Transcript: AI and Drug Development - Keeping it Real with Pharma

The following transcript was taken from a recorded conversation between Dr. Jo Varshney, CEO of VeriSIM Life, and Dr. Grazia Rovelli, Senior Researcher and Project Leader, Italfarmaco Group. Read on to understand how established companies are approaching AI and integrating it into their strategy as well as their research and development workflows. (Edited for clarity and length.) 

Dr. Jo Varshney

Thank you and welcome to this webinar. I'm very much looking forward to sitting down with you and keeping it real, as we say. So before we begin our discussion about anything and everything good and bad and ugly, AI - I would love to talk a little bit about what we do at VeriSIM Life.

Our core objective is to reduce the translational gap using our technology, BIOiSIM®, and ultimately bring (treatments) to patients in a cost and time-effective manner. We are a team, well-established, with more than 100 years of drug development expertise, spanning from software engineering to drug development and experience in bringing new drugs to market. We have several different clients globally, and we’ve been well-established, and it's an honor to be well-funded by incredible investors as well as (to receive) government funding such as from NSF and the NIH. And, thanks to several different award organizers, we have also been recognized for all the hard work that our team has been putting together in terms of the approaches that we take when it comes to artificial intelligence and machine learning. Over to you Grazia.

Dr. Grazia Rovelli

Thank you Jo and VeriSIM, for asking me to join you today for this webinar. I'm very excited for our discussion. So just to give you a quick overview about my company: Italfarmaco was originally founded here in Milan in 1938, I believe. And now, as you can see from this slide, we do have a global presence.

Our main manufacturing sites are in Italy, in Spain and in South America. And in terms of headcounts itself, I can count about 4000 employees among which we have about 350 people working in the R&D department. I myself am part of the preclinical drug discovery team and it's about 40 people, including both chemists such as myself and biologists. 

Here you can see our pipeline. Just to give you an idea on the indication we're mostly interested in. So our most advanced molecule is getting a start and we are hoping it's going to launch in 2024 with indication to treat Duchenne muscular dystrophy. Fingers crossed. Yeah. And among the other products we have in our pipeline, we have small molecules, but also conjugates for the areas of oncology, neuropathy and fibrosis.

And concerning the products we currently have in market, they are mostly in the areas of cardiovascular disease, CNS anemia, women's health, among others. So hopefully that gives you a little bit of an idea about Italfarmaco, and I can pass it back to Jo.

Dr. Jo Varshney

Thank you Grazia for this introduction about Italfarmaco, and we wish you the very best with [the development of your assets] 

So the core crux of this discussion is, are we there yet or are we just hyping up artificial intelligence? As you can see, we have been covered in the two different types of news cycles.

One [point of view] is, should we believe the hype and can artificial intelligence hold the power to improve clinical development and clinical success?

So given your extensive knowledge on drug development and different programs exploring different targets, when was the big moment for Italfarmaco to discuss the AI strategy and what was the process in terms of, from the management to the communications folks, as well as the challenges someone like yourself faces to educate and sort of adopt artificial intelligence?

Dr. Grazia Rovelli

Yeah. So I would say, first of all, that Italfarmaco is very open to new technologies, and I can speak especially to, you know, our preclinical research efforts. And so, you know, we're in our most recent projects, we're trying to adopt new technologies, say, for example, fragment-based drug discovery, or we're looking into products. So just to give you an idea, the management is really open to embracing and trying out new technologies. And obviously, artificial intelligence has been on the news very, very much over the last, you know, maybe couple of years. And of course, as someone who wants to be attentive and and open to new tools, obviously artificial intelligence was something that our board made a priority for us to integrate within our company, starting from preclinical drug discovery projects.

And so that's essentially where my role was created. So I joined the company in 2022 and as you know I act as a computational chemist, and most of my time is devoted to scouting partners that are in the artificial intelligence for drug discovery space. And that's how everything started. And so I think it's been a bit of a learning curve, both for me, for myself and for the company. And I think it's been very productive over the last year so far.

Dr. Jo Varshney

Yeah, I think that's actually incredible. Like not many companies come together on a very novel methodology for software, right? Because in pharma in general it is all code, right? It's not a robot or an assay. So it's incredible to see the management being supportive of those changes and creating a role for someone like yourself.

And let's dig in a little bit about the challenges from the scouting versus how to communicate internally. What kind of issues that you were facing while having these conversations?

