this, like, roughly a tokenizer, a transformer, something that looks like a language model, and then something that kind of looks like an image diffusion model. And they're all just, like, stitched together. The tokenizer is, like, not your kind of typical, like, words-of-exile tokenizer. This is, like, I have a bunch of atoms in a molecule, and now I want to, like, pull those into what I would call tokens for my, like, LLM-looking trunk. And then that conditions this, like, kind of big diffusion model, which will then emit the image, which is some 3D structure.

23:48So, is it atoms or is it amino acids that are the input? It's an interesting question as well. So, we have, like, all these different input tracks. So, like, one thing about biology is the data is inherently multimodality in a sense. You have this, like, you know, kind of token sequence representation. Each of these tokens has, like, a set of atoms that kind of dangles off. And then you also have, you know, some properties of the different atoms. Like, an atom might have, like, a different charge. It might have a different element type. So, like, periodic table of atoms.

24:20And then these kind of all get bunched together into tokens. Once tokenized, you can kind of process this in very standard ways. But then, ultimately, you have to get back to these, like, 3D coordinates. So, like, in order to predict the structure, this is just some 3D object. And that object goes through, or, like, to emit that object, you go through what looks like an image diffusion model, where you kind of go back from tokens back to the atom representation. I see. So, the tokens go in, the transformer establishes the relationship between the different tokens,

24:52and then the diffusion model turns that latent representation into a 3D structure. That's exactly right. Yeah. Okay, great. So, that's CHI 2? That was CHI 1. Okay, CHI. So, CHI 1 folding model. Yeah. It's, like, and, like, all this bio stuff, it sounds, like, kind of scary. Like, atoms, tokens, amino acids. Like, at the end of the day, my background, personally, is, like, theoretical computer science. So, that's what I spent, like, all of my earlier years doing, transitioned to this, like, pretty late in my PhD.

25:24But I think, like, the background that you need is really similar to the background that you'd need for, like, any other field of machine learning. There are all these domain-specific things that you learn about. But, like, one analogy or, like, anecdote I like to say is people think you can't work on, like, AI bio unless you're a biologist. But it's kind of like you can't work on, like, video models unless you're, like, a director or something. Like, there are all these, like, super domain-specific things, like, oh, yeah, to understand, like, lighting in a video, things like that. But at the end of the day, these are just machine learning problems. And, like, they're all solved the same way.

25:55Okay.

Chai 2 and Design Capabilities

25:56So, then CHI 2, there's a jump in capability as well as an architectural change, right? Yeah. What we've disclosed about CHI 2 is, like, it is an all-atom diffusion model. So, we're trying to predict, like, you know, atoms in 3D space still. But we're doing it in such a way that, like, the model actually has the ability to, like, design atoms, place them, decide which atoms actually are there. So, like, one way to represent an amino acid, like a protein token, is by, like, which atoms are present. So, in the CHI 2 case, we were just predicting, like, all right, show the model, let the model just kind of pick what atoms it wants to keep,

26:30and then map that back to what amino acids there are. What are you able to do with CHI 2 that you can't do with CHI 1? Is it just, like, better? Or are there new capabilities it brings? It's design, right? So, CHI 1 lets you say, hey, I know the sequence of amino acids, right, that text string, and I know the structure. That you would get it from, like, the genome arm. Right. Exactly. CHI 2 says, okay, I have a target structure, right, that I want to design a binder to. CHI 2 will then generate, you know, candidate molecules, candidate medicines that bind to that target.

27:03And so, this is kind of a design model or design family of models. And I think that's where you really cross the threshold of usefulness, right? Like, I mean, CHI 1, alpha, very useful because you can, you know, you can at least intuit and reason about the structure and see what you're looking at. But, you know, the ultimate goal here is to design medicines, right, and design new molecules. And I think CHI 2 really crossed the threshold of performance for doing that with antibodies a year ago. One analogy here would be, like, kind of, like, back to, like, the image domain. So, like, CHI 1 would be, like, you know, there is a cat in this image.

27:35Like, thanks, CHI 1. And CHI 2 is, like, I'll show you a background, maybe. Like, I'll prompt you with some, like, image information. Like, hey, put a cat in a field. And CHI 2 will actually just, like, give you back an image of a cat in a field. And you're, like, that's a good-looking image. Or it's not. You might have some other model which kind of ranks the image. But fundamentally, it's the generative problem. So, taking that analogy a step further, it's maybe more like you showed a background. And then it generates, there is a cat. And then it generates an image of the cat at the same time. And it makes sense that there is a cat in this field and also that the cat works in the image.

28:10So, there's a, it is a, it's an interesting problem because you have to generate two things at the same time. Exactly. Both the sequence and the structure. Then you, if you, I don't know if you can, but could you talk a bit about, like, how that works? Like, how do you do that? So, you code, you co-design the sequence in a way that the structure also fits and makes sense. One way to think about it is kind of like the classic way of doing this. Let's talk about both. And structure prediction, like, all right, I know the sequence. Like, I can, from that, roughly figure out the 3D shape. And then there's kind of like the inverse folding problem, which is like, given a 3D shape, give me back a sequence that would fold into this.

28:44And now you kind of like need to do both things at the same time. But I think like similar principles apply. Like, you can kind of have the model, like, think a little bit about what should this structure look like. And then you can have some other part of the model thinking about, like, now what sequence would maybe support this? And then, like, a nice thing with diffusion is, like, you can do this pretty slowly and pretty iteratively. So, you can give the model a lot of time to think about, all right, if I change the structure like this, how should the sequence change? And you can kind of just play this back and forth and back and forth. And eventually, it ends up kind of converging on something that's self-consistent.

29:14It's almost like an EM algorithm. Yeah, exactly. So, you have this model now, Chai2, which is able to predict or to sample a structure and a sequence which generates that structure. And just because you can generate a structure, like, doesn't necessarily mean it's necessarily accurate enough to do something. So, do you have other scaffolding on top of that? Are there additional problems? Like, are you one-shotting these things or are you, you know, needing to generate thousands of them and then you have a ranking or scoring?

29:46Or, you know, how, like, just having a candidate is maybe, let's say, not enough. So, what do you do once you sample a structure? Traditionally, what's done in, like, when co-design and, like, protein structure design, like, started to become a thing, we're, like, kind of at a loss for metrics. This is, like, how do you know that your protein, like, you designed some, like, sequencing structure. Like, how do I know that this is legit or not? Like, I can tell you it's, like, anything. It's, like, totally out of domain now, right? Exactly. Almost by definition. Yeah, yeah, yeah. And, like, as a human, you can look at this thing and be, like, I don't know.

30:18It checks out. Like, even biologists are, like, I have no idea if this thing actually folds. Like, maybe some of it looks right. Even our biologists are surprised, by the way, with, like, some of our designs that, like, do end up working. What was done at the time is, like, we kind of came up with a bunch of metrics. And, like, alpha fold, it really is what enabled this. So, you'd take the sequence that you predicted. You'd run that through some, like, totally, like, distinct structure prediction method. So, this is completely independent of your model. And you say, if an independent model thinks that this sequence folds to a similar structure,

30:49then it has a higher likelihood of being correct than, like, you know, just whatever the prior likelihood would be. So, you can take your sequence now and you can measure, like, how consistent is this structure prediction method with the structure that you actually predicted for that sequence. So, you can now compare your design to an independent model structure prediction. And that became, like, a really good way of gaining conviction that your design model was correct. And people kind of, like, game these benchmarks for a while and kept pushing, pushing, pushing. It turns out, like, it's easy to get self-consistency, consistent design of structures.

31:20If all of your proteins look identical, there are a lot of problems that this creates. But then people started adding more and more on top of this. Yeah, that's an interesting point that I think some people have acknowledged in the community. So, how did you solve that? Yeah, you can see that if you sort of use your Oracle and also your sampler at the same time, you eventually will converge. What do you do to stop that or to convince yourselves that you're doing something valuable? One of the nice things about structure prediction methods is that usually you have some calibration

31:50and kind of how confident the model is in its prediction. It turns out these models, they can give you a pretty well-calibrated confidence prediction. So, rather than just say, this is what I think the structure looks like, I'll say, this is what I think the structure looks like. And kind of, like, here are the parts that I'm not really certain about. And you can kind of aggregate this down to, like, a single scalar. And typically what people do is they'll look at, like, okay, like, not only how self-consistent am I, how much does this independent folding model even like the structure that it output? So, that was one way of early on, I'd say, to, like, just gain confidence.

