#108: Alex Zhavoronkov | Can AI Fix Drug Discovery?
1 hr 15 minAudio
Can AI make drug discovery more likely to succeed? Alex Zhavoronkov believes that is the question that matters most. Faster chemistry and lower costs are valuable. But choosing better targets, designing better molecules and stopping weak programs earlier could change the economics of building a biotech company. In this conversation, the Insilico Medicine founder and CEO joins Christian Soschner from the World Government Summit in Dubai to explain how his team connects generative AI, biology and experimental validation. Alex walks through an idiopathic pulmonary fibrosis program that, he says, reached the preclinical candidate stage in under 18 months for under $3 million. He explains how Insilico combines target discovery with molecule design, why commercial strategy shapes scientific decisions, and where clinical trials still create bottlenecks. The conversation also explores his larger ambition: developing medicines that treat disease while addressing biological processes involved in aging. For biotech founders, investors and pharmaceutical leaders, this episode offers a practical way to think about AI: follow the evidence from a model’s prediction to a laboratory experiment, a development candidate and a clinical trial. WHAT YOU WILL LEARN Why Alex prioritizes probability of success over speed and cost. How generative AI connects target discovery and small molecule design. What the fibrosis program reveals about the economics of early discovery. How pharma’s future priorities influence pipeline decisions today. Why robotics and experimental validation matter as much as algorithms. How aging research could inform treatments for multiple diseases. CHAPTERS (00:00) Can AI change how we discover medicines? (02:07) Alex joins from the World Government Summit (04:21) Longevity, aging clocks and the search for new targets (08:22) How generative AI designs molecules (15:02) Drug discovery costs, productivity and failure (23:28) What Alex’s success estimates actually cover (24:50) The fibrosis program and an AI discovery case study (36:31) What the $3 million figure includes (38:09) Regulators and predicting clinical trial outcomes (40:33) Alex’s proposal for faster development (44:45) Rare diseases, recruitment and clinical bottlenecks (48:55) Audience question: AI and the microbiome (53:22) Building a global company around talent and research (57:34) Inside Insilico’s automated laboratory (01:01:39) Could a discovery lab fit into two rooms? (01:03:12) Medicines that address disease and aging (01:04:57) Life extension and the future of generative AI ABOUT THIS RECORDING Recorded in February 2023. Company milestones, pipeline status and forecasts reflect the conversation at that time. The cost and timeline figures describe a specific preclinical discovery program, rather than the full cost of developing an approved medicine. CONNECT AND EXPLORE Watch the conversation on YouTube Connect with Alex Zhavoronkov Explore Christian’s work and discuss working together Follow Beginner’s Mind for conversations about turning breakthrough science into companies that last. Send us Fan Mail Support the show Join the Podcast Newsletter: Link Your Next Move: Link
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
Christian SoschnerWhat if I told you that there is a way to reverse the aging process and potentially extend our lives by decades? Sounds too good to be true.
Alex ZhavoronkovSo we made a little bit of a history. Generative AI is going to transform our life beyond recognition in every way. I think that we can at least improve the probability of success by 50%. And that's more important than time and cost. So in two years, uh, well, two and a half years, 11 preclinical candidates, it's a pretty cool number. So that is how long it took us to get to preclinical candidate stage uh with our antifibrotic.
Alex ZhavoronkovUh and again, this is the traditional approach, this is our approach. So under 18 months, uh the ability to very rapidly predict success or failure and also interpret, I think that's uh that's key.
Christian SoschnerWell, today's guest, Alex Taboronkov, CEO of InSilicon Medicine, is on a mission to make this a reality. With the help of artificial intelligence, his team is working to uncover the secrets of aging and create drugs that could not only extend our lifespans, but also prevent age-related diseases like Alzheimer's or cancer. In this podcast, you learn about the incredible potential of artificial intelligence in aging research and how in silicon medicine is leading the way.
Christian SoschnerAlex explains the potential to reduce drug development costs, increase the probability of success of drug discovery, and even create drugs for diseases we once thought were untreatable. Get ready to have your mind blown by the groundbreaking research and insights shared in this episode. Don't forget to hit the subscribe button and join us on this journey to revolutionize the way we age. You are live directly from the world's government summit.
Christian SoschnerDid I get it right?
Alex ZhavoronkovThat's correct. So uh I'm dialing in now from the World Government Summit. I apologize for any background noise uh because I had to step out of the meeting. Uh, and here it's uh that there are no meeting spaces. Um uh so very happy to connect. Uh my name is Alex Jabrenkov. I'm the founder and CEO of a company called Insilica Medicine, uh, which is a uh global company. So that's what that's why I am in the Middle East right now.
Alex ZhavoronkovUh, one of our sites is in Abu Dhabi, and uh which is a fantastic place to set up. Uh and uh right now I'm presenting from Dubai.
Christian SoschnerWhat is the World Government Summit all about?
Alex ZhavoronkovUh it's all about uh bringing together uh world leaders from all around the world to um ensure that they can achieve common goals. Uh they also invite uh business people and scientists to ensure that uh uh they have the support from the uh private and public uh uh sector uh for all kinds of um uh discussions about the environment, healthcare, and other um areas where global collaboration is extremely important.
Christian SoschnerSo it's uh practically similar to the World Economic Forum. Uh it's uh that's just that's correct.
Alex ZhavoronkovThat's the Middle East version of that.
Christian SoschnerYeah, I had the same thought. It's it's a good thing to have that we are part of the summit now with this uh podcast. So it's uh thanks for coming on by Alex. I read in your uh in your bio that starts directly jump right into it and into what you're doing with in cyclical research. Uh I found a lot of uh articles on the internet where you describe that you come from longevity research. I think we have that in common. I did uh something similar, not in the academic field, more in the field of martial arts, because the most important question I had was in the it was back in the 90s.
Christian SoschnerUh, how can I avoid to get sick? And since in the 90s in Austria, there was not many people really interested in that topic. I switched more to martial arts to find how I can improve my life, and now it's called, I think, longevity research. Uh, what did you do in this field?
Alex ZhavoronkovSo I am uh participating in this field more from the academic perspective uh rather than for from the perspective of uh self-improvement. Uh I think that longevity in general is probably the most impactful field anybody can be focused on, right? Because uh uh if you figure out a way to extend the life of uh people just by one year for everybody on the planet, you generate eight billion life years, right? So that's more than any surgeon or a doctor can uh um achieve in their lifetime.