Dr. Grazia Rovelli

Yeah, so I would say that possibly the biggest challenge we experience every day is the limited resources. Obviously, you know, as I mentioned, we’re a mid-size company, not Novartis or Pfizer or AstraZeneca. And so, you know, we need to really think about how we want to employ those resources in the most effective way possible.

And so that's for sure. You know, (it’s) something that I've been thinking about a lot and something that I have tried to make the best for, for our company. And when I say limited resources, it's both in terms of money and people, because as I mentioned, my position was created just in 2022. And I have my supervisor, who was just kind of interested in these topics. So he gives me a hand and he's super supportive. But at the end of the day, it's the headcounts that are really limited, right?

Dr. Jo Varshney

You are the AI for Italfarmaco!

Dr. Grazia Rovelli

Yeah. Yes but it is, you know and I think figuring out a strategy on how to make that successful has been a little challenging. But also I want to say that the company has really been open to our suggestions. So initially, say, for example, we were thinking of maybe acquiring internally for our new project some licensing access to some artificial intelligence platforms.

But eventually, as I was speaking to more and more people such as yourself, very soon I realized that was not ever going to be the most effective way of proceeding with this. You know, So the way we're looking at collaborations such as the one we are trying to establish with you guys at the time is that it's going to be a learning experience, right? It's going to be a tryout, if you will, of the most compelling technology that we've found out there and see what it can do for our specific projects and needs.

Dr. Jo Varshney

Yeah, I know. Thank you so much for the shout-out and we're very excited to establish this collaboration with Italfarmaco as well. Since we are on the topic of working with us and the scouting, how was your experience? And as I said, keep it real, so be honest, don't hold back. This is about a webinar for us to learn how we can improve ourselves and you know, what did you like? What were things we could improve and more importantly, how did you come about the decision for the company to proceed with a company like ourselves who is across the globe?

Dr. Grazia Rovelli

So I think the process was as simple as networking. You know I was sending out emails. Many of them got responses as many of them didn't, which is fine. And so just starting from there and trying to to build a network of people and trying to see, you know, sometimes I have to say a little challenge sometimes is that it's kind of hard from websites of some companies to understand what they exactly do.

And so I've met people that do super, super interesting work but maybe weren't necessarily what we were looking for at the time. But then, you know, the relationships were built. And so those are still going to possibly be useful in the future. And so by contacting, I would say between ten and 15 companies, eventually, that's how many people I managed to spoke to.

And then I think it came down to about four, four or five companies that we selected to to be able to sign confidential disclosure agreements, because at the end of the day, that's something that is absolutely necessary to be able to figure out exactly - to be able to communicate our needs in terms of, what are the challenges we expect from a specific project.

And on the other side, obviously, learning more about technology and ideas that come from the companies we're talking with. And I have to say, without naming any names, I was a little disappointed at some point by one or two companies after signing the CDAs. I realized there's a lot of marketing going on here and the ideas are like, okay, I mean, I understand what you're doing, but at the end of the day, I thought this was going to be more, more like edgy and more exciting.

But then, you know, I figured it's kind of like standard of what can be at this point in history, standard artificial intelligence for drug discovery, you know. And so eventually, I think what made all the difference in selecting our partners just came down to ideas and technology, really, and people, I would say. So, you know, I really enjoy speaking with really passionate people such as yourself and your colleagues at VeriSIM and other other companies, too.

It's really kind of refreshing in a way, because you read a lot of articles about how skeptical we should be about artificial intelligence - which is something we need to keep in mind - but also, I think finding someone who's really passionate and really convinced that they can make a difference into this very impactful research area - I just really, really enjoy it.

Dr. Jo Varshney

Yeah. Likewise, we, I think we feed off the energy of folks we talk to. So thank you so much for sharing that. 

We really talk a lot about, you know, what our board of investors say “sizzles.” But it really comes down to, can we really do this and can we solve this problem for the client? So really appreciate it. So switching the gears from the differentiation of different companies selling to talking about the actual noise in artificial intelligence.

And, you know, how do we and more importantly, I think we as a company, know the signal, right? We have so many different case studies, blah, blah, blah. But I think the biggest thing that I often see is like, how does the pharma find the signal and what are they looking for in that signal? Because every day we get some new AI company or some new AI thing. But how did you come about figuring out like, okay, this is the methodology we need for Italfarmaco? Because that's a lot to take in for a person like yourself.