32:24And then, like, another thing that people often do is they'll look at, like, the diversity of their generations. Because, again, you could have a model that's perfectly consistent, gives you great confidence predictions back, might be the same structure every time, like, same sequence every time. So, you also want to see, like, okay, how diverse are the solutions? How many of these new problems can I solve, in a sense? If I had a whole lot of money to validate, how would you do that? Can I go and, you know, do cryo-EM or something like that and try to figure out the structure, you know, sort of get some ground truth on that?

32:56It's more that the feedback loop is really slow. So, you can validate a few structures like this, but it might take months. And it's just not, like, a very scalable direction. So, I think that's, like, a problem for the field as a whole. And I think people are spending a lot of time, even, like, especially at CHI, I think thinking about how do we validate these problems at, like, bigger scale? How do we, you know, basically increase the throughput of our validation or increase the cycle time? Because if you're waiting months to figure out, hey, was my model correct? Like, it's just, it's hard to iterate in a research environment that way. The good news is that this is getting a lot better, right?

33:27Like, there's a whole network now of wet labs that you can work with that will run, you know, these assays, these experiments, and tell you things about, say, you know, does your protein that you came up with bind to its target well? And so, you know, thankfully, we're not at years, right? We're down to, like, weeks, which, you know, not as fast as, like, LLM land, where you can just, you know, scale up and eval with and throw more compute and get results back in hours. But, you know, fast enough to where you can start to recursively self-improve. And, you know, I think we also spend a lot of time, like, you know, figuring out what are the metrics that we can compute, you know, in silico,

33:58like on the computer that are predictive, perhaps, of lab success. But, you know, your question about cryo-EM, yeah. I mean, also, you kind of have to measure the structure. And as you know, that's, like, so expensive because you have to kind of freeze the protein and shoot these electron beams at it and see how they bounce off. I remember there's, like, this really funny anecdote. We'll see if I can share it. But, like, the, you know, the paper in Chi-2, we actually, you know, did that. We took some of the, you know, the proteins that the model predicted and ran cryo-EM, and we got the results back. And we were like, wait, the results look wrong because we had overlaid the kind of prediction

34:32over the point, the electron cloud, the point cloud. And we didn't see any difference. And point being, like, we're getting the point now where these structure prediction models are within, you know, a few angstroms or less of the actual atomic positions that you validate. And in this case, it was a 0.33 angstrom error, which is one-third the width of an atom. And we were like, this, like, can't even be right. Like, clearly, they just sent us back the wrong design. They just sent us back our design. Yeah, exactly. Did you check for data leakage? Yeah, in this case, like, there were no... So, like, we actually chose these targets specifically, like, to have no known antibody binder.

35:08So, like, if we did get a hit, like, it was definitely the first antibody hit to this target. Yeah. I think that's one of the things I didn't realize about biology was, like, just how much of it is literally feeling around in the dark. And that's not even a metaphor. You literally can't see, like, how these things look, right? So, structure models are so huge because now you can, okay, you can actually predict within an atom, you know, how these things look. And that enables you to then do things like CHI 2 with the design models. This, to me, is AI for science is one of the cornerstone problems, right?

35:38Yeah. You don't know, you fundamentally don't even know how to measure your problem. Yeah. In a lot of cases. So, it's very difficult to validate. Yeah. So, you're getting these sub-Angstrom predictions with CHI 2, CHI 3, what, why CHI 3, what's

Chai 3 and Binding Affinity

35:54better or what? Yeah. I think, like, with CHI 3, so, like, honestly, like, there was a CHI 2, there's a CHI 2.5, there was a CHI 2.7, there was eventually a CHI 3. And, like, each time we saw better and better performance. And I think, like, the main thing with CHI 3 is, like, we look at CHI 2 and, like, we look at the targets it can solve. There was, like, a lot of internal discussion after CHI 2. Like, hey, we made, like, successful molecules, binders, to half of these 50 targets. What about the other 25? You know, what can we do to make those better?

36:24And then, like, you know, we were split. We're like, all right, should we, like, study these targets that we missed and, like, figure out exactly, like, are there properties of these that we can look at? Or should we just bet on the models? Like, will the models just get there if we put more time into, like, you know, just be bitter less and pill in that sense and just really bet on the models getting better? And we definitely took the latter approach. Like, we bet on the models getting better and we just pushed as hard as we could on that front. So, you're scaling up the model, the data, whatever, to just build more accurate models?

36:56Yeah. Is it accuracy? Is that the main thing? Is it binding affinity? What do we... So, I think binding affinity is a big one. Like, you can't just bind weekly. In order for this to be, like, a useful tool, especially for our partners, we need to start producing molecules that are, like, at or very close to therapeutic grade, which means, like, they have to bind really tight. They also have to be developable. They have to have, like, all of these nice therapeutic properties. And developability, I think that we talked about... He mentioned CHI 2.5, right? Which we released, like, a few months after CHI 2. There was a study we did on the developability of the molecule, which, you know, for the audience,

37:30like, obviously, the molecule has to stick good and stick tightly, but, you know, there are these other properties you care about, and to use the non-biological terms, right? Is it safe? Is it stable? Is it easy to manufacture? Does it, you know, self-aggregate? And we've been pleasantly surprised at, you know, how much we've been able to climb and push the performance in those areas. It seems like one of the reasons that you want to do antibodies is because of the developability. Yeah, you get a lot for free there, right, with that antibody framework. Yeah. It's interesting.

38:00I mean, to me, there are many structure prediction molecules out there, I mean, models out there. I feel like it's these other ancillary factors, actually, that are going to probably be the most impactful in the usefulness of a product. Yeah, right. Yeah, absolutely. The nice thing about structure prediction is there is a ground truth that you can compare against. For design, you don't really have that. You're like, here's some new, like, disease molecule. Give me a binder for that. And, like, if you want to know if this thing really binds, you have to send it off to the

38:33lab and wait a while. For structure prediction, you can be like, all right, the model hasn't seen this sequence before. It's never seen anything close. Does it actually, like, fold up into the correct shape? And we can just kind of hold that out of the data set and check. So, I think I've always thought of structure prediction as this really nice speed run kind of benchmark to, like, validate ideas on. Right. Sorry, I didn't mean to say, I meant, you know, sort of structural models in general. Yeah. But, yes, exactly.

Building the Design Suite

38:57So, maybe we can talk a little bit more about start getting into the product side of things. Thank you for coming. I actually, I mean, like I said, I really think this goes throughout not only for, you know, sort of structural models like this, but also virtual cell and whatever. It's really all the other stuff around the direct development process that is going to have the biggest impact. So, can you talk a little bit about that? Yeah, I think that's actually a good thing to talk about after Chai 2 because I think Chai

39:292 is where it started to get really fun from a product perspective, right? I think with Chai 2, we crossed the threshold of usefulness where after we, you know, released that paper, we had a lot of, you know, you know, pharmas and biotechs approach us and say, hey, this model might be able to do some stuff for us. Like, can we use it? And then we're like, oh, man, like, we should build a product, right? We should build something to let you use that model. And that's right around when I joined. And there was sort of this, you know, mad, mad build out to both, you know, build the product, which we can talk about the shape of, and also go and secure the compute, actually,

40:00so we can go and serve those models to our partners. And, you know, I think another third piece there that was really interesting is, you know, around security and IP, right? I think we want to be a very neutral platform that anyone can design medicines on. But as you guys know, like, pharma is this notoriously IP-sensitive industry, right? And I think when I joined, a lot of people told me this can't be done. Like, they're not going to put their data in a platform and, like, have all their new medicines be generating out of it. And having a bit of a background in security helped a bit, whereas, like, no, actually,

40:31if you, like, just are really aggressive about how you, like, segment data and set up, like, single tenancy where you're, like, almost deploying a separate version or a separate account in the product per customer, you can actually, like, build a platform and then go and ship it to them. And so, you know, through the summer of last year, we started doing that, right? And, you know, we'd been working with, you know, or talking to Eli Lilly, and, you know, they were, you know, one of the first partners to really work with us closely on that. Kind of, you know, it made that V1 of that design suite, right, that you can use to engineer

41:05some of those molecules on. And, you know, maybe it's worth talking a bit about that design suite, right? I think, you know, I think we have these really, really powerful models now, right, that can do, like, all of these crazy things if you condition them in the right way, if you kind of give them the right context about, you know, the structure that you're going after or maybe the constraints around the model, right? Like, hey, I want to design an antibody that hits this GPCR protein, but, you know, doesn't collide with the cell membrane and also targets the specific epitope on that as well.