Alex ZhavoronkovSo I am focused on the discovery of novel therapeutics uh that may address certain hallmarks of aging. So we look at both uh um uh we look at both uh protein target discovery, so identifying the uh various uh possible protein switches that might be important in uh uh aging and longevity. Uh and then we also have uh a way to design um novel therapeutics, novel small molecule drugs that go after those targets.
Alex ZhavoronkovUh so we have uh kind of two hands. One uh searches for promising areas to intervene in, and another hand which uh allows you to very rapidly generate uh chemical matter um uh that can be turned into a drug. Uh and for that, I study um and I uh contribute to research in artificial intelligence that uh works both in biology and chemistry. Uh, we have a platform called Pharma AI at EnSilico that helps accelerate uh uh uh drug discovery RD.
Alex ZhavoronkovBut many of the programs that we go after are in age-associated diseases and longevity. Uh we also contributed quite uh a bit to the uh development of a variety of what is called what is called aging clocks, um the uh biomarkers of aging utilizing artificial intelligence. So you can actually train train deep neural networks uh um or other uh AI models uh that allow you to predict chronological age in a healthy state of the patient, let's call it uh biological age, or you could uh predict uh um uh subjective age, so how old do you feel, uh, or even uh some other features uh like for example perceived age, uh how uh old do other people think you are, right?
Alex ZhavoronkovIf you are talking about uh imaging biomarkers or psychological biomarkers. So we can look at many, many different data types to build those comprehensive age predictories. Uh, and later we can um design certain interventions for uh biomarkers that could yield valuable targets, either protein targets or even sometimes non-biological targets. Um looking at both again interventions uh and um uh diagnostic applications.
Christian SoschnerThat's a great thing. Um, I found your company when I was doing uh some research on LinkedIn about another topic, ChatGPT, and then I came across an article that's uh Regina Hoditz uh liked where you basically compare your solution in drug development and drug discovery to ChatGPT and DALI from OpenAI. Uh and the thoughts that I had in my mind was so now that I know how to work with ChatGPT and it can produce articles and questions and stuff like that, uh, is it really possible to have an artificial intelligence where you just enter the questions that I want to cure cancer and the artificial intelligence then spits out the drug and the target and everything is solved?
Christian SoschnerWhat is the state of the art in uh generative artificial intelligence in your field? So sure.
Alex ZhavoronkovWell, uh again, generative AI as a term has been around for quite a while. Uh, I think that the term generative artificial networks, um uh generative sale networks, sorry, was uh again was coined by Ian Goodfellow in 2014 uh in a very famous paper that he co-authored with Joshua Benjo, probably the highest-sided uh researcher in um uh AI. Uh and uh um uh in that paper, they've described uh how to utilize generated the serial networks to uh imagine handwriting written characters, for example.
Alex ZhavoronkovSince then, in just a year, the technology has progressed to the level where you could generate pictures uh that are of very high quality, where you can describe uh um what you want to see, and GAN will create it for you. Uh, that was the year when we got into uh generated the serial networks as well with molecules. So just like with a picture or uh with a handwritten character, you can uh describe the molecule you want, and the GAN will generate it for you with the desired properties using multi-parameter optimization techniques.
Alex ZhavoronkovUh and uh then since 2016, this technology has progressed uh pretty dramatically, right? So in 2020, you could generate pictures of people that are indistinguishable from photographs, right, or almost indistinguishable. Uh and uh nowadays uh with these tools like DALI and Mid Journey, uh generative technologies, uh now also transformer neural networks and other forms of generative AI, uh became consumerized.
Alex ZhavoronkovSo now consumers got access to those tools that we were using professionally for quite a while, uh, and they have propagated uh into many industries. Uh so Midjourney and DALI, for example, they're now used to generate very valuable uh images uh from a single prompt. Uh ChatGPT is now outperforming humans in some of the writing tasks, especially when language is not your first language.
Alex ZhavoronkovUh but in um our field, in drug discovery, this technology has progressed quite a bit. So we uh synthesized and tested our first AI-generated molecule in 2017 that led to some considerable investment, but also uh established us as a player in this field. Uh, so there we uh designed small molecules against a specific target. So you basically take a protein crystal structure and tell your AI to design molecules that would be very selective uh and active against that specific protein target, uh, and also could be drug-like.
Alex ZhavoronkovSo something that is not a poison and uh can get into the um uh into the body uh via uh oral administration, so not injectable, for example. Uh and we've uh synthesized a number of those molecules, tested some, uh, and uh published our first paper 2018, submitted 2017. Uh, and then in 2019, we demonstrated a race where uh another company challenged us to design small molecules against a very well-known target in fibrosis.
Alex ZhavoronkovUh, and we very rapidly generated small molecules, tested them against the target, uh, and demonstrated that the generation conditions could were confirmed using experiments. Uh, so that's 2019, and then we productized it. So we developed software around it with many, many, many different generative models uh for chemistry and for biology, and released it to the pharmaceutical companies. And now 10 out of the top 20 pharmaceutical companies have used our tools in one way or another.
Alex ZhavoronkovSo it's actually pretty widespread uh in drug discovery. In longevity, uh generative techniques, uh, we used them for the first time in 2017, where we used age as a generation condition. So that's where you can um uh generate very high-quality synthetic biological data for a person that you that does not exist, right? So, for example, you can take yourself as a template, say, okay, well, I'm Christian.
Alex ZhavoronkovUh, here is my gene expression profile, and I want to see how this gene expression profile looks like in 20 years. So you say, okay, well, generate according to this gene expression, uh, 1000 versions of Christian uh skin, for example, um uh and show me different ages, right? So uh from let's say you know 40 to 50. And uh the system will generate a bunch of those gene expression profiles.
Alex ZhavoronkovUm, you as a human would not be able to interpret it, but you would be able to uh trick pretty much every AI system that is trained to predict age using gene expression, right? Um, and this approach allows you to also identify valuable uh targets, valuable genes uh that makes your gene expression profile younger or older, right? So just by basically um uh moving this dial, uh, which is age back and forth on several biological data types, you can actually get valuable insights into what's making you older or younger, and also try to find causal relationships.
Alex ZhavoronkovSo um I think that generative AI will also help quite a bit study aging. We found some targets this way already. Uh, and um we currently have the first drug that we've designed uh using generative AI for a novel target uh already in human clinical trials. Uh so we just uh got the top-line data from the phase one clinical study where we said where we test safety. Now we're going into phase two for a very broad indication with no cure.
Alex ZhavoronkovUm, but the way we found this target was also using aging research. So I think generative AI in uh uh biomedicine is very well established and very broadly used.