Dr. Grazia Rovelli

I think I can start answering this question, remembering like an episode - I think it was maybe my first board meeting that I took part in last year or the year before, and someone in the board had these comments, which I think were really interesting at the time. And they said, oh you know, like at the end of the day, you're going to pick whichever AI company you want and it's going to be fine, they’re all the same. And it’s like no, I’m not sure that's the case. And at the time, you know, it was the very beginning of this process. So I wasn't super equipped in answering that. But as time went along, I realized that that couldn't be farther from the truth. Like, there is so much that you can kind of pick up from each company that is different, or like the most interesting companies that differentiate one from the other.

And I think a way I've been thinking about that lately is you can really tell if the founder of an AI company is coming more from the biology side of things or from the chemistry side of things. And so that's what really affects what the the philosophy of the company is. And at the end of the day, you know, as I said, and you can disagree with me but I feel like you’re very similar to biology-first company and you really care about translatability of drugs which I think is great but at the end of the day, you also need to do some chemistry.

But I really think that you can tell from the background of the founders and maybe, you know, the CSO or the main people in the company, what's the main focus and what they care about in their philosophy and therefore what is really, really fundamental in their technology, right? And so I would say to (the question of) which one is, you know, biology-first or chemistry-first, which one is best… I don't think that there's an answer to that. I think that - the way I've been thinking about that and the way I think my colleagues also agree with me is that it really depends on what's the project, right, that you're going to approach, if that makes sense. I think, you know, if we think about the project, we would like to start off with you.

We have selected a very ambitious target, I think.

Dr. Jo Varshney


Dr. Grazia Rovelli

And so I think that your kind of like broader approach is very suitable for that, right? Whereas for more of a characterized, maybe for a little bit more of a crowded space in terms of what's out there, in terms of patents and whatnot, maybe - and this, you know, is still to be proved. But I would guess that maybe a more chemistry-first approach might be better.

And so this is how - this is how we've been thinking about this, and how we have tried to set up our collaborations with respect to different projects that may have different challenges.

Dr. Jo Varshney

I've spoken to so many folks in the last couple of years, but I think putting that differentiation from chemistry-first versus biology-first is the first time for me (hearing it), and if it's okay with you, we will be using that (term) quite a bit for ourselves because you know, there's no disagreement there. In fact, it's a very astute observation because for us it's about translatability. 

And we believe, as you said, right, I think there's a lot of great technologies out there who are already out there developing the best chemistry, best in class, first in class, and they're very well appreciated. And we believe that the biology still needs to be figured out because at the end of the day, the patients have to have a positive response.

And unfortunately every drug has a side effect. What side effects are critically - and can be avoided is something we really feel very passionate about. Like, okay, can we improve those odds and can we make the biology better? So definitely we'll be using that quite a bit. And thank you so much for sharing that observation since it's something that everybody in pharma and journalists should realize.

Like, you know, the applications are so critical when it comes to understanding how it will solve a real problem in hand. And I often find that folks, it's kind of interesting - you mentioned that the board is coming in with a problem to solve - and unfortunately, many companies will say that hey, we'll solve this problem for you, even though they're not equipped or built in a manner to solve their problem.

Because, you know, there's a lot of discussion about data, and that would be my follow-up question. But it's not just the data in my opinion, and I'm very curious to hear your thoughts about - it's also the knowledge. It's how you think the AI should be working on a specific problem. 

So, with that, I wanted to get your opinion about, what do you think about the data in the pre-clinical and translation space? Do you have concerns about whether AI is really ready to be helpful in this space? And more importantly, how do you see with your partners solving problems with the data present in the organization as well as using publicly available data?

Dr. Grazia Rovelli

That's a really good question. I feel like it's a question that everybody somewhat has in the back of their mind any time AI gets brought up, you know. I would say that the concerns I feel like at this point, everybody has this kind of like mantra in their head, which is like, you know, the AI is going to sort of - is trained on something, and it's never - or at this point may not be able to, you know, give you an answer outside the range of applicability of the specific algorithm we're talking about.

And I feel like that's probably accurate. And I think this is something that people should keep in mind. And I think this is very clear also to non-technical people. You know, I don't think anyone really, really believes - or no one has ever told me that they believe - that these kinds of technologies are at the point where they're just going to give us an answer. And so I feel like the data that is out there and is publicly available - companies such as yourself try to collect and bring the most out of them - is potentially very adequate for a lot of applications, but maybe not quite there for everything.