41:36And, you know, we looked at it and we're like, I guess we could put a chatbot around it. That'd be, like, really easy to talk to. But, like, really, like, you're trying to build something almost very visual, right? And you can finally build something really visual with some of these structure prediction models. And so if you kind of look at the Chai product, it looks a lot less like a, you know, a chat GPT and a lot more like Autodesk or SolidWorks or Figma, you know, if you've used those things where you can kind of load up your molecule. There's this almost, like, Photoshop-esque, like, design suite. You have this equivalent of a paint tool to kind of paint your epitope.

42:08You have this equivalent of a content-aware fill tool to kind of get your binders generated from Chai. You, of course, have a lot of the scientific analysis and plotting and whatever to understand the results of the models. But we've just been surprised at, like, how much complexity is actually just in, like, doing that right so that you kind of don't shoot yourself in the foot when then you're then prompting these models to give you binders. So are you sitting with people who are designing these antibodies, you know, and feel like they're complaining to you or whatever?

42:38How do you convince med chemists to use your tools? Because med chemists hate AI tools. Like, they're notorious, like, I don't want to touch this thing. Or, like, I don't understand it. And they will not touch things, which they do not understand. Well, it helps a lot to have the models working really well, right? So when we, you know, when we had the results of Chai 2 and Chai 2.5, I think, you know, that's enough of an activation energy where, you know, pharma companies and scientists within these companies are like, oh, let's try it. Actually, can you, Chai, can you guys just try running the model against a few of these

43:11targets and let's look at the results? And then we do that and the results are good. And they're like, OK, let me let me try to get on that product and let me try to use it. No, I think pharma is, like, incredibly pragmatic, actually. Like, I've been very impressed with everyone that we've we've worked with so far. They're they're very, like I was saying, pragmatic about this. And they're like, they're they're willing to be proven wrong. And like, I actually don't blame them for not trusting the models. Like, I have used these models and like, yeah, they've like rightly so. Like, I am pretty skeptical when I like see any release. I always have been.

43:42So like, you really just like need to show them the proof and like they can give you this target that they are interested in or maybe it's more of something they've worked on in the past. They probably don't want to like share IP right out of the gate, but they can be like, hey, you know, I've had trouble with this particular target in the past. Let's see how you guys can do on this. And then once you show them the proof, they like almost overwhelmingly are willing to accept that. I come from a cybersecurity background or, you know, have worked on security products before. And those were dark, dark years because you spend a lot of your time actually selling to people who are surprisingly not that technical.

44:14You think cybersecurity people are very technical. In many cases, they are not. And it is this kind of like uphill enterprise slog to this very unsophisticated customer. I think we've been just pleasantly surprised or I have by just how much I enjoy working with our partners and our customers. You know, these are scientists who have been spending, you know, five, 10, 20 years of their life working on one target, right, often in some cases. And they've studied everything about it. You know, they're very sophisticated. They're very smart, right? You know, getting to collaborate with them is just a goldmine.

44:45And we learn a lot about how to make the product better. You know, this anecdote, we, you know, a few months ago, we were actually showing some of the results that we, from a target data with a pharma partnership. And one of the scientists in the room, like, started tearing up and crying. Oh, wow. You really hit the head with that one. And she was like, and we were like, what's wrong? And she's like, no, I've just been, I've literally spent 10 years trying to get an initial binder to this thing. And you guys were able to help me do it. Oh, that's awesome.

45:15And, you know, that feels really special. To answer your question, you know, we, you know, there's, of course, the teams of scientists and computational biologists that we're working with within, you know, each of our partnerships. There's also the people we have within the building, right? So we, I think one of the things that I really appreciate about Chai is how cross-disciplinary it is. Like, you know, we have people who are maybe engineering experts and less bio experts like myself. We have great, you know, AI scientists, but, or ML scientists, but we also have a bunch of scientists that we work with and have joined Chai to sort of help us both, you know, test

45:49the limits of the models, right? See what is Chai 2 actually capable of? What targets can it do? What can't it? And form some of the research direction there. I want to add to that. Like, in like the Chai 2 days, like we kind of started with like a bunch of engineers and people who have like AI bio experience. We didn't have a hardcore lab scientist and like one of our first hires on that realm was Nathan Rollins, who I think he started working in the Baker lab at 14, graduated from Harvard at like 18 and got his PhD by like 21 or something like this in the Marks lab.

46:21And he was like super skeptical about Chai at first. And then, you know, the results started to come in. He's like, okay, this is, this is kind of interesting. Like this could work. And then like once the Chai 2 results came back, he was like, I need to bulletproof this. Like nobody celebrate yet. Like all this. So I think like it's, it's been really nice to have that level of rigor and to just have people who have really like, they've spent the time in the lab, they've designed proteins themselves. They've literally, in the case of like Andy, led several therapeutic programs, brought drugs to the clinic themselves.

46:51And like we have all these people internally at Chai just like using the products and like really battle testing that. So if you don't have your own platforms, right? I mean, so you don't have your own programs, right? You're a pure platform, your partnership model, right? Yeah. How do you battle test something if you basically aren't, you don't have a use case where you have to continuously push it forward? Or if you are just pushing things forward, when you just end up with your own candidates, if you're successful, and then what do you do about that? I mean, we have benchmarks of our own internal cases, right? You know, there's a set of targets that, you know, have our known therapeutics, right?

47:24That have known therapeutics against them. There's a set of targets that we pick to sort of push ourselves, right? And so we're constantly refining that set and adding to it. And that's what that internal science team that we have helps with, right? Is expanding that and almost running the experiments to try to get initial binders there. We don't care about going and developing those drugs. Like we just do that in service of validating and making our models better. And then, of course, there's a loop with our partners too. Would you consider yourself hit discovery or are you, do you, I guess, using some jargon, hit to lead, lead optimization?

47:55Like where do you live in this? And, you know, hit discovery might be like one part of it, which you can do hit discovery. But the later, the other parts of this are, I think, oftentimes much more bespoke and kind of special. I mean, how do you balance that? And it seems much more difficult to me to be general than it does to solve general lead optimization than it does to solve like hit discovery. I think ideally, like we really want to be able to, rather than think of this as a bunch of stages, I think part of the reason why we think of it that way is because the initial

48:28molecules are usually like not good enough to be drugs. And like really, like we're kind of at the inflection point now. We're really seeing this internally at CHI where the models are getting pretty close to like producing molecules that could eventually, or like are very close to drugs. Um, so we, we try not to make too much of a distinction in between, okay, hit discovery, lead optimization, all of the different parts of this kind of preclinical pipeline are, are like, you know, the, the light, the North star is to just really produce drug-like molecules

48:59straight out of the models. Of course, this is going to be hard and like, they're going to be like tons of roadblocks and like, you need to be able to like actually prompt the model to do this. You need the whole RL stack to like learn different properties, things along those lines. Uh, but I think it's very achievable. Yeah. And I think to add to that, right. Yeah. The site notion of target discovery and hit discovery and optimization where each of these has a gate and takes a few months to a few years is this very like waterfall model, right? Where the cost of trying things and getting things early is very expensive. But I think to what Matt's saying, right, if you start to get in a regime where you can

49:31have models give you really promising candidates, you can start to make that look a lot more like a loop, right? It's, it's akin to like becoming more agile and software development. Internally, we kind of have two, you know, North stars, right? Um, and it's at first pass, they almost sound like contradictory, but you know, the, um, you know, the North star in research is to start to de novo one shot, you know, better and better and better medicinal candidates that are as close to being ready for, you know, the next phase as possible. Um, but you know, also within product, we, we do want to sort of expand into whatever

50:02these iterative workflows look like, right? Where maybe I get a binder, I get some results from the lab. I'm using that to condition my next run of the model. Um, and I think, you know, they sound contradictory, but I think they're actually not because I think what's going to happen, you know, the research is going to get better at identifying a de novo candidate for like a specific class of drugs, right? Say like anti-agnists, right? Like blocking things, right? A little bit easier, maybe. Okay. We can get to a state where we can one shot pretty good drugs there. But now the next problem is like agonists, right? Like how do you reliably one shot hitting a switch like on a cell, right?