Christian SoschnerThat's that's great. I had um a couple of weeks ago, I had a conversation with Jack Scannell. He coined the term IROOM's law uh in 2010. And we were talking about the big problems that we have in drug discovery and drug development. And I would like to hear your opinion on what generative AI and artificial intelligence can do. Uh, I mean, the basic principle you also said in some interviews that 90% of drug candidates fail in clinics. And when we look from uh coming from the lab bench, we are probably about 99% failure rates.
Christian SoschnerAnd amazingly, the expenses for one truck in the last 10 years went up from roughly $1 billion per truck uh to three to four billion dollars. Uh how do you see that artificial intelligence or how can artificial intelligence help to solve these problems to make uh drug development faster, more accurate, and uh also cut back the costs? Where is uh what is the role of generative AI in this field?
Alex ZhavoronkovSo sure. Thank you for asking this question. Uh, because I would be able to answer first in numbers, right, with real facts, and then give you a little bit more of a futuristic view. So, as you know, our company uh we raised uh just over 400 million, but mostly uh it was in the last two years, uh so from uh mostly biotechnology investors who understand biotech. Uh, and um last year, just in 2022, we nominated nine preclinical candidates.
Alex ZhavoronkovSo eight internal and one for an external collaborator. Uh and in small molecules, that's a very large number. Usually a big pharmaceutical company would nominate uh, you know, four, five, six. Uh sometimes when they're lucky, they can go. I don't know a big pharma company uh that nominated 10, for example, in a single year. Um, it would be uh you you usually they have a lot of programs, but many of them are external and many of them are in biologics.
Alex ZhavoronkovSo in small molecules, uh you usually don't have this space of RD. Uh usually to get to this point, to a point of preclinical candidate, you would uh with all the failures, you would probably spend uh uh, well, is a biotech probably around 100 million dollars, right? If you're going after new targets. Uh and uh in pharma, it's gonna be even more expensive because they they they usually waste a lot of time and effort on uh uh all kinds of bureaucratic things, right?
Alex ZhavoronkovAnd just people do not focus and work as hard and fast as in biotechnology companies. So this improvement in uh and and acceleration in uh uh drug discovery is real, so it's not just blah blah futuristic, right? So nine preclinical candidates in a single year, it's we've demonstrated that. In 2021, we nominated only two preclinical candidates. Um, and this year we're expect to uh you know hopefully exceed our internal record.
Alex ZhavoronkovUh and some of these uh drugs went into clinic. So usually by this time the failure rate is very high. Uh and again, we raised only 400 and we managed to um uh get to pre-clinical candidate stage with now since uh we started doing our own discovery, um we managed to uh progress 11. Uh it's a pretty big number. And uh you can compare drug discovery to space exploration, right?
Alex ZhavoronkovSo first you need to launch your first uh um uh first rocket, right? Reach the orbit uh and or you know suborbital flights, uh, but still with some payload delivery. Uh and the most difficult one is your first, right? Or your second. But then when you demonstrate that it consistently works, uh you can do much more, and also people start believing in it. So I think generative AI already transformed drug discovery, and we have contributed significantly to this process because we also, in addition to uh using generative AI internally, we also gave it to other people in a form of software.
Alex ZhavoronkovSo now a lot of people are using our software. Uh for target discovery, it's so easy to use that high school students can use it, and they do. Uh for small molecule generation, uh, it is not as easy to use, uh, and we only deploy it within big pharma companies. And but they those that understand the value, instead of you know trying to say, oh, we are we we we can do a better job. Uh by the way, if some some some of them say that, they can also implement some of their models in our software so it's so flexible.
Alex ZhavoronkovUh, but many of them are afraid of doing that because then um if you don't outperform, well, you would have to rely on our software. Uh uh usually it's a good good idea to do both. Uh, and um I think that the cost component and the time component is not as important as the probability of success component, right? Because when you are using generative AI for both target discovery and small molecule chemistry, I think that you can uh uh significantly improve the probability of success because very often you would go after maybe not so hard targets, but when you have a choice of a target uh and you can very rapidly fast forward, um you can find those combos uh that have not been previously explored.
Alex ZhavoronkovUh, because either people thought that it's uh maybe you know uh not a very competitive target, uh, or it's a very difficult target, or this target might not address uh a substantial patient population, or it's just too new. So they try to ignore this target because uh there is very little biology and biological evidence available around this target. And we manage to matchmake uh the target with uh the generated molecule uh in a most efficient way.
Alex ZhavoronkovWe also try to forecast into the future and see which targets are going to be hot because many of them are made for sale for big pharma companies, right? So we would not be able to progress many of them ourselves into phase three, for example, right? It just we won't be able to raise so much capital. Uh and um when you are trying to do those three steps with generative AI, I think that. That you are significantly improving the probability of success, even from the strategy perspective, because many of those uh programs within big pharma companies they fail because of strategic choice.
Alex ZhavoronkovSo they actually just decide to pivot uh their strategic direction and refocus, for example, from oncology to CNS, uh, or from you know one target discovery philosophy to another, they want to say to okay, we're gonna go only for uh genomically validated targets, and then you know 70% of the pipeline is stopped or you know cut, uh, and they just refocus, and a lot of that is uh uh wasted due to um strategic reasons.
Alex ZhavoronkovSo when you also try to extrapolate uh the kind of hotness of the target space into the future and try to see what is going to be valuable for big pharma companies uh from the strategic perspective, uh you can also minimize that risk. So I would be uh guessing right now, but I think that we can at least improve the probability of success by 50%. And that's more important than time and cost. And that's at least, right?
Alex ZhavoronkovBecause for us, internally, pretty much every program we tried uh has succeeded. So we failed a few times, but more than 90% of the time we succeeded to get to a preclinical candidate and uh into the clinic.
Christian SoschnerUh 50% the probability of success from in which stage are we talking about clinical uh trials or are we uh already earlier?
Alex ZhavoronkovSo it uh I'm talking about earlier, much earlier. So from target discovery stage all the way to let's say phase one clinical trials. Um and the reason I can say that is because currently we have just a few of those programs, right? And uh uh so far the probability of success was almost 100%, right? So um 50% reduction in probability of success, uh in probability of failure is um uh my personal conservative estimate.
Alex ZhavoronkovAnd of course, we did it on a tiny fraction of the cost uh and in a fraction of the time. So in two years, uh well, two and a half years, 11 preclinical candidates, it's a pretty cool number.
