Aired:
June 13, 2024
Category:
Podcast

A Venture Studio with Big Pharma Backing Seeks to Fix Drug Development with AI

In This Episode

Mati Gill, CEO of AION Labs, sits down with Amar Drawid to discuss how the AI venture studio works with its Big Pharma partners.

Episode highlights
  • The biggest drug development hurdles the industry faces.
  • Challenges to attract academics, entrepreneurs, and life sciences professionals with the best solutions to build new companies.

Transcript

Transcript of 12 LSDNA Mati Gill

The Life Sciences DNA podcast is sponsored by Agilisium Labs, a collaborative space where Agilisium works with its clients ranging from early stage biotechs to pharmaceutical giants to co -develop and incubate POCs, products and solutions that improve patient outcomes and accelerate the development of therapies to the market. To learn how Agilisium Labs can use the power of its generative AI for life sciences analytics,

To help you turn your visionary ideas into realities, visit them at labs .agilicium .com.

You're tuned to Life Sciences DNA with Dr. Amar Drawid.

Amar, we've got Mati Gill of Aion Labs on the show today. Who is Mati? Mati is the CEO of Aion Labs. Prior to Aion, he spent 11 years at Teva Pharmaceuticals as the head of government affairs, corporate and international markets, where he developed Teva's innovative strategy and built a network of partnerships. He's got an MBA as well as a BA in law from Reichman University. And what is Aion Labs?

Aion bills itself as a venture studio and has developed an unusual alliance with top pharma companies including AstraZeneca, Merck KGaA, Pfizer and Teva, as well as Amazon Web Services. It's working to use AI to solve major problems in drug discovery and development. And before we get started, I just want to encourage our audience to hit the subscribe button if they'd like to keep up on the latest episodes or.

hit the like button if they enjoy the content and give us your feedback. With that, let's welcome Mati to the show.

Mati, thanks for joining us today. We're going to talk about the challenges drug developers face and how Aion Labs is seeking to address those challenges using AI. So most people know that drug development is expensive, requires enormous amounts of time to go from the lab to the patient. So what are the most significant challenges that AI can be used to address? So thank you, Amar, and thank you for the opportunity to have this discussion,

really a thrill and exciting for me. So I appreciate you inviting me. I think our partners and in general, the global pharmaceutical ecosystem is looking for new ways to address exactly what you mentioned, Amar, is how do we bring new hope to new patients, drugs that are previously, for targets that previously we haven't been able to drug.

And to do so by failing less and make healthcare much more accessible by being able to do so in a record time for much less money and for failing less across that process where we all know that only 10 % of drugs that go into clinical trials actually make it to approval of an FDA. So the thinking is that by using new technologies and by taking a data -driven approach, we can actually make better decisions,

what drugs we want to develop and be able to do so in a much more efficient and fast pace to be able to actually bring hope to patients. Throughout my career in the pharma industry, I've seen a lot of technologies that have promised to accelerate drug discovery and development and reduce costs. By and large, I would say that there's been some success, but not a whole lot of success in terms of the acceleration.

And still, the cost is, I guess, almost getting to $3 billion of drug at this point, 10 to 15 years of development. So, what's the case for AI being fundamentally different than before? Well, I think, Amar, you just made the case yourself by giving the numbers and the statistics there on how long it takes and how expensive it is. And really, the reason why it's so expensive is because in the $2 to - to $3 billion per average for

drug that's proved that makes it through, we're paying for all the failures that don't go well because we actually want the pharmaceutical industry to go and innovate and to be incentivized to continue to innovate and to bring new drugs to the market. That's something we as a society want to continue to incentivize doing that. We want to incentivize the private market and industry to do that. Now in order to do that, you need to give them a good return on their investment.