50:35Or by specifics or ADCs, right? And I think, you know, there's kind of this, uh, levels of abstraction that we're going to have to climb with the product as like the models get better. One of the things, uh, I was, I got, I got very existential like a few months ago because I was like, man, all this stuff we're building in the product to like visualize molecules and do this, like maybe I'm just going to have to throw it all away when like Matt ships like Chi 4, right? Um, but you know, I think that's, that's kind of the reality of like building products now, right? You're actually using them less as an end in and of itself.

51:05Like maybe you'd have built software that was supposed to last like 20 years. Now it's supposed to last maybe one year, but it is the bridge to deliver value and kind of enable the research that then gets you to the next thing. And so I'd imagine we're probably going to rewrite our products at higher and higher levels of abstraction, right? Like maybe like right now we have something a little bit more akin to cursor where you're, you know, inspecting the molecule in the same way you're inspecting the code because you really need to verify like the bonds that are forming and the, the, the properties of the things that you're getting. But then, you know, you get to a point where that stuff is solved enough where now the product

51:38is actually just helping you orchestrate these like campaigns of hypotheses, right? Or maybe you have like one target and you're like orchestrating a bunch of different epitope choices or whatever against that. And then maybe you're going up one level of abstraction where you're now doing a whole campaign against all of the targets within a pathway, right? And I think what's really exciting about that is if you, if you have like these really good primitives for structure prediction and binding and design, and you can kind of compose them, then you can start to just like grow into like the outer loop of science, right? And then, you know, maybe the thing runs itself and you start to really get to some really,

52:13really, really cool drugs at the end of it. I actually want to push on what you just said about epitope prediction, because I think a lot of people in the field would argue this might be the much harder problem than finding antibodies and binders. Where do you think that the state of the art is in general and also with regards to Chai in terms of epitope prediction? And like, is this a problem which has a reasonable, solvable time horizon? Oh, and also maybe can you define epitope prediction? I'll think of this at like some different levels. So the most basic level is, okay, I have some disease that I want to target and what proteins are actually responsible there?

52:44Like actually figuring out biologically what's going on, like what should I be targeting in the first place with a drug? Because once you figure that out, it's kind of like a structural biology problem at that point. You're like, all right, this like set of proteins is responsible and like what's going on there? Well, this is interacting with some other protein that it shouldn't be interacting with. And conventionally, you'd just like want to block that interaction or something with anybody. Uh, but that's kind of where these proteins interact and like the type of interactions that you want to disrupt. That's typically like the epitope. It's like the actual site on the protein that you want to block.

53:14This is a ridiculously hard problem. Uh, I'm with you on this. This is like the harder problem. Uh, like just the amount of context that you need and like the global understanding that you need, you need to get in order to like actually figure out what's interacting and how. So, but maybe let's take a few specific cases. Let's think about what about, um, SARS-CoV-3 comes around or the new flu or whatever. What would you do there? I mean, is that something that you think you could actually reasonably tackle? In that case, like, yeah, you could just run a structured prediction model maybe and like see where the model thinks this thing will bind.

53:46Uh, if it's highly confident in that, you might say, okay, here's like the site that we want to block. I think in general, still very hard. And even like structure prediction, it's getting really good. And like a lot of people think alpha-fold-2 like solve structure prediction. Not really. Like alpha-fold-2 got like, I think 11%, the multimer version of this got like 11% of antibody antigen prediction cases correct. That means 90% of the time it's wrong. Yeah. I mean, but, but, uh, alpha-fold-2 solved a certain class of monomeric proteins with MSAs.

54:17Absolutely. Yeah. Yeah. So, I mean, the, and that's the, the MSA, I think might be the key point here because MSAs are sort of the, the, the magic, which makes it all work. It's like a, it's a template in some sense about like what the structure should be. And antibodies almost evolutionarily can't have a template, right? Yeah. Everyone has to have unique antibodies custom to the things that they've experienced over the course of their life. Yeah. So. Right. And just, just to clarify, I had to understand this myself. So maybe I can help the listeners who aren't familiar. An antibody, the whole point of an antibody is it, it can, it could be used by the immune

54:50system to identify new things that it hasn't, the body hasn't encountered before. So the design of antibodies as opposed to other types of proteins is to, the system is designed so that you can quickly recombine different components of it in order to, um, match, uh, proteins that are from unknown pathogens more or less. And so this is why you, it's not conserved in evolution the way that other parts, other proteins are. Yeah. So, so like back to the, the epitope prediction problem.

55:21Um, I, I think it's still hard. I think like there, there are a lot of cases that, that are maybe tractable, but I think in general, like if you want to discover this for a new target, uh, still, still a really difficult problem. Maybe virtual cell would be like the closest thing to the state of the art there, but that's still, still a ways out.

The Economics of Drug Discovery

55:37I wanted to dig in a little bit on the product because I, I, there's something I don't understand about the economics of basically all, all the structural stuff that's happening right now. And obviously a lot of people think it's very, very valuable. So there's, you know, I'm not grokking something, but when you look at the cost of developing an antibody, you know, it maybe is a couple million dollars, right? When you go from, you, you, you've identified a target somehow, and then you say, okay, I need an antibody to match this.

56:09And then I have to sort of optimize it in various ways. And then maybe I try it in, I mean, with antibodies, you go to the animal typically faster. If you look at how much does it cost to drink, bring, if you like are pressing it and pick the right target and the right technology to get all the way to drug, it might be half a billion. And typically that $2.6 billion number is amortized over all the failures as well. So if you look at just the cost of that one success, depending on the, the disease, maybe less, but you know, half a billion might be a good median number or something.

56:42So you're, you're saving like a couple million dollars in a half billion dollar campaign. So why is this so attractive? I would maybe challenge the premise a bit, like in a few ways, right? Like, okay, sure, if you're trying to get an antibody for like a very simple kind of target, like maybe, right? But I think what we've been most excited by is our partners using antibodies and, you know, more sophisticated ways, right? Like in, for example, in CHI 2, we showed like GPCR agonist activity, right? Where you can really hit the switch on a, you know, on a cell doorbell protein, so to speak,

57:15right? In a very precise way. Very, very, very hard to do that with antibodies if you can't be that precise, right? So you're unlocking a new capability. Yes, right. I would think about it as less like, oh, I'm taking the existing drugs that I can do and making them faster. I mean, there is some of that too, right? But it's like, no, they're just like, hey, how do you go after like better targets, right? That are, you know, maybe more precise, more effective, right? I see. I think like also on top of that too is like there are drug modalities that you just can't discover with immunization. Like you're not going to design your like crazy, multi-specific, warheaded, super intense

57:48formats. These are really things where you kind of have to design these from first principles. Even just with bi-specifics in particular, like both arms need to now bind different targets and you've kind of like have this multiplicative effect on your binding rate. So like if you have a one in a billion chance of finding a binder in arm one and a one in a billion chance in arm two, you're not, this isn't going to work for the traditional lab that. Exactly. I think the other thing I'd think about is, right, you're not just helping your partner with maybe one drug, right? There might be a portfolio of targets that are, they're going after a portfolio of drugs

58:22that they're trying to make. And the nice thing about the platform approach rather than that we are developing individual drugs is we can sort of scale with them as they pursue more targets in addition to more ambitious targets. So it lets you concentrate your learning in a subdomain of that and so that you, everybody benefits from that. Exactly. That's the, but I, okay, so I didn't, so what are some of these capabilities you mentioned a few? Are there more that are really interesting that you guys are chasing? Yes. I mean, and we talked about like, you know, cross reactivity.

58:53We talked about selectivity. We talked about some of these like really interesting additional modalities with bi-specifics, right? There's a set of things that, you know, our partners have been asking us for that we've been working on that I can't get too into because then that starts to reveal some of the targets that they're going after. But I think the point being you can just, once you get precise, like you can start to do some really, really cool drugs. It's a new technology, right? So like technology in pharma means like how do you deliver your therapeutic?