Christian SoschnerThat's um that's really amazing. So for my understanding, um the numbers that I have in mind are that uh when we look earlier before pre-clinics, we have a success rate of 1%, and uh your programs have a success rate of 100%, basically. And I understood you right. So it means that uh you beat everything out that doesn't work very quickly and move only those compounds forward into pre-clinical development and clinical development that have a real high probability of success already. Is this the right understanding?
Alex ZhavoronkovYeah, well, that's the that that's that's uh I would say a little bit more ambitious way to explain it. Uh but um why don't I visualize uh and share a screen? I just need to have uh screen sharing privileges. Um uh and I'll just show your um followers uh one slide, which is very important. That's a good idea. Uh you got the right. So here you go.
Alex ZhavoronkovUm that is the slide which describes this from Stephen Paul's paper, um, uh formerly at LA Lilly when he published it, uh, in Nature Reviews Drug Discovery, very famous journal, uh, which describes uh uh these different steps of uh drug discovery and development. Uh and on his paper, he described just those steps from target to hit all the way to submission and launch, um, and showed that it costs about $1.8 billion, right, from target to hit. And here are the probabilities of success on the top.
Alex ZhavoronkovHere is the number of years it took. Uh, it usually takes. And here is on average uh the number of millions of dollars that every step costs. Uh I added my own estimates to this slide, uh disease hypothesis and target discovery. That's usually performed in academia, not in big pharmaceutical companies. They very rarely discover good targets that get progressed into um late-stage clinical trials. And here the probability of success I estimate it to be one to five percent, predominantly because um currently the NIH, National Institute of Health, uh, their budget annually is $45 billion, right?
Alex ZhavoronkovSo they actually spend quite a bit of money, but there are very few good targets that uh reach the clinic. And usually it takes one to 10 years, uh, could cost you billions of dollars. For example, Alzheimer's, we still don't know a single good target that would work for everybody, right? Uh and the disease hypothesis is poorly understood. That's why we have to study aging to understand Alzheimer's. And Alzheimer's understanding may help us understand aging a little bit more.
Alex ZhavoronkovUm, so what we have done so far is we've taken one of our programs uh from discovery to phase one topline data received, right? So basically we can refer to it uh as phase uh one complete. Um and uh we've done all this uh for for this program. We haven't failed yet. So fingers crossed. Uh we did it in um uh under three years, right? So usually it would take you um uh you know five and a half, six years, sometimes a decade.
Alex ZhavoronkovUh, and it would take you a lot of money, right? So basically close to a billion. Uh we have done that on a fraction of the cost. So to reach the preclinical candidate uh stage at that time, um uh just for idiopathic pulmonary fibrosis application uh indication, it took us about three million dollars. Um of course we have tested it for many other indications, right? So we're taking it also for chronic kidney disease, that specific target.
Alex ZhavoronkovUh and uh um we're doing indication expansions into skin fibrosis as well. It's one fibrotic target. Um, but just that is a pretty cool number. Uh and let me see if I can um show some other interesting slides just to explain uh this process. Um and again, I will have to apologize one more time to your uh listeners for the background noise because I'm at the World Government Summit.
Alex ZhavoronkovUm so that is how long it took us to get to preclinical candidates stage uh with our antifibrotic. Uh and again, this is the traditional approach, this is our approach. So under 18 months, under $3 million for IPF only. Uh, and then um under 30 months uh into close to phase one complete when I made these slides. Now we're we're here. Um and again we did it at a fraction of the cost.
Alex ZhavoronkovTo my knowledge, is the first uh AI generated drug to reach uh human clinical trials and uh um go through the phase one clinical trials. Um so we made a little bit of a history, hopefully. Okay. Uh and and and and uh even if we you know do not succeed in phase two, it was uh a very um uh cost-effective journey compared to everything else. Uh, but we of course are aiming for uh uh major success.
Alex ZhavoronkovUh and uh so this is the first drug where we managed to get uh all the way to humans. We actually conducted um a phase zero study. So a small study in a small number of uh humans, eight humans in Australia, uh healthy volunteers. We call it phase zero, because again, the target is new, right? So before and we are a small company, we wanted to ensure that we understand the target uh and the compound distribution a little bit better. And then we start uh uh phase one trial uh in 80 healthy volunteers, just of the top line data.
Alex ZhavoronkovUm, and these are all slides, but we uh we have um progressed with our target X. We haven't disclosed it, and we have several other very promising targets, uh including targets for COVID. So for C3 like protease, uh that one is ready to go human right now as well. Uh and it's a very selective, broad spectrum molecule. So it works pretty much in every um COVID strain that we've uh tested it against. Um, and we also have a lot of cancer targets uh that are specifically tailored for um licensing to be pharmaceutical companies.
Alex ZhavoronkovSo we try to have very high-quality data packages that would be very attractive to um uh to those kind of companies. And just to show you a case study um uh that we published openly for everybody, so you can go back and trace our work. Um and we try to publish uh as much of our work as possible uh uh open uh with open access. Um so we went after ALS, which is a very popular disease actually for all AI-powered drug discovery companies.
Alex ZhavoronkovUh, but we try to do it in the open. Um it's a rare disease, but uh there are still a lot of patients uh out there. And very often when I speak with uh my friends, somebody knows somebody who had ALS or has ALS right now. Um we decided to use all of our uh software tools, Panda Omics, Chemistry for Intuit, and Clinical, that are available uh commercially right now for target discoveries, small molecule chemistry, and prediction of clinical trials outcomes.
Alex ZhavoronkovUm and utilized Panda Omics, uh, where we utilized massive amounts of uh data uh available to us from public repositories for ALS and other diseases. Uh we used multiple uh AI algorithms and AI models uh to come up with target hypotheses, also got some textual evidence that some of those targets may be implicated on ALS, uh, explored various drug filters uh uh for for uh exploring targeted uh uh target uh whether the target is good for small molecules or uh uh biologics, uh whether it's commercially viable.
Alex ZhavoronkovAnd then we partnered with wonderful key opinion leaders in this field, Mary Chutkovitz, uh head of neurology at um uh at uh Harvard MGH, Jeffrey Rostein, uh head of the Brainst uh project, uh, and uh very prominent neuroscientist, uh neurobiologist from uh Hopkins, uh Bailu uh from Tsinkhwai University, and Kia Jong uh from Mayo Clinic uh to work with massive data sets from ANSER ALS Consortium, uh, where uh a lot of human um ALS cells were reprogrammed into baby cells called IPSCs, uh, and uh uh genetic profiles were uh evaluated, uh, and also they were later programmed into full-scale neurons, also with uh all kinds of molecular data coming from the samples.