Dr. Jo Varshney


Dr. Grazia Rovelli

And then maybe another point I can make to this is that, you know, obviously we and all pharmaceutical companies that, you know, establish collaborations and partnerships with companies such as VeriSIM have internal data that they would like to sort of exploit in these kinds of applications. And there is always a bit of a concern about what happens - not exactly what happens to the data when we share it, because you know, if there’s a CDA you’re not going to go to someone, to some competitor and go “they measured this and that” or whatever. So that's not really a concern. The concern is often, you know, so what if this company then learns from our data and then uses that learning or that improved A.I. algorithm to work with some other competitor and eventually, you know, give them an advantage.

And so I think that that's the biggest concern for most people, both, you know, in the science side of things and in the management side of things. And then that's why we always ask exclusivity clauses to be put in contracts and things like that. But I feel like companies, like in the case of VeriSIM have been very open to listening to these concerns. And I think in general that this is really the case and at the end of the day, you know, we have to - that's probably the way forward.

Dr. Jo Varshney

Right. I think, you know, there are so many things to unpack here. So let's start with first like, you know, the training and what happens to your data. 

I think that for the audience, it's a good question to ask any AI provider, like what are the ways you are ensuring the data secrecy or data - Is it mingling with other data sets? And if there are, you know, potential challenges with that, since a company like ours undergoes what is called SOC-2 and then we have several audits with the risk assessment, and we create certain database separations based on the type of project and contracts we signed. It's an easier discussion to have, but I think it's an important discussion in general to have.

The second is like all the system learns, right? And I think the system has been learning forever. As experts, they're learning from each other. They're getting inspired from other patents. We don't talk about it, of course, but we all know, like everybody and tell me if you have a different opinion or you disagree. But everything is an inspiration to somebody else's work and, you know, in the records we put the references, we put that kind of credibility. 

I think when it comes to AI, it's often like a black box. So the more explainability you have in your validation and your models and how the data came about, where it is coming from, I think that can ease some of the gap of like, okay, this generalized learning is happening and, and OpenAI is a great example, ChatGPT, right. There are several concerns and they're addressing in their guardrails… but when it comes to data, it's all about how do you collect the data, what do you do with the data? And more importantly, what do you do with the data after the work is done?

And those are the questions I think everybody should ask to ensure there is no secrecy about that since it’s your data, right, and we don't want to be doing anything without permission of the client. And then the other aspect is when it comes to pharma, you're right. Like I don't think many companies equip - unless they're selling the data, there are companies out there whose job and business model is just to sell data… So, you know, my opinion is always, stay away from the companies which are doing the analytics part as well as also selling the data, because that's a place where it's really hard to kind of control how the data could be used, or misused, however you want to see in those kind of business models.

But when the objective is to improve the model's methodology ultimately for the outcome for the client, I think that conflict hopefully gets reduced because, you know, we are making sure that the models are answering the questions that are of interest to you. And then the ultimate thing is generalized learning. Yes, there is a component of artificial intelligence to learn, but without that learning, we're not going to get any further.

And it's just like the human mind, right? We all have to learn from a book, from peers. And it's the kind of learning which is not very focused on a specific target of a client. And I think there's often this misunderstanding like, oh if they get to know my target, they're going to [help my] competitor. But I think it's more about like, are we using the right models?

And I think the best example is a CRO. A CRO company gets to access data from every client, and the assay that they're developing is based on what the clients need and then they improve those assays for the client. And that improvement most likely will not be relevant unless it's like neck-to-neck exactly the same molecule, exactly the same target, which often cannot be the case because of the patent rights and such.

So there is a fundamental assay and then you're improving on that. And I think as long as the systems are built in a way that, okay, there is a generalized system and infrastructure to take into account what specific outcomes the client is doing - so there is a very specific improvement, but it's not a general improvement. So I think having those conversations during the technological vetting process is very important to helping to reduce that gap.

And ultimately we all want one thing, to get the drugs to the patients faster, and it's economically very valuable to every client and every company. Now it really comes down to who is going to be the first and who is going to be the last. 

And I think that comes to my next question is the answers, right? Like you already mentioned that, you know, we are working on a very ambitious target. Yes. It's a really ambitious target to get out there, we're really excited. I think those are the kind of things all of us, and even though I'm quite nervous, I think my team is really excited. Like they're just like, “this is great.”

So how do you get to discussing those answers and getting comfortable like, “Hey, I know this is the technology, or this could be the technology which helps address the answers.” And where were the challenges in kind of reducing the gap from just a general technology to “can they actually answer that question for me?”