59:24And so this is maybe a kind of thinking about like CAR-T is a technology, right? And so this is maybe a new technology in the sense that you can have these highly, highly engineered. Right. And that comes, you know, from the mission of the company is to really turn, you know, drug discovery from a scientific experiment to an engineering discipline, right? How do you sort of get to the precision engineering phase for biology where you can start with, you know, almost declaratively define the thing you're trying to get and have the model fill

59:55in the gaps and get you that. So what is the biggest blocker from going from science to engineering?

Engineering and Infrastructure

1:00:01Oh man, there's so many things. Like that's the thing about, you know, micro-heterogenicity. Yeah.

1:00:08Micro-heterogenicity. What is that? I don't even want to talk about this. Like the amount of headaches. Too late. You already went. Yeah, you're getting it out. Okay, so like just like when you're actually parsing, like first of all, file formats for biologists, like I just, they just don't care. There's like no standardized, there are standardized file formats. Are they the best? I don't really know. But there's like also just like a lot of information that you want to pack. And I have this structure. Here are the people who solved it. This is the method I used to solve it. There's like a lot of stuff going on. And then depending on the method that you used to actually figure out what this 3D structure

1:00:39is, you might have like multiple copies of that structure. Part of it might not have really been resolved. You're like, it could be here. It could be there. I'm just going to give you like both options. So like the actual just parsing problem on the engineering side of like working with this type of data is like really difficult. This seems like something that LLMs can excel at though. They don't know all the edge cases often, right? This is more back to just like a simplicity approach. Like LLMs are very good. I will absolutely give you that. Then you're thinking about like, do I really want to like, should this function have 20 special cases or should we be like really principled in how we approach this?

1:01:12And should we be, I guess, more of a opinionated, opinionated, yes. Like how opinionated should we be in how we do this? We want a strategy that's easy enough for humans to understand. And like when we're reading through the code base, we really need to know what's going on here. What are the potential problems? And like sometimes that just comes down to looking at examples. But then I think, okay, once you've kind of figured out all the infra work and how you get data into the models, there's then like scaling the model. There's then scaling the infrastructure around the model to train bigger and bigger versions of this.

1:01:42And that's like a lot of work that Neil and the product team actually leaves. Yeah. I mean, that would have been my answer is the infrastructure part. I mean, you know, not to beat a dead horse, but compute, right? Like getting the compute and using it in the right way is such a challenge, you know, especially for startups. This has been such a theme. Yeah. I think like anthropic is single holding back science. It's not going to be higher. No, I mean, and to that point, like we, you know. I mean, they're also accelerating science, but it's like this weird doing. No, totally. Like one of the things that I help a lot with at Chai is buying compute for the company.

1:02:16Worst job, man. I would not recommend it. It is very stressful. But, you know, even September of last year, right? You're back to the hardware job. Yeah, I know, exactly. In the wrong way. But, you know, September of last year, we started to really notice like things are getting tight, right? We were doing a lot of our inference on, you know, spot and on-demand markets. And we'd have these days where you just like get these capacity crunches and we're like, okay, we should probably start to get ahead of buying some compute for ourself. And I mean, I think everyone probably says this, but man, it was hard.

1:02:50Like, I think I didn't realize how much of a power law, you know, this is, right? Where, you know, there's say 10,000, you know, B300 units that are shipping everywhere, right? The hyperscalers and the, you know, the biggest AI labs are buying 95 plus percent of it, right? And then you kind of have the startups like fighting over the scraps. And I think the other thing that's really interesting, especially if you look at these later compute versions, right? The Vera Rubins or, you know, the B300s, like a lot of this stuff has been built very

1:03:21like LLM for it, right? Like you have these, you know, systems with like huge KV caches where you have like 72 GPUs that are all acquired to talk to each other, right? And, you know, obviously some performance gains there like help us, right? But like, it's kind of interesting just how much the compute market has kind of gotten LLM pilled. I think there's like a whole probably set of, you know, compute stack and inference optimizations and things that need to be made for this class of models. And, you know, I think this class of models is going to be like just as big, just as impactful

1:03:51as LLMs. But it's almost like the compute market like kind of doesn't realize that yet, both in the capacity sense, but also in like the software stack sense. So we actually spend a lot of our time, you know, even just like doing basic like optimizations of compute to like get them to work better for the types of models that we have. Yeah, I know that some structured models are more recursive than LLMs, for example. And so that, which changes sort of like the, maybe the compute to memory ratio that you need and things like that.

1:04:22What are some of the like sort of cool or interesting optimizations that you've done there? Depending on the type of model. So like we can go back to like a try one type model. In that case, we're following the full two, three architecture. And there you're like, rather than doing attention over like this, like normal sequence representation, you're in a sense, loosely doing attention over this pair representation. So you can think of this as like a sequence of length L squared, rather than like typically length L. If you're doing attention over that, the way that you actually batch this up, it ends up

1:04:52being L cubed. Now you're in like a pretty, pretty heavy compute regime. Uh, so the amount of flops that you're putting into every token stays, it's pretty high. The amount of memory that like the memory bandwidth, uh, like overhead of just transferring that, uh, from like SRAM to whatever, that's a real bottleneck in these architectures. So like even something as simple as like a layer norm, uh, can, it can take a long time actually. Like that can be a significant amount of the compute that you're using. Uh, so I think like on our side, we've spent a lot of time just like optimizing and engineering,

1:05:23taking engineering very seriously so that like these operations are, you know, at least better. We're always looking at like a, how do new chips perform compared to the older versions. Sometimes that's even different for training versus inference. And like, of course, Neil knows this really well. Well, so, so there's, you know, what you're doing on the individual GPU. And then there's like, how do you like orchestrate fleets of GPUs? Right. And, you know, uh, you basically shard your computation. Right. And so, you know, when you're designing a molecule on CHI, it's not necessarily like one call, right? Like it's, it's a lot of, a lot of GPUs being thrown at the problem, right?

1:05:54Across, um, across a lot of compute. And, um, actually I, I would say that one of the hardest things to get right in, in software engineering is durable execution. Are y'all familiar with that term? Can I go on a little? Yeah, sure. Go first. Yeah, yeah. Ultimately, if you're like computing a lot of data, you know, model calls across like a very wide set of infrastructure, you always run into these problems where like some part of the infrastructure is flaky, right? Like maybe the bucket you're grabbing your data from like goes down or like your database has a blip because they're like too many transactions against it or you're like GPU errors out, right?

1:06:29I've been at companies before where you like spend so much of your time just dealing with this shit, right? Like you, you're basically putting like all of these cues and like all of these retries and you're like duct taping things together and you have a, and it becomes this mess where now what used to be like a, ideally like a pretty simple like computation that's just distributed, you're ending up spending like 95 plus percent of your time on all of this queuing and retry stuff, right? Um, we're huge fans of this company called Temporal. Basically, you know, there's this idea like, look, if you're just trying to get something,

1:07:00a really long running job to run, at the end of the day, what do you need? You need a queue, you know, you need your flaky thing like pulling off of the queue. You need some retry logic to put things back on the queue if they fail, right? And then you need some whole like orchestration system to just like tie all the queues together and monitor them. What's really cool about Temporal is like, this is a tech, a company that's kind of invented a framework for doing this. And, um, one of the, I think one of the technical decisions we made early on that was very helpful was to run as much stuff as we can on Temporal, right?

1:07:31So whether those are, um, you know, calls out to the database from the app, right? To make sure the database transaction goes through without failing. Okay, let's have side effects like sit on Temporal so that they get retried smartly without us having to like write our own queue logic, right? Or things related to model calls or things related to orchestrating really long data pipelines. Point being like, you know, one of those primitives, like just like, hey, you need to get durable execution right so that you're not stuck in like retry hell. Uh, a really deep, like engineering thing that like you wouldn't realize if unless you

1:08:03were like me and Jack, you've been like burned by this like many, many times before. Um, and I think like we're at this state now, right? Where we've, you know, we've raised another $400 million. Uh, I have to go buy another compute cluster. Like, you know, like we're going to have like really, really, really large runs and inference and, and, and, and training sets. And so, um, getting those foundations right is what's actually going to let us do more ambitious things. And to kind of answer your question, actually, that's a lot of the bio, the, the, the bottleneck to making, uh, making biology more like engineering is just like having the

1:08:34right engineering primitives, um, supporting it. I have an analogous tangent on the model side. Actually, one of the things that's, uh, kind of nice about those problems is they're like super visible. Uh, so like, at least, you know, like, Hey, this, this crash, this failed for us, we just see like loss curve didn't go down or like we see weird gradient behavior or whatever. I think a lot of these same principles, like, you know, engineering first, like that also applies on the research team. One thing that I like to say is kind of like complexity and being bitter lesson pill, they're like fundamentally at odds. For example, I think like outfold three, I might get this number on, but I think it was like 23 submodules.

