00:10:39 · Oliver Vince
[00:10:39] S1: Yeah. So we we kinda have three parts of the the business. So we have the data expansion piece, that's basically all data collection piece. We're still doing that. We're still working building partnerships around the world. These different companies recently signed some in some very exciting locations that we're gonna talk about soon. We have also... You know, that's what the Trillion Gene Atlas is based on, and that's grown... And that's on track. That's growing very, very quickly. It's on track to be completed by the end of twenty twenty seven. Then the sort of... We're still scaling the models. We're still building newer models, getting better at those models, and evaluating them on tasks that people previously haven't done, and that includes tasks that we... What we call long horizon. So rather than saying to the model, can you design me one molecule that does this? It's saying, can you think about the patient? Can you then, you know, analyze, you know, what needs to be done? Can you then design a molecule? Can you then assess safety? And then can you do all these things? And what you're working towards is not a model that just does one task at a time, but in the similar way to, like, you know, these these frontier models today, you you input a patient file on one side, and out comes something that's on the validation on other side. Now to be clear, we've got versions of that such as the antibiotics and Claude Science, but we're at the point where it can do the whole thing yet, there's a lot of work to do, understand that. But then there's sort of validation that we are actually at the point where we can start treating or we think we start getting close to treating patients. We have this pipeline of medicines that we are working on, we are developing, and that, you know, we'll... We will work with pharma companies on to to develop further. And and there... There's all this sort of stuff to be built. So back to sort of realistic and optimistic. Like, realistic is kind of like, you know, we've got a very, very strong validation of all the technologies, but we've not yet really lent into it at the scale that will be required to to deliver the impact of this technology to those who need it.
00:12:26 · Sebastian Johnson
[00:12:26] S0: And is that a point in time that you look forward to is like the the very first patient? You know, is that something where where you have that milestone? What does that look like working backwards? Like, how far off is that?
00:12:38 · Oliver Vince
[00:12:38] S1: It's hard to say exactly, and it's also difficult... You don't want to promise things because it... You know, it's people's lives. Right? So you you don't want to... But but it's it's... You know, in the next couple of years, I expect we'll be there. And, you know, my hope is that when we are doing this, it's medicines that that are curative. So they're curative for diseases that previously or currently cannot be treated. And they start to demonstrate the potential of AI in this space, which is that, you know, in my view, even if your first medicines sort of equal or match what can be done today by humans, the fact that an AI model has got there means that if you continue to work on it, it will go through that. And And again, realistic and optimistic, right? Realistically, we don't know exactly how that's going to scale, how that's going to work, but optimistic that there is a better way of treating disease than we have today, and that if we can get that and we can invest in this technology, you know, we can get to a point where, you know, certain diseases which are very commonplace today and are absolutely disastrous are kind of not a thing anymore. As in, the people will still get them, but in the same way that most infections in the West, at least, they're not a thing. As in if you get an infection, you go to the doctor, you get some antibiotics, whereas previously it would have killed you. We hope to move more diseases into that space, basically.
00:13:59 · Sebastian Johnson, Oliver Vince
[00:13:59] S0: Yeah. That'd be amazing. Right? It's just a a such game changing way of approaching disease. When you talk about it, do you see yourself becoming a drug company, or will it always be partnerships with pharmaceuticals? [00:14:15] S1: I think in the long time, it will be both. I think that we will always focus on concentrating where we believe we can contribute, where people have already built the infrastructure, so they've built manufacturing plants, they've built hospitals. We should not try and rebuild them, and we should try and work with them. But where we believe, you know, for these AI models to have the most impact, they need new infrastructure or they need new ways of interfacing with the patients or they need new data and that sort of thing, then I think it's our priority and our responsibility to build it. And I... You know, it's a bit of a... We've built quite a lot of different things already, and I'm more focused on sort of the end outcome, which is demonstrate AI can cure disease. And it might need other things to help it get there, but it's the premise of that actually working. And, you know, all these pieces, you need to build AI, you need it to focus on curing, not just treating, and you need to actually get to the point where you're treating disease. It doesn't, like... It almost doesn't get more complex than that. So the... You need to do all of it, and I don't know what you identify as when you do all of that, but it's definitely exciting.
00:15:18 · Sebastian Johnson
[00:15:18] S0: And when I look at the landscape of other companies in this space, whether it's isomorphic in London and is also some in The US, maybe like evolutionary scale. It feels like, and maybe I've got this wrong, you've taken a different approach to data and training the models and that yours... You seem to be betting really big on building this in this almost a unique proprietary dataset, was I think a lot of those are the ones that are kind of using public data. Firstly, is that right? And then why have you taken this this different approach to some of the other companies in this space?
00:15:51 · Oliver Vince
[00:15:51] S1: That is correct. I mean, we are, for want of a better word, like, trying to build, like, general biological intelligence. Like, we're trying to build models that that get generally smarter, and I think other people have approached the problem a bit more, like, linearly. So a bit more saying, like, look. Here's how the drug discovery process works today. We're gonna build a model for this and a model for that and a model for this, and we're gonna sort of treat each one like a bit of a game, and we're gonna play each one, we're gonna get really, really good at each phase, we're gonna stick them together. And what that means is in each of the individual things, they get very, very good at that, but they sort of, in our view, you know, I think that's very promising, I don't, you know, I don't want to be negative on any of it, but I think that has certain limitations in the fact that you're replicating the existing process. Now we've taken a bit more of a leap and said, look. What if you just train this thing to understand biology and you learn the ability to talk to it? What would it be able to do for you? Right? And so an example of that would be normally when you treat antibiotics, you would take the pathogen, would you do a bunch of modeling on the pathogen, you'd find the target, you'd find the mechanism of action, you'd find the thing, you do the docking, and you do all these steps, and then you design a peptide or an antibiotic that worked against that. And then you, you know, you test the sort of five steps in that process, and you build a model for each one. What we did is we just built this model, and we say... We showed it lots of pairs of pathogens and, like, antibiotics and just said, here's a new pathogen. Give me a new antibiotic. And it just jumps from that, and ninety seven percent of the time it works for... In the lab, and it works, you know, equivalent to last line in in mice and that sort of thing. And we've no idea in principle, like, how it did that jump, but we know that it's able to jump from a to b. Right? So in kind of in the way that you're not quite sure exactly exactly how ChatGPT took your prompt and jumped to something at the output, but you know that it didn't, you can check that it did. We... Is...