Alex ZhavoronkovUm, and also we used a lot of samples from our own uh databases that are properly uh integrating and annotating massive amounts of public available data, uh, ran it through Panda Omics and found new targets and also older targets with drugs to repurpose. And then we tested those drugs very rapidly in a fly model from Mayo Clinic. Uh so you know when it comes to rare diseases, um, and especially the diseases of the brain, there are very few uh models where you can test your hypotheses, right?
Alex ZhavoronkovSo you can test it on mice, you can test it in human neurons, you can try to grow them, uh, or you can uh use a model organism uh like uh flies, for example, and do very quick tests. So we very so you find you can find more information about this project at als.ai. Uh we use this fly model from Mayo uh and demonstrated that uh in uh uh uh occual fact ocular neurons uh where um the uh G4C230 phenotype is observed, very similar to um many ALS uh um phenotypes that are observed in humans.
Alex ZhavoronkovUm many of the drugs that we predicted, uh many of the targets worked and rescued the phenotype. So AI can pick the targets and very rapidly demonstrate strong rescue, moderate rescue, and mild rescue. And a few targets, of course, didn't work, right? Um and we've published it openly. Uh, and some of those uh drugs, some of those targets, they do have uh already existing drugs. Uh, and we've published those uh for repurposing.
Alex ZhavoronkovI know that uh there there are some companies that are not trying those drugs and targets lemocly. So uh that's one of the ways to test your hypothesis very quickly. And yeah, then we actually did our own chemistry, and uh, I'm gonna pause here because uh that's gonna be a long story and uh with a lot of uh chemistry terms, and maybe it's gonna be boring for your readers, for your listeners. Um, but uh we managed to demonstrate that we can very rapidly with AI identify new targets uh uh and validate those targets either using uh knockouts uh or real chemistry, uh, and for a very uh important disease with no cure uh and then published uh rapidly, right?
Alex ZhavoronkovSo we didn't even um think about commercializing this uh when we published. Uh and for many other diseases, we do something similar, uh, but we do it internally.
Christian SoschnerIn the in the first slides, if I remembered right, you said that uh you cut back the time that you need to bring uh a drug candidate into the clinics and uh instead of an ERA map, right? It was about 414 million in a traditional way that uh a company needs to move a company phase uh a track candidate into phase one. And with your technology, you did it for three million dollars. Is this uh did I remember that right?
Alex ZhavoronkovThat's correct. Wow. So we managed to get to uh preclinical candidate stage just for IPF indication because uh there you need to to to claim a preclinical candidate in our world, uh you need to demonstrate efficacy in uh several models, so at least three. Uh and uh um we used uh one in vitro model and two mice models to demonstrate efficacy in uh uh for for idiopathic pulmonary fibrosis.
Alex ZhavoronkovFor other types of fibrosis, we of course uh did many, many other experiments that increase the cost pretty dramatically, yeah.
Christian SoschnerSo, but for uh proving what artificial intelligence can do in track development, I think it's a good example because you cut back the cost to 1%, or you can uh do hundred tracks uh at the same price of one with the traditional way. So it's basically uh it's a very amazing feat uh feat. What was the response from regulatory regulatory authorities when you approach them with uh your artificial intelligence uh approach? Was it uh an operator for them to say, okay, it's uh it's cool, move forward, or did you have some challenges to solve?
Alex ZhavoronkovSo, of course, uh we got some interest from the regulators, especially in Asia. So uh in uh China, the NNPA was very interested in our uh results and uh uh we're in active discussions. Uh and uh in the US, uh we we got some interest. So I actually did uh gave a talk, uh um uh virtual talk at the uh to the to the FDA uh to one of the big departments, uh, and we are trying to establish this connection.
Alex ZhavoronkovUh and basically there our program one, which is uh um going to human clinical trials phase two, is going to be the highlight, right? So that's what we need to focus on right now. Um and uh have the tool that we really want the regulators to use. It's called InClinico, which predicts the outcomes of phase two to phase three transitions and also interprets some of the um uh likely failures and successes, right?
Alex ZhavoronkovUh so we're trying to uh um make this tool available to the regulators as well. Currently, we're selling this to um uh hedge funds and banks, to the financial industry, and also piloting with them to uh improve the tool. Um, because of course uh the industry that gets uh hurt the most uh by failures is the uh the VCs and and hedge funds that are investing in those uh uh and banks that are investing in uh in the companies uh that are you know running phase two clinical trials, that's where most of the failures are.
Alex ZhavoronkovUh and they usually either win a lot or or lose a lot, right? So um uh the ability to very rapidly uh predict success or failure and also interpret, I think that's uh that's key to uh the financial industry, but also to the regulators. Because very often clinical trial uh phase twos actually succeed, but uh later they fail in phase three because the clinical study was mismanaged somewhere, and our tools uh again hopefully will be able to uh allow to identify those mismanaged mismanagements.
Alex ZhavoronkovYeah.
Christian SoschnerWhat do you think in future is it possible to completely um avoid clinical trials to just predict uh the success with artificial intelligence and not do human trials anymore uh unless we are in phase three, or uh is it uh just uh no, you won't be able to do that.
Alex ZhavoronkovYou need to do human clinical trials for sure. Yeah, uh even uh with something. So right now there are pretty much no medicines that work 100%, right? And uh many of the medicines that are that you think work, they actually work for the wrong reason. So uh I think that until we figure out how to simulate this entire world at the atomic level and also recreate it with many, many different people, we will require human clinical trials to be conducted.
Alex ZhavoronkovI just hope that the regulatory authorities will become much more aware of uh generative or of the potential of generative AI and actually recent achievements and start paying closer attention to um what we do, for example, uh, and try to get more involved and try to accelerate, right? So I think that uh currently the regulators would play a much more active role at accelerating uh the pace of technological progress.
Alex ZhavoronkovAnd very often we should actually think about uh how to uh make really rapid pilot runs where you try to set records in certain areas, right? So if I were uh a position right now, uh focused on um you know improving the pharmaceutical uh industry, I would try to identify a few opportunities, for example, in rare diseases, where AI could be used end-to-end.
Alex ZhavoronkovAnd I would try to partner with a few companies, well, for example, like our ours or with ours, to try to go from discovery to uh the clinic in the shortest amount of time possible. So you basically design a strategy, you say, okay, well, this is the disease I want to go after, preferably with with a target that is not obvious, uh, for example, like ALS, and uh then give us green lights to conduct uh rapid clinical studies with novel small molecules and observe how uh those studies span out and also allow us to get the approval as quickly as possible, right?