1:09:08And at that point, that's a really difficult system to optimize and study. You're like, all right, what happens if I change? Like if I tweak this thing in submodule 30 or like 21, what, what happens to the whole system? And you can always think, Hey, we can make this better by like adding module 24, but like, should you, or should you think about just like removing things and lowering that complexity down? But I think that's like a pretty fundamental thing at CHI is just like the engineering culture and just being like very simplicity biased. Have y'all seen the picture of like the SpaceX engines? It's like Raptor one. It has a bunch of pipes and like Raptor two. We have a picture of that, like on our office wall. Cause I mean,

1:09:42it's just true, right? Like how do you delete, delete, delete more things? But the only way you can accomplish that is, I mean, the reason alpha fold two and alpha fold three worked, they were small models, relatively speaking, they were very compute intensive, but they were very data efficient. And like the, there was inductive bias after inductive bias brought in by human intuition and probably like hard fought experience. Um, it was, they're incredibly efficient. If you try to knock down those things, you know,

1:10:15they're not like a house of cards. Like everything is a incremental improvement on top of it. In order to get beyond that, it seems to me like you really just need new sources of data. Uh, you would need at least treat data fundamentally different in a way that is much more efficient. Um, I mean, I, I mean, I actually kind of surprised to hear that you have scale to that degree because it suggests that you're doing something very different from what the community is thinking, the way the community is thinking about it. I don't know if you can comment about that, but we're pretty first principle people. Like the whole research team at CHI, uh, except for me and Kevin, really,

1:10:50um, like we're the only people with quote bio background, even still, like we're, we're pretty far removed. So I think like we, we try to like look at every problem as a core ML problem. We try to think of like, what's the analog in other spaces. So like, um, even for image models, like CNNs were built to process images. So like images should be looked at in patches. Like that was the nice inductive bias there. And then people were like, well, you can just kind of tokenize this thing, throw it into transform and it's going to work. And like, it did end up working, uh, even like on a relatively small data set. But I think, um, for proteins in particular, it is really hard.

1:11:22There's not as much structural data. There's a ton of sequence data. And like, that's one of the unlocks for like ESM working. Uh, you can get that to just run on a transformer. If you try to do the same thing with like experimental structure data, good luck. You need alpha fold. Yeah, there was the, there was that Apple paper where they distilled on the alpha fold, which it was actually really cool that you could distill on a very large data set and you could get, you know, good signal, but you know, it didn't generalize at all because it wasn't reasoning. It was really pattern matching. Like one of the things, these like triangle layers you were talking about, for

1:11:53example, uh, they do have a very nice inductive bias. Maybe it's not the triangle inequality, like the paper originally proposed, but it's a clean inductive bias and it unambiguously is like one of the things which made it work and it just comes at a huge cost. Yeah. Yeah, no, I think that's, that's definitely true. These layers are pretty costly. Um, and like that kind of limits what you can do with the architectures. They're not like, not only are they like costly in terms of compute, they're just like not efficient on modern GPUs either. You have small hidden dimensions, large sequence dimensions. Like it's like exactly the opposite of what GPUs are designed to process.

1:12:28One takeaway from like triangle layers is you're kind of just trading off parameters for compute in that sense. Like that's like one mental model for thinking about this. Uh, I might want to like throw more compute at the problem and just trade that off for parameters. Cause like I won't be able to hold as many, like, I can't literally store these large pair representations and still do normal attention. Uh, so I think there are fundamental things you can abstract from the ideas like alpha fold. Um, but you can kind of just like tweak these and start building off of them in your own way. It sounds like you have quite a bit of research, um, like fundamental research going

1:13:02into this direction for, I guess, audience looking for a nerd snipe and, uh, ML engineering for new problems, probably something very, uh, it's a very, uh, different research direction than a lot of the communities going in. Yeah. Yeah. I think what we built at Chai is like, it's, it's very unique in a lot of ways. Uh, but also very tied to like what CoreML is, is good at the kind of what I was saying before. Like we try to map every problem into like a CoreML problem. We think, you know, how would you approach this if, if it were an LLM or something like that? Um, but yeah, like at the end of the

1:13:32day, we really, really value simplicity. Uh, and we, we really encourage people who don't have a bio background to like not be scared of this stuff. And I think that extends into the product too, where, you know, there's a balance to be had here, right? Between like, how general do you make the product? Like, do you build a cross reactivity workflow and a selectivity workflow and a by specifics workflow? Or do you all say, no, like let's make the model general enough to say, I'm going to like condition on arbitrarily binding or avoiding something. And then you just have a very general like screen in your CAD suite where you can say, Hey, I just want to avoid or bind to

1:14:06these parts of these different structures. Right. And I think, um, you know, kind of like the, the ML team, like I don't, I don't have, you know, a formal bio background. Most of the, the product and platform team doesn't have a formal background either. Now there's some amount of like maybe regretting my words that I'm going to have, right? Cause I'm sure there are, you know, a million nuances and it, you know, I don't want to come off as, you know, too, too brash or naive there. Um, but you know, I think, I think it's, sometimes it's helpful to not be burdened by like all of the, Oh, these, this nuance and this nuance and this nuance. And you can, you get to kind of bet and be maximally, uh, general because, uh, you know, that's kind of what we're seeing in the

1:14:39research. You can, the models are very general. That lets the product be very general. I'm thinking back to like, in my, in my CS theory days, my first advisor was like, uh, we're working on some problem and we, we needed like a polynomial time algorithm for something. Uh, and he, he would always tell me like, never underestimate the power of polynomial time. Like, cause it's basically like, you're allowed to choose like whatever exponent you want. And my first paper was an end to the 20th time algorithm for this problem. And I was like, Andy, I did exactly what you said. He's like, wait a minute. I didn't mean it like that. Yeah. But I think like, it kind of like, you can really help yourself. Like you can free yourself

1:15:12up a lot when you're like, all right, I can kind of do whatever I want and then kind of simplify it later. Uh, and I think that's really like a pretty fundamental way of thinking about things that we, we leverage a lot at CHI. The space of binders of protein design and binders in general is actually a fairly crowded space. I, I'm curious about what your general outlook of the, the field, the industry is. I mean, I can go back to like some anecdote. I was at maybe NeurIPS three, four years ago, right? The one right after RF diffusion came out, I was talking to someone in the Baker lab and they're like,

1:15:44man, I just one shotted. I don't think they even use one shot. One shot wasn't even a term back then, but they just like, I just got picomolar binders out of RF diffusion and just like through the cryo. Great. Right. It didn't seem like that just solved the problem. Like it's not like, Oh man, now every, yeah. But there are lots of people who I think have seen that you can actually do protein design, at least in some categories quite well. I'd say like, is it many proteins or mini binders? Um, ironically, nano binders are actually smaller than or larger than

1:16:19mini proteins or maybe like a little bit harder. Antibodies are typically considered even harder, but there's this like, is this something which can and will be commoditized at least in some part? How do you compete? Like, where does this, where do you, where does the field go from here? I mean, I think the answer is it's kind of all of the above. Like, I think there probably will be some commodity layer for, for certain types of modalities or drugs, right? I think at the same time, we're going to be able to do even more and more and more ambitious drugs. And you're going to, it's just like what's happened in LLM land, right? Like you have your, your open source models that are

1:16:53maybe general and helpful for some things, but people are still buying frontier models, right? And actually, if you look at the amount of value captured, it's actually the, the closed source frontier models, you know, the whole pie is growing, but it's growing so fast that even as the open source models like share expands, that the, the frontier models are still able to capture the majority of the value, right? Like raise your hand if you're using an open source model on your day-to-day. Right. And, and what are the reasons for that, right? One, like if you have, you know, more intelligence, you're going to go after harder tasks, right? I think if we have more, you know,