Dr. Grazia Rovelli

Yeah, that's a really good question. I think more than anything, what I can say to that is that it comes down to asking a lot of questions. I'm sure if we were to ask your computational chemists, your artificial intelligence architects and all of these people, I feel like they must have been so fed up with me at some point.

You know, I've asked so many questions at some point that I can't, you know, I've lost track. And I think that's just how you gather as much information as you can, and figure out that eventually no one, you know, at the end of the day - drug discovery is a risky business anywhere anyway. You decide to approach it but it is, as you know, a very, very ambitious, risky, expensive business. And so I think that also is reflected in the adoption of any technology and also and even more so, something as young as artificial intelligence. And so, you know, I think the most you can do is asking a lot of questions, asking as many case studies as you can read as many articles from, you know, a potential company you want to collaborate with, even though, you know, I feel like there might be a company that hasn't specifically worked on a target, that target that could present the same challenge challenges that you think you might encounter.

So you might not get the exact answer to your doubts. But if you can see how a company works with the achieved, how open they are in discussing even like the minor details of a specific genetic algorithm or neural network or whatever it is, I think that is very telling on how your collaboration is going to go and the the if you're going to be able to be successful at the end of a project, I think, and then it's a bit of a gamble.

And also I think the people component is essential to me. I am really driven towards people. I want to have a good relationship, an open conversation, where I can say, I'm not convinced about this. Please tell me more. Or no, I don't think this timeline makes any sense. If you encounter someone who's open discussing those those questions, that's very important to me.

Dr. Jo Varshney

Yeah, wonderful. I think this is this is a fantastic segue to my next question. You already mentioned “success”. What does success mean for you? Because, you know, we find often that the bar for an experiment is very low. You know, there's a lot of room for variation in animal studies and experimentation. But we often find that the bar [for AI] is like way up there. It's like every time a [project] outcome comes, the next step should be like, patients should have that. But I think it's really important that you have what you think success means when it comes to an AI partnership and how how can we kind of convince the general pharma groups like how to think about success when it comes to working with the AI companies?

Dr. Grazia Rovelli

Yeah, So that's that's a good question. So I think most of what the expectation for collaborations with artificial intelligence companies is, is that we're going to get something faster than normal, cheaper than normal, you know, more efficiently, more likely to succeed. So I don't think we expect 100% success rate. I'm hoping we're going to get 100% success! But I don't think that's the expectation. The expectation is de-risking, you know, making things faster and more financially optimized, let's say. And then and I think maybe a little more on the science side of things, I think that the expectation is that these new tools that we now have are going to allow us to tackle targets that were very hard to address before, for example. Or, you know, design new compounds that are going to have reduced toxicity from the very, very early stages of a project.

So again, that goes in the sense of de-risking a project. Right. And so I think, you know, I can see how you're saying that you feel like the bar is really, really high and in a way it should be. I think. But also I don't really think that anyone thinks that at this point in history an AI driven drug discovery project has going to have 100% success rate.

Dr. Jo Varshney

Right? So I love it. We will keep the bar high. I like the way you mention it's like more likely to succeed because I think that's the kind of openness and transparency, even though the onus is on our companies to be open about it. Like we believe based on where the models are, in the accuracy. This is what we think based on all the information we were able to gather and the comprehensive knowledge the system has, that this is the chance to get more likely to succeed versus not. And having that kind of conversation during the work as well as in the [project] details, so there is less ambiguity and not just feel like, okay, we just got the final product and we don't know, wait a minute, should we trust this or not?

And I think it really comes down to the trust of like, okay, can we really rely on the system's outcomes or do we have to do extensive work to validate the system's outcomes? So that gap, if we can help reduce that as an AI company, I think that could reduce that. The bar is high, but it's also a rational bar rather than, you know, unless just get 100% accuracy, then, you know… It's it's just a very different mindset.

Dr. Jo Varshney

That brings me to my next question, and this has been quite on my mind for a while, and at JPMorgan there was quite a bit of discussion about workforce reduction… The fear of AI adoption will reduce reduce the workforce because we don't need humans. And [this fear] also stops many companies to have an external AI collaboration because they're like, “Well, I don't want my job to be impacted.” So it's a very serious question and it's on many people's minds. And, you know, even though they don't say it, this is the reality because we are humans, after all, we have families and real concerns.