Alex ZhavoronkovAnd for example, if you can tangibly demonstrate that you can do this within you know four years, for example, well, you just cut down the uh time by you know 60 percent, and demonstrat you can demonstrate that the effective drug can go into uh humans, uh and I'm talking about to the to the market in a certain amount of time, and then you can benchmark against that. So you need to have a few of those uh you know going to the moon uh cases, preferably in uh smaller scale for more rare diseases, but with poorly understood biology, um in a race format, just to get uh better at uh um uh just to get better at uh you know regulating for bigger bigger diseases, right?
Alex ZhavoronkovUh so I would try to do that. And currently I don't know of any politician or regulatory authorities that would be looking at this at scale.
Christian SoschnerNow it's interesting. I mean, uh with large-scale studies in fields like COVID, for example, are 40,000 patients or 40,000 people with these. I was involved in two companies um who did research in rare diseases, and the timelines in clinics were just uh not very attractive to VCs. I mean, when I approach VCs and say, look, we need five to ten years to recruit enough patients, um, they are not very happy with that timeline and uh are reluctant to invest. Where do you see artificial intelligence in in the field of clinical trial to really speed up the I mean when you don't fancy the patients?
Christian SoschnerWhat can you do with that in artificial intelligence?
Alex ZhavoronkovSo sure. Well, again, if you go for China, for example, some of the disease, rare diseases are not so rare. So you can do um uh faster recruitment. And I actually think that uh for this kind of exercise uh in AI, uh that's one of the areas where international collaboration could help a lot, right? So currently we see a lot of uh what we call you know trade wars uh between US and China specifically, uh and actual real wars, uh where Russia invaded Ukraine, for example.
Alex ZhavoronkovUm but with the uh with rare diseases and races for AI part in AI part drug discovery, I hope that authorities could collaborate and maybe we could recruit some of those patients uh much quicker. Um and also I'm pretty sure if the uh um you know an agency like the FDA would uh prioritize uh RFL uh in a in a bigger way and specifically uh you know look at how to um uh at how to uh conduct such a such an experiment, right?
Alex ZhavoronkovI'm pretty sure we could come to a conclusion very quickly, would require just one day of a seminar uh or or of meetings, workshops to come up with a proper strategy. Um I'm pretty sure that we can also get some significant financial backing from the private sector, because uh if um uh the the regulators would demonstrate that at the end there is some potential uh profit involved, right?
Alex ZhavoronkovSo uh the VCs who would be funding something like this would have a big payout, plus uh a seat at the table for future uh races like that. I'm pretty sure we would be able to do that, but currently nobody is uh motivated enough to uh uh to go after those races. Of course, the COVID case was probably the best demonstration of uh abilities and also draw drawbacks and uh inefficiencies in the industry.
Alex ZhavoronkovSo even the first drug that has been uh properly approved, uh Puxlovid, uh, so Pfizer drug, um it's still a very old molecule that is very rapidly repackaged uh as an oral um uh as an oral C3-like protease inhibitor with another drug together that helps with uh you know SIP inhibition with better um stability of the molecule. Uh so it's definitely an imperfect uh drug that probably wouldn't go through in a normal environment.
Alex ZhavoronkovUh but during COVID, you've seen that they've managed to basically conduct all the clinical work in uh a year and a half, right? So that is something that the regulators are capable of. So now let's try and do that with a novel target, novel molecule, right? So why don't we uh try to increase the novelty and do another race? Uh not wait for another pandemic.
Christian SoschnerYeah, I know the pandemic was amazing. I mean, to see the whole industry move so quickly. We have one question from the audience. I will uh uh allow M Kayesh. I hope uh I speak the name right. Uh maybe to ask the question himself. Uh Kay, you have the micro. Um would you like to ask your question? Hi.
Azzam KayassehUh I'm really enjoying uh talking. I am Azam Kayase from Dubai. Uh, thank you very much. Uh I'm really uh I am proud of you, what you are doing. Uh uh this is I asked you before, but I want to know after I listen to your lecture. Uh uh artificial intelligence microbiome uh-based therapeutic tools. Can we apply what you are talking about on the microbiome new molecule?
Alex ZhavoronkovSo absolutely. So if you have a very good microbial uh target uh so expressed by bacteria, uh RAI can perform as well on this target as it does on human targets. As long as you have the crystal or uh you have a um uh predictive crystal structure from alpha fold or a zetafold or some other algorithm, um, we could do similar things on microbial targets.
Alex ZhavoronkovPlus, um uh we could look at new target discovery uh utilizing very similar tools like Panda Onyx. And uh what we did can do uh specifically um so about 2017, I trained the deep neural network uh to predict human chronological age in a healthy state using uh microbiomic uh data using whole genome sequencing data of human gut microbiome.
Alex ZhavoronkovUh and we got to reasonable accuracy for prediction of uh human age. Again, that's one of the features that is present in pretty much every data set, right? Uh and we demonstrated which bacteria are more senopositive and senonegative, so contribute to uh you quote unquote looking younger to the deep neural network or quote unquote looking older uh to the deep neural network. Uh and that showed you that uh deep neural networks can actually zoom into specific uh bacterial species uh that may uh perform certain uh certain roles, play certain roles in uh different biological processes.
Alex ZhavoronkovSo in this case in aging. Um there, of course, the causality is poorly understood, but we can at least uh narrow down to specific bio uh specific microbial species. Um and again, chemistry is chemistry, so uh in in chemistry we can develop small molecules, uh discover small molecules for bacterial targets, um, no problem. Uh for for for disease target research, uh, we can possibly narrow down to individual bacterial species, and then within bacterial species, uh we could identify possible targets just by looking at what uh um proteins are important for the various uh biological processes within bacterial species.
Alex ZhavoronkovSo there are many ways to integrate AI uh in drug discovery using microbiome. Uh and there are many companies uh, in addition to ours, uh, that are actually doing maybe even a better job at target discovery using microbiome. I personally do not prioritize uh microbiome uh as a data type uh within Silico because we have many, many other data types to focus on. Um it's of course promising, uh, but I think again, for from the drug discovery perspective, currently we are prioritizing other data types.
Alex ZhavoronkovSo we like to have uh human proteins as targets.
Azzam KayassehThank you. Thank you very much. I'm looking to work with you or under your umbrella. Thank you very much, Wilden. Thank you.
Alex ZhavoronkovWould love to explore any collaborations, thank you.
Christian SoschnerTalking about collaborations, uh, you mentioned that your company operates in the Middle East, in China, and uh I also think in the United States and in Canada. Uh, why did you decide to go this way?