1:17:25intelligent bio models, we're going to go after more, more crazy bio tasks, right? But then also too, like, I mean, a lot of the reason I don't use the open source model is because like, you know, I don't get like cloud code, right? I don't get like cloud, you know, I think there's like a product layer to be built that is just as important as the model layer. We learn a lot from our partners and, you know, the people in the building as well, just like, what are the really tough things that they get stuck on using the models, right? And some of them are like, you know, the dumbest things, right? Like, you know, I want to be able to better visualize this piece and like focus on that. And

1:17:57some of them are actually like very sophisticated things that we then have to build some like pretty vertical product for. And look, maybe in the fullness of time, like AGI, like one shots everything and doesn't matter. But I think there's quite a bit of, of time until we get there. Right. And I think the product makes a huge, huge difference for that. That'd be my answer. I mean, you probably have a more model forward answer. No, like, I think like, like biology is slow, which is like one kind of nice thing. And there's like not that much labeled data. So like you could take all the publicly available sequence information out there that might give you

1:18:29a good base model, but you still need some measurements on that data. That's still pretty time consuming. And then you need to like iterate on that. So I think there are even just data blockers there and unlocking, like if we really want to do this zero shot design candidate, start generating molecules that are almost ready to go into the clinic. I think that it's more to that. And then just like, you know, AGI might not solve that right away. I think there are definitely like some technical blockers there. But even in the space of, you know, specialist companies, I mean, I'm not going to like just start naming them, but there there's, I think, I don't know, probably 10, 15 protein

1:19:03design startups. I think that the two things, which it sounds like Chai has gone on is like one, all in a product and two, you are not trying to do your own platform. If you don't have your own data mode, you know, is that going to like help you one out in the end? Or is that going to be a, you know, a blocker? I don't, I'm just, I'm just curious about that. Yeah, that's, that's a great question. Yeah. So, so Chai definitely no plans of like starting a pipeline. Like we take the partnership model pretty seriously. And we, I just like from a personal stance, I love the incentive alignment and just being like, you know, we make the

1:19:36models better, the partners succeed more. And just like, you know, that iterates on itself. So like, I think that's like a pretty unique part of Chai is like one, just being able to partner with a lot of people to getting like the feedback on the product. So like, you know, knowing that it's very real, this is in like, like legit big pharma hands and like they're actually running campaigns on this stuff. So I think it's interesting. We really have to be model forward, model focused. Like we need to keep delivering value. So that puts a lot of pressure, like on the research team, the product team, first of all, to like, to serve these things,

1:20:08the research teams always shoot for like better and better versions. The way I think about this is like, if you're a bitter lesson pill forward, kind of like thinker or company, then there kind of comes a certain point where there's a lot to do on like both the model and data side, but I don't think either is exhausted. It would be stupid to say like, we don't need any more data, but it also be stupid to say like the models are stuck. We only can like use data to solve these problems. So I think there's like tons of room to grow on both sides. We're taking like both very seriously. And I would also maybe push back on the no data moat premise, right? That'd be kind of like saying,

1:20:41hey, like all the enterprises that work with Anthropic, like you're not letting like Anthropic train on their data. So like you can't like build models that are good at enterprise workflows, right? I think, you know, one, we are investing in this, right? You know, there are ways to turn compute into data and get more and we're doing those, right? But then also too, okay, what is the kind of data that you're trying to get, right? And I think what is kind of cool about, you know, working so closely and supporting so many of these partners is we get to really learn about, you know, what is like the stuff that would be helpful in research, right? And so rather than

1:21:12doing research in a vacuum, you know, based on what would hypothetically be cool, we're able to sort of kind of do informed research based on like, you know, what our partners have just been very organically asking us for help with. I see. Do you, I assume that you aren't allowed to train general models based upon your partner's data. Do you train specific, specialized models for like, does, is there a Novartis model and a Pfizer model? Yeah. I mean, like a lot of these, a lot of these deals, you know, and this is all public, right? We are working with them to, you know, train or fine tune a version of our model for

1:21:43them. And I think there's probably like so much more we can do there over time. My brother started a company called Applied Compute, a great company. They're kind of doing this thing for, you know, design for LLMs, right? And helping enterprises really understand the value of their language data and do that for specialized tasks. I think there's a whole world where we could potentially do that for biological data. What is the value there? Like, what is the lift that you get from using their data? I mean, is it just that it's more data or is it more that there is specialized to a problem? You know, they have, they have a lot of like scientific, you know, data that they're getting

1:22:15from experiments that can maybe help our models do better in like particular classes of candidates that are targets that they care about. Yeah. I mean, even something as simple as like, they might just have some preferred way of doing things that might not be like native to the CHI model. And they can like, you know, kind of like ask the product team and in a sense to just be like, hey, we like, you know, our, our designs have property X. Can you make sure that they have those? So I think like even things as simple as that actually have like a pretty big impact for them. Yeah. So, I mean, this goes along with the, a pet hypothesis I have that all AI companies and

1:22:52especially bio and scientific ones are actually consulting companies. Pharma, I think is particularly the case because you're developing a new drug, right? It's almost by definition new, right? So like the existing stuff has to be customized in many cases, right? Unless you're doing something that's just reiteration of old stuff, but a lot of the big pharma are, are pushing the boundaries of science. Yeah. I mean, certainly like we aim to make the models very general. We aim to make the product very general. We mean, aim to make it powerful, but yeah, I mean, there is integration

1:23:24work, right? With every, with every customer. To answer your question, you do get some defensibility just by doing that, right? And I think what is, what is nice about building, you know, trusted relationships with these partners is hopefully, you know, if we execute really well over the next, you know, the first year, then they'll continue working with Chai to, to do more ambitious and more, more drugs past that. I mean, it's going to be hard to switch, right? I hope so. Yeah. Just getting the security review done. Yeah. Yeah. Yeah. Like maybe, maybe one other interesting point is like, if you think of this, like on a per token basis, I don't know if there's another

1:23:58domain where like the downstream value of a token is like as valuable as it is for pharma. Like, you know, you're thinking about like the actual drugs that come out, like these can be like multi-billion dollar assets. So like the case of GLP-1s, I think the two GLP-1 drugs combined like maybe a trillion dollar asset, like revenue stream. Yeah. I mean, I think up until I think three months ago, right, GLP-1s like total revenue was more than all of the AI labs put together. Yeah. I don't think people realize that. Like I didn't realize that. It's crazy, right? But yet the market cap way lower. It's like crazy how relatively speaking the market cap is.

1:24:30And you know, I didn't realize how much of like a VC business, you know, you know, pharma is in, right? They're in some sense, like taking really ambitious bets. You know, I think one of the things that was really cool with, you know, is like if you study the history of Silicon Valley, right? Like obviously people think of Silicon Valley with software, but you know, in the 80s, one of the, one of the biggest venture outcomes, one of the first ones was, was Genentech, right? And because it is such a VC model, right? You get the string of tokens that can then give you so much value downstream.