So let me ask you, how did you get fearless in a way and more importantly, how do you see… like do you think this is a real fear, a real threat to the workforce, or is this something, you know, just some vibe right now in the industry?

Dr. Grazia Rovelli

That it's a very tough question. And I feel like I can you know, I can answer this from the point of view of a mid-size pharmaceutical company. So I'm not entirely sure this is going to be applicable to, you know, the big the big giants out there. But from my experience, I don't really see that as a threat to either myself or my biologist colleagues or my colleagues in chemical synthesis and all of that.

If anything, to be honest with you, I feel like if we're able to get to, you know, more and more molecules faster in vivo and in vitro to like further stages closer to the clinical development, I feel like we might need more people. I mean, it's just, you know, as an example, I feel like talking to you guys, we've really been clear saying we cannot do all the experimental work on this [project] in 2024. [Work] will need to change because we need to allocate the resources we have to this faster pace that we expect coming out of AI.

But the other thing I want to say to that, I think it's really important to keep in mind that this is where the field is going. And so if you want to be, you know, up to date with what's out there, if you want to be competitive in the job market, you need to to some extent at least, you know, train yourself, read as much as you can about this, you know, technicals less and more or less technical depends on, you know, the single individual on how much they want to delve into, you know, Python coding or whatever it is.

But I really think this is something that especially I feel like, you know, people my age, you know, people that are approaching this field or have been in this field for maybe like ten years or so, you need to be absolutely open to learn how this this new technology works. Even if it's just like, you know, my company has got this platform in licensing, how do I use it to make my work again more efficient, faster?

And so on And it can be at a very high level, as I mentioned. But I feel like we need to be able we should have an open mind to retraining and learning, continuously learning especially, you know, I feel like as a researcher, one of the things I love most is learning continuously. And I think a lot of my colleagues are like that as well.

Dr. Jo Varshney

I actually share a quite a bit of the similarities of what you mentioned. And I do think that there is a lot of discussions about workforce reduction and improving the efficiency. Like if one person can do the job, you know, why should we hire five and how do we figure that out from the AI perspective. But from specifically our company perspective, we need the experts because we are the enablers for their expertise. And I think that's what you're going with. Like we need the experts because without the expertise, you can't really innovate. And I think of our technology kind of like being a game of chess with a computer, you know. First the machine is learning. But as you learn from the moves of the machine, you are getting better. And we need that also because without that, at some point, some random discovery from an AI system still has to go to the patients.

But and also there's a whole regulatory aspect that we need hours to discuss because, you know, FDA is not just going to be like, “oh this came out of AI, we should just [grant it] testing in humans.” There's there's a lot of checkpoints that we have to go through. And I think that our experts are ready and should get more ready. And from our side, we should make things as easier for folks on the other side to learn and adapt and make it as user friendly, as clear, as comparable to what they are used to.

I think if we can, as an AI industry grow, we all can agree like look, instead of being like we are here and they are there, I think there has to be a better collaborative, integrated methods to really foster more collaborations because it's not just one company, it's every company who is trying to figure out. And then, you know, there are a lot of companies trying to do everything in-house. But and I think there are challenges to that. Of course, there pros that everything stays inside, but then you're missing out on the learnings, missing out on the blind spots because, you know, everybody has favorite targets, favorite patients, favorite everything.

So another question… Most of the conversations that I see or hear around AI are on the target discovery. Chemistry. And you know you coming from chemistry, do we need a lot more on the chemistry side or do we have to get to the biology or are we over indexing on biology or are we over indexing on chemistry? I wanted to hear your thoughts because, you've spent more than ten years in chemistry and how do you see where AI can be most helpful in the drug [development] problem?

Dr. Grazia Rovelli

I have to say we do not engage a lot in targeted discovery, so maybe I don't have the, you know, the broadest overview on that. I think there's there's a decent amount of companies out there that do that. And but I don't have a sense on how risky that is in terms of, you know, like then eventually going into the lab on, on a target that you've discovered with artificial intelligence and you know, and developing something against that target. So I'm not sure I can I can speak to that. 

In terms of chemistry, what I can say is that there's there's a few companies that are doing some exciting things, you know, by coupling artificial intelligence and robot synthesis. I feel like it's scaling up a parallel chemistry or a AI driven chemistry that's kind of exciting. And I mean, still a lot of be shown in that space, but that's something that obviously, you know, you're going to need like a very specialized entity to do that. I'm not sure that you will ever be able to internalize that kind of technology, but that's something interesting. And in terms of also, you know, synthesis driven and drug discovery, I find that quite, quite exciting.