Alex ZhavoronkovSo we try to go uh after the best talent in the world. Very often we hire through competitions. Uh and uh in Europe, for example, we used to hire through massive national competitions hackathons. Um and then uh, for example, when Russia invaded Ukraine, we realized that much of this valuable talent from Europe could be moved to the Middle East. So I uh managed to get quite a bit of the quote-unquote AI refugees um uh originally into Abu Dhabi.
Alex ZhavoronkovAnd then we realized that that's a great place to uh uh actually uh host talent, specifically AI talent, and we started hiring, and we grew the team uh to about 40 uh here. Um and that's one of the reasons for international expansions, expansion into the Middle East. Uh China is a very special case, so much of the biomedical research uh and contract research infrastructure is in China.
Alex ZhavoronkovSo there over the past 20 years, they invested over half a trillion dollars into uh biomedical infrastructure and research infrastructure. So you don't uh need to own your own labs, you can actually hire uh contract research organizations to do it for you. As an and for an AI-focused company, uh not the the absence of this uh uh labs uh is actually a competitive advantage.
Alex ZhavoronkovSo you have to, if you want to move quickly, you can you can actually move very quickly by outsourcing many, many different experiments to many different labs, uh, do it in parallel, and also have redundancy, right? So very often one lab may screw it up, another lab may succeed. Which one do you go with, right? So then you repeat the experiments in both. Very often, if you do it internally and if you succeed, um in places where you should not succeed, uh, and you proceed with the program, uh, it will be very damaging in later stages.
Alex ZhavoronkovSo we try to introduce quite a bit of redundancy when we are talking about uh um experimental data. So China is very important, it's a very important place to do the experiments. Uh, and uh if you do have resources there overseeing um the large number of full-time equivalents within contract research organizations, you can get higher quality and faster speed and even uh create an ecosystem of those contract research organizations that is very seamless and is better than your than having your own lab.
Alex ZhavoronkovOf course, we do have our own fully robotics uh laboratory as well, but uh relying on those CROs is key. And of course, Montreal, Canada is a great place to hire AI talent. That's the capital of artificial intelligence. One of the capitals, uh the second one would be Toronto, um, so still Canada, so proud to be Canadian. Uh and uh um uh and uh Taiwan, for example, great part of China where um you can hire really amazing computational chemists.
Alex ZhavoronkovUh and in Hong Kong, we do have our target discovery group, which um utilizes a lot of RAI tools to discover new targets. But actually, it's really unusual to have this geographic distribution, but for us it works and it makes us a much more resilient organization. And I think that this level of diversity helps us innovate uh at a broader scale.
Christian SoschnerYeah, and you can uh make use of uh the best from all countries. So it's uh it's pretty amazing. It's pretty amazing the approach that you have. It's uh I like it. You mentioned before you have a fully intelligent robotics lab. Can you talk a little bit more about that?
Alex ZhavoronkovSo sure. Uh I already tested if I do show you the video uh via Zoom, it's not gonna be of good quality and it's gonna lag. Uh so I'll just talk through it and maybe you can post a link to this lab uh in comments. Uh so we we we we realize that now that we have demonstrated that AI can very efficiently identify new targets uh that actually get validated using real experiments, we can now do it at a higher scale, at a broader scale.
Alex ZhavoronkovUh and um we wanted to automate this process of target discovery. So we are not reliant on this incremental data for uh target discovery, but we can um now perform experimental validation of our algorithms much faster. So we have a lab where you can throw in a sample. Imagine that uh a sample comes from some animal model of disease uh of disease, right? So you take uh a piece of animal tissue, I just don't want to use the human analogy, um, where you've got uh uh, for example, certain cancer of a specific tissue.
Alex ZhavoronkovYou throw it into the robot, the robot picks it up, uh, grinds it, microplates it, does quality control analysis, then sends it automatically into another room where part of the sample is being placed in the incubator. Uh, so you grow some of those uh cells uh or even primary tissue or store it. Uh and but part of the tissue goes into the um uh high-resolution imaging system, so it's getting image.
Alex ZhavoronkovUh, you also um incubate uh uh some of this tissue with contrast markers, so you get uh high content uh imaging with uh um fluorescence, so part of the sample gets destroyed, uh then it gets sent to another room uh where you get uh uh omics data, so transcriptomics and uh methylation data uh and several other data types. All this data goes back into AI. AI decides which targets are important in that specific um uh sample uh and which uh drugs already work on this target, picks those drugs from the libraries uh and tests them in a high performance manner, uh different incubation times, different concentrations, and then every sample goes through the same exact uh uh cycle.
Alex ZhavoronkovSo it gets imaged, uh it gets well, high resolution and also uh fluorescence imaging, and you get uh um epigenetic data, and you get uh um uh gene expression data and plus other data types. And you then you compare your predictions to the actual experimental data and see if the target if the targets were found and if the compounds work. And in parallel, you also try to uh uh find some novel biology, so find some novel biological hypotheses that might be uh valuable in the future.
Alex ZhavoronkovSo this lab is more of a target discovery plus target validation lab. Now we are uh miniaturizing it, we're trying to make it smaller. So you can even do this at the hospital level. Imagine if you could do personalized drug discovery for individual patients or perform some uh already um established uh personalized medicine services or um decision support services for physicians and at the same time discover targets.
Alex ZhavoronkovSo you could place those labs in regions that never even aspire to be a player in biotechnology.
Christian SoschnerUm how small, how small can you make it?
Alex ZhavoronkovUm I think I can do it in two rooms.
Christian SoschnerIn two rooms, this is practically you can can do it everywhere then it's not yeah.
Alex ZhavoronkovCurrently it's the entire floor which is automated. Um and it's a pretty complex system, so it's many rooms uh performing different tasks uh with many different types of equipment. I think if I try to place this equipment in 3D and also um uh make some of my own equipment, uh I would be able to automate this process with just tools.
Christian SoschnerSo it's possible to run the processes in every hospital of the world, basically with uh every major one that can afford something like that, yeah.
Alex ZhavoronkovOr if there is a government program that allows you to uh uh to do something like that. So uh currently that's a dream, right? But and I and I don't I cannot guarantee you that we will have a hospital-based robot. But you know, three years ago I did not even dream about the lab that we have.
Christian SoschnerYeah, this would be my final part uh for this conversation. So looking into the future. Uh, but before I ask that, uh we're coming to the end of the podcast. Uh, is there any question open that you would like me to ask? Is there anything that I missed that you think is important to share?