1:25:01Just a general shout out to Outposting's blog series about like finance and funding and yeah, really fantastic. Yeah. Before that, I knew a lot of those points, but I did not realize just how deep that rabbit hole went. Yeah. It's a, yeah, I mean, it's maybe the biggest, single biggest problem in biopharma is actually just the funding model. There's also, have you, have you heard of Aram's law? Yeah. Oh, yeah. Yeah. Yeah. Yeah. It's more backwards. Yeah. More, more is a lot backwards. So it's like in like compute,

1:25:32you know, it's kind of scales. So you have like this nice exponential scaling, log linear scaling of compute and you have the exact opposite in pharma. So like the cost of actually making a drug in pharma is kind of like increasing exponentially. So like the amount of money put in per drug is growing at kind of like an exponential rate, which is, it's pretty interesting to see this. Yeah. Which guarantees at some point, the marginal return on a new drug development will be negative. Exactly. So unless someone, I mean maybe Chai, figures out how to, you know, fix this. I think that

1:26:02we might be on the verge of sort of flipping some of these. Phase transition, bending the S curve. Yeah. Just to double, maybe belabor the point, but that pharma and VC fundamentally both are optimizing a portfolio. Yeah. And I think that's the connection there. Yeah. Thinking of pharma as like sophisticated capital allocators, right? Where they have this portfolio of targets and they're allocating between them. I think that was a big reframe for me. Yeah. And I think we will just see more of that in the future, right? And hopefully they can

1:26:32take, you know, in a sense of VC taking riskier bets, like hopefully pharma can take riskier bets and pursue really, really cool drug targets in the future. That analogy is actually like one, the kind of like VC type investor-ish model. It's like actually how we think a lot about research at CHI as well. Our research team is relatively small. I think definitely compared to like a lot of the, like the isomorphic deep minds, like our research team is like, you know, in the around 10 people. So like we're, we're a relatively small team, but we kind of think of it as almost like an investing job where like you're investing ideas

1:27:05towards compute. In the same sense, you're really just capital allocators in that respect. Yeah. I actually think maybe this is too cute, but I would even make the broader point, which is I think every, we kind of think of everyone at CHI as a bit of a capital allocator. So I think one of the things that surprises people is we're, we're pretty small. We're, we're only 30 people. And that's because everyone we hire onto the research team or the engineering team, you know, especially now that they're in some ways like very empowered with AI, a lot of it is just like allocating, you know, their attention into the right ideas and allocating their compute.

1:27:35This is actually, I think a characteristic to some extent of machine learning AI projects projects and also science, right? Whereas if you're building like a API for some B2B SaaS company, that's not building foundation models, whatever, your limit is mostly people, right? So you're the resource you're allocating is almost entirely people. Whereas if you're building hardware, you're building AI models, you're building something scientific, then your constraint is those, the resources that are, you know, the bottleneck is, you know, the lab, it's the

1:28:11compute, it's other things. And so that you have to really be in that mentality of, I have these limited allocation of, I have some shots on goal. How do I allocate those shots? Well, I would say yes and no. So I agree. It's a bit more like that, right? But like, let's going back to the example of building an API for, you know, a B2B company, right? That API has incremental cost. You have to support it. It adds complexity to the product. It's another thing you have to go market and sell. Maybe you should actually be allocating that into like a different bet, right? A different thing on your product roadmap that you should be prioritizing

1:28:42instead of the other thing. I think in a world where like building things just gets like really cheap and, you know, increasingly free, the scarce thing is the attention both that you can put into it, right, to keep your product simple and grokkable and that your customer can put into it to like really understand how to use it. I see it less as like a binary thing and more just like we're all kind of as engineers going to be a little bit more like allocators of attention. Yeah. Which is what executives are. We're all going to speak coming. Well, I mean, like, there's a podcast with Satya at Nadella, right? He says, you know,

1:29:13Microsoft wants to make everyone a manager of infinite minds, right? If you like really take that to your extreme, like everyone's going to be an executive. I mean, I certainly feel like an executive when I talk to Claude every day, right? Yeah. A little suite of interns who are all going out and eagerly solving problems. You may or may not have actually wanted, but they're solving the problem.

Bottlenecks and Final Takeaways

1:29:32Yeah. So we have two typical questions that we asked that we've already kind of asked one, but I'm going to ask it again, maybe more directly, is if you, and you can both answer this, if you could remove a bottleneck from your problem space by fiat, what would that be? Oh, that's, that's an interesting question. I think one thing that'd be really nice, like, just, uh, I'm like always in research land, very hard to turn off. For me, it's probably just the validation loop of protein design in general. Uh, so like just being able to say like instantly,

1:30:06like, Hey, this thing works, this thing doesn't, there's still a bit of walking around in the dark that you're doing. Uh, just to like, you know, you have, you have some ways. And like, I think at CHI we've taken this like very seriously, but it's probably along the lines of just like validating hypotheses and like, you know, knowing for certain that things work. Yeah. That's unsolved problem for sure. Yeah. Yeah. And it would be hugely valuable. Hugely valuable. Yeah. Yeah. Yeah. I'm going to take a much more abstract answer to that, which is actually like talent obscurity. I think, you know, there's a lot of smart people going and working on LLMs. You know, there's a lot of people that are working and becoming software

1:30:38engineers for, for SAS. Right. But I think just like not that many like smart people go and work on bio. You know, I didn't work on bio like in high school cause I was like, Oh, I could like pick up my computer and program apps. But if I want to work on bio, I have to like go study and get good grades in school and like maybe get a PhD or whatever. Right. And you know, maybe that's one reason for it. I think another reason is, you know, a lot of the stuff is really obscure, right? Like we threw around a lot of big words during this podcast. You can't really visualize the things. It's one of the things we care a lot about at CHI is like, how do we make the whole thing feel

1:31:10visual on our website and in the product? Um, and you know, part of the reason we're here is like, I, you know, I think, you know, more people should realize like you don't need to like have like a super, super, super specialist bio background to contribute to this like computationally. Um, and so, um, you know, I think a lot about like talent flows and like where talent goes in the economy and right, you know, in the nineties, everyone was flowing to talent and, you know, since the two thousands, people have been flowing to tech, but, you know, big tech like ate up a lot of the talent, you know, until, you know, a few years ago. And now

1:31:40maybe like LLMs and the big AI labs are eating up a lot of the good talent, but it's like, you know, at the meta level, like how do you allocate talent better? You know, selfishly, I want more talent going into bio. I mean, we probably want more talent going into manufacturing and physical world things and these other problems that, that the U S has. But, uh, yeah, I think communicating that better would be the thing that if I had a megaphone to, to talk to everyone, I would, I would try to do that. Okay. So, and then that leads to the second question, which is, and maybe the answer is

1:32:11the same, but what is the takeaway, one takeaway that you would like to people to have from the episode? Yeah. I mean, I think, um, you know, biology has been this somewhat obscure feeling field where you're stumbling around in the dark. You don't know what you're looking at. You're dealing with, um, non-determinism in your experiments. You're having to do a very long and iterative trial and error loop across a very, very long amount of time. And, um, at some

1:32:41point you're crossing that threshold of what you can do computationally when you can get folding models down to being within, you know, an angstrom, right, where you can get design models to give you, you know, uh, hit rates, you know, north of 50%, or now you can put them, you know, in a, in a 96 well plate and actually have like 48, uh, interesting binders. You start to get to the point where now you can declaratively precision engineer what you want rather than betting on, you know, nature or trial and error to get you there. Um, and I think that, um, look,

1:33:11we had the same thing happen in software where you can write code and you can deterministically get an outcome or an electrical engineering where, you know, you, instead of your schematic being drawn out, you can put it in cadence design systems and get it on, on, um, you know, it made in software, right. Or, or CAD for, um, for mechanical engineering where you can sort of precision engineer your part and get it printed or manufactured. Um, you know, the same thing is happening in bio and it's happening very quickly. Yeah. And that really opens the door for a lot of, uh, really interesting people, or maybe it wasn't as escrutable or accessible before, right? Like

1:33:45software engineers, like myself, researchers like Matt, you know, obviously we're still going to want the specialists, but, um, you know, the, uh, the, the generalists can often really accelerate the, uh, the precision engineering happening in the domain. Yeah. I think for, for me, like the base takeaway is that the field is actually working, uh, and like, like not only does it have commercial traction, but like the research is like actually showing signs of life. Like it's not even just showing signs of life. Like the signs of life have been shown. We're actually in a place where like the models work, they're delivering value. And like, there's still tons of really interesting

1:34:17research problems to solve. So I think there's a lot more low hanging fruit in this field than there would be in other fields. And I think the amount of impact that you can have, especially like as a researcher is just like unmatched in this, in this field for us, we're all very mission driven. Uh, but even if you're not like, it's a lot of fun puzzles to solve. Like there, there's like this kind of 3d geometry angle. There's like, if you like diffusion models, there's like a million problems to solve in that regard. We have this LLM looking trunk in like Chai 1. Uh, there's just so much of like core machine learning is touched by these problems. Um, we're still, although we've made a ton of

1:34:49progress, there's still a lot to be done. Uh, and I think it's just like one of the most interesting fields to be working in, which like, while also having some of the largest impact on just like humanity. Cool. Thank you so much for having us. Thank you for making a long journey. Yeah. 22 minute walk. Yeah. Um, and you know, we look forward to tracking Chai's progress. Awesome. Thank you guys. Awesome. Thank you very much.