Ideally we we spoke about biology first and chemistry first companies. Maybe there's going to be a super company that does both, you know, to be credible at the same time. And but I have to say, I don't know that I've found that yet.

Dr. Jo Varshney

Yeah. And those companies, I think it's just it's a major investment to get both sides of the equation. Yeah, but you know, we are always excited to see how, how the field of AI really evolves. 

And maybe to that question, since we are biology first, we care like, you know, how does this work in the actual patient population, how often do you see that in from the AI perspective? Are there many companies and are there specific challenges or problems that you think are key things of an interest to pharma company?

Dr. Grazia Rovelli

Something that I feel like is not super common, that I have to say in your technology I really like, is metabolite prediction and, prediction of toxicity of metabolites. So you know, you're thinking ten steps ahead in terms of, you know, not only the toxicity of what you're actually synthesizing, but what's going to come up later through metabolism. So that's something I don't think is super common. And I think it's it's very interesting for me that that you do it. 

Something else is, what if I'm going to do some animal testing in a specific animal model, how translatable, again, is that to humans? Right. Especially right now, we have a series of molecules that is showing some side effect in mice. And so we're wondering, is that going to be  relevant in human as it is in mice? And that's kind of an open question. I don't know that we have found a way to address that quite yet, but that's something relating animal models to human trials. That's something super, super important on the biology side of things.

Dr. Jo Varshney

This is something we care a lot about that translatability it in mice, monkeys rabbits etc. How do they translate into humans? So we'd love to dive in deeper in these different conversations.

This has been fantastic. One last final question. It takes a monumental effort for an organization to change something that works and has generated billions of dollars and creating a new technology mindset. And you are one of the few members in the company… And for the broader audience, just a big shout out for how you become who you are and what other folks can learn from you because I think you have the great mix of the background and knowledge of AI and are always really asking the good questions and vetting the process. So is there any secret sauce you're comfortable sharing with the audience here?

Dr. Grazia Rovelli

I feel like, you know, just be curious, read a lot. I've been watching so many of these webinars, workshops, podcasts on my own. I listen to so many podcasts. I really enjoy learning through this kind of channel. So be curious. And and the more you listen to people talking about this, the more you're going to be able to ask the right questions. I think that's key not only in this field, but I guess in life. I would [also] say awesome networking, super, super essential. The more people you talk to, the more you're going to have ideas on how you could potentially apply certain technology within your own research. Again, super crucial. And then I would say something that I've been thinking about going into this webinar, I really think it is valuable to embrace both the passion from someone like you, from someone who's like super, super passionate about what their technology is and what they're trying to achieve. 

And at the same time also embrace the skepticism of, you know, who is not maybe as into the field and maybe they don't know exactly how these kinds of algorithm works, what they can do for them. And and I think a having a good balance of those two can really, really… I'm not saying it's easy, but I have to say, keeping both those opposite tension in your in mind, I think it's it's pretty important or it has been for me at least.

Dr. Jo Varshney

I really appreciate that you see in me and our companies that we embrace it. One of the first folks [I shared the idea of VeriSIM Life with] seven years ago, or actually almost ten years ago, mentioned to me this is never going to work. What is AI? What is this [BIOiSIM] thing? Had I listen to that I don't think we would be talking to each other [today]. So I think you're right. Like having that balance is necessary. And more importantly, always keeping in mind that the experts have been there, done that, and they've been around to really navigate the process and they have learned from their mistakes. We don't want to make the same mistake. So I think having those open dialogs and having that open mind, as you mentioned and being curious is fantastic advice. Thank you so much. 

So [based on] this conversation, is AI real or is it a myth? How are you thinking about it given you guys are embracing AI?

Dr. Grazia Rovelli

I want to say it’s real, 100%. You know, people are adopting it every day. We're working with it. But still a lot to be demonstrated, you know, a lot to be achieved. But I'm hopeful that it's going to keep its promise. 

Dr. Jo Varshney

Fingers crossed! We are really excited to partner up with you. And hopefully this year or early next year, we should be able to really showcase some of the work and maybe we can sit down and have another discussion and have improved your company’s success success rate. So thank you so much, Grazia. I think this is a good time to segue way to questions from the audience and ask you to join us.

Prefer to watch the video? Gain access to the webinar recording here.