Alex ZhavoronkovWell, I think that you covered pretty much everything. I must say that uh the most important disease that we have uh in the world is actually aging because we will all lose function and die. And in this process, we will also get a bunch of diseases and trauma, and uh uh it's never a happy ending. So it's always loss of function. So I think that we need to look at um and prioritize therapeutics that have dual purpose that can treat a disease but at the same time target um important uh biological processes that are implicated in aging.
Alex ZhavoronkovAnd what we do uh at Ancilico and what I do uh in general um is looking for those dual purpose therapeutics. So I think being dual purpose aging and disease, that's truly key for identifying blockbuster drugs, because uh the probability that your uh drug is gonna work for many people at the same time is gonna be higher if you are addressing a very broad biological process. Um and second, we need to figure out aging, right?
Alex ZhavoronkovBecause that's that's the main killer. So uh this year, 60 million people are gonna die of the due to aging. It's more than any war or any cataclysmic event or any plague. Um, and we need to figure out how to uh increase health span and significantly delay death.
Christian SoschnerDo you do you think it's possible uh to delay death? I mean, there's this uh there was a science fiction book that I read, I think, 40 years ago, um, where the assumption was that one day uh some pharma industry in an alternative reality will develop a drug that basically makes death a conscious decision. So people don't age, people don't get sick anymore. Do you believe it's possible to come to that point in our reality somewhere in the future?
Alex ZhavoronkovSo with pharmacological means, I'm talking about small molecule drugs or biologics uh or uh even RNA therapeutics, you won't be able to dramatically increase lifespan. So I'm talking about maybe you know, plus minus plus uh 40 50 percent is probably possible if you combine a bunch of uh therapeutics that A intervene with your metabolism and B um uh improve uh certain repair functions, uh and maybe even um uh stimulate certain uh certain biological processes that are just too expensive for the body to run all the time or where we do not have a biological program.
Alex ZhavoronkovUm so I think that significant life uh life life extension is possible using pharmacological memes but not dramatic to go dramatic you need to look past pharmacology it's just pharmacology is the most uh credible way to uh extend lifespan and um it's also something that is very sustainable so if you are very good at discovering and delivering novel therapeutics for diseases uh and you're also prioritizing aging as your primary uh outcome measure so the effects on uh uh on on on on the health span and lifespan of your patients um I think that this is the area where you can achieve commercial success uh and then fund other areas uh of longevity research that may give you uh more longevity dividend compared to just for pharmaceuticals yeah I think there's a lot in lifestyle um pharmaceutical is one way lifestyle is another way and probably with uh Nona Linko but with Elon Musk research or research that's going in that direction maybe one day it's possible to upload consciousness uh into into the cloud and download it to an uh artificial intelligence or to an uh to a robot yeah I wouldn't bet on that uh but uh um it's definitely a very promising area of research for other purposes also aging and uh Alzheimer's CNS diseases uh but yeah I could look at uh maybe gradual replacement of neuronal tissue uh as the uh you know way to rejuvenate the brain but uploading that's something that uh we may not be able to master i mean creating a copy yes but is it gonna be you most likely not why not create a clone then right yeah yeah that's a big question I have one final question left uh when we are talking about the future how do you envision generative and robotics impacting the future of healthcare and medicine well I think it's already impacted in a massive way uh so and partly uh due to our work in generative chemistry and generative biology in uh uh you know 2016 2017 uh 2018 so seminal work uh and now it has propagated uh so I think that generative AI is going to transform our life uh beyond recognition in every way right so we are actually moving into the new uh era of knowledge management so instead of retrieving knowledge uh through search and through by browsing the different directories we can now uh generate uh knowledge and uh complete response to your queries with reasonable accuracy this accuracy is expected to improve over time uh and you would be able to get uh um purpose built systems for discovery so just like our Panda omics platform but even more uh reliant on generative uh AI with much better user experience uh and uh uh and perhaps conversational AI.
Alex ZhavoronkovUm and yeah I think that we are getting to the area where generative AI will play a role in pretty much every area of drug discovery and development. Most important uh uh area where I see the application of generative AI is the ability to generate high quality synthetic patient data. Right? So currently people are just too sensitive about their healthcare data being misused for all kinds of purposes. By the way I I do not know many cases where people were hurt by having their health data accessed they're hurt by having their financial uh records for example accessed or you know credit card information identity information uh but healthcare information very rarely uh there are a lot of hypotheses but uh insurance companies are um uh regulated right and uh prohibited from using genomic information for discrimination so for drug discovery I think we should be able to use this information um however since people are so afraid of sharing their biological data we can now use ai to generate high quality data instead of taking them for taking it from specific individual just by like by by asking ai to generate specific biological data sets with desired properties uh trained on publicly available data that's that's a great future that's a great future alex thank you very much for your time uh and uh on the world uh government summit and uh I wish you and your team all the best for the future uh you're doing an amazing work and I believe you your team will move the healthcare industry globally forward and enjoy the summit.
Alex ZhavoronkovWell thank you Christian I again apologize uh to you and uh your um listeners uh and people who watch this uh for the background noise and also for the kind of presentation from the field um I just couldn't change it and I couldn't uh escape so uh thank you for bearing with me it's a nice it gives a nice flavor to to the episode I like it I like it thank you very much Alex and have a great day thank you bye bye you bye bye thanks for joining me on this journey to discover the groundbreaking potential of artificial intelligence in drug discovery and anti-aging research if this episode has blown your mind and opened your eyes to the incredible power of generative AI then smash that like button and hit subscribe to never miss out on future episodes.
Christian SoschnerTogether we can spread the word and inspire a revolution in the pharmaceutical industry that will change the world as we know it. So don't be shy leave a comment and share this video with everyone you know to help us reach more people and make a real difference. Thanks for being a part of the movement towards a brighter healthier and longer future enjoy your day
Invest in the wrong startup and you lose one times your money. Miss the next Google and you lose ten thousand times. Most boards are built to avoid the first mistake. Venture capitalists play by different rules. They...
The trait a Silicon Valley AI investor rewards most in a management team has a name almost nobody uses: XQ. IQ brings the idea and EQ reads the people across the table. The execution quotient shows in something as small...
In 2017 another investor told Ivy Nguyen that artificial intelligence was too small a niche to dedicate a fund to. More than ten thousand AI pitches later, every pitch deck says AI. Single funding rounds have grown...
A seed investor at Novo Holdings would rather read your rough idea than your finished pitch deck. Most founders spend a year building the deck, forming the team and waiting until the story holds together. By the time it...
What does it mean when the most respected adviser in Silicon Valley tells you that you are the best chief executive he works with, while your company is heading straight into a wall? Bill Campbell said exactly that to...