So if already the two to three billion dollars is what it takes on average to bring a new drug to the market, we're understanding it doesn't actually take those costs for each direct drug directly. It's because we're paying for the industry to innovate. And part of innovation is also failing. And right now, as I mentioned before, you have a higher failure rate than what is conceivable if we are able to really drive insights from artificial that

can be driven from big data analysis, artificial intelligence, machine learning platforms that can help us to just fail less and help us choose the right targets, choose the right drugs for the right targets, the right mechanisms, be able to develop them in a much more precise manner in the preclinical as well as in the clinical stages, and then we can actually fail less. So there's a huge opportunity there to be able to lower those costs of bringing new offers to patients, and I think that's one of the opportunities.

opportunities that our industry sees with the potential of AI and it's a new market. It's a new technology. Ultimately, we're trying to develop a drug. pharma knows how to develop drugs and this is a new technology that is at the toolbox of the pharmaceutical industry to be able to use in order to ultimately develop a drug. and You know from our perspective if it's developed in a wet lab or if it's developed using computational tools

it shouldn't ultimately matter. The FDA still needs to approve a drug. It doesn't really matter how it got to that point, right? As long as they abide by the protocols to get there to that end point. And so we're just trying to add with the AI industry, we're trying for AI for pharma industry, we're trying to add new tools into the toolbox of an industry that very much needs them. And do you see any specific area in

the phase in the drug development process where you think AI can work more than the others or you think there's applicability across the board? So I think there's in the drug discovery and development area, there's applicability and there's an opportunity for AI to really impact the whole spectrum from the beginning to the end, ultimately to the FDA approval and then beyond even in post -market surveillance.

There's an opportunity throughout that whole spectrum. The way we try to look at it in our venture studio of Aion Labs, we try to segment that process into four different buckets according to that value chain. So we create companies here that address big challenges that go across that whole spectrum, either in the way that we're able to understand diseases and to bring new insights into disease mechanisms so we can actually

discover new targets and new ways and methodologies to attack diseases by understanding them more in depth. And then secondly, drug discovery. And I think that's the low hanging fruit of our industry is how do we actually use AI to be able to discover new drugs and previously undruggable targets and be able to discover new drugs in a much more fast pace. And then to be able to make preclinical studies our third segment, to be able to make the preclinical process much more precise,

because when you actually solve cancer like we have many times in a mouse, that doesn't mean you're actually solving it for human beings. As we know, we haven't done that yet, but our industry jokes that we always say we've solved cancer and cured cancer in animals, mice, whatever, many different times with a lot of different problems that ultimately doesn't reach the end goal of being able to cure it in patients as we all know. So preclinical studies is an area that there's a lot of potential for AI and

computational technologies to be able to impact. And then fourth, as we discussed before, it's the clinical processes. It's all the phase one, phase two, phase three and beyond of clinical trials that are the most costly part of drug development. And then our failure rate is 90%. So if we're able to use big data and analysis, machine learning platforms, artificial intelligence capabilities,

to be able to lower that attrition rate from 90 % to something much less. Even if we lowered from 90 to 60 % failure or 50 % failure, that's a huge amount of value that we're able to capture by bringing these new type of tools. And we're seeing companies coming up, both the ones that Aion Labs create, but of course all across the industry, we're just two and a half years old where we've created nine companies up until now. But we're seeing...

we're seeing companies come up all across the region, both in Israel, as well as in the United States and elsewhere across that whole spectrum. Absolutely. I agree with you. We probably have created the best drugs for mice already, working on a lot of mouse models. But I really like the way you approach it. And for the audience who don't know the drug discovery process and the development. So you usually identify the disease mechanism and what are the genes

that you want to focus on, right? Those are the targets and there's applicability in terms of making that a faster process. Then once you know the drugs or the targets, then you want to find either a small molecule or you want to find like a biologic drug that will be really attacking that target. And so there is a lot of opportunity in that area. And then as you move into development, there's the clinical development where...

of India, a lot of the animal models, we want to make sure that there is the efficacy and then the toxicity is low, right? And we're basically enhancing the drug there and then finally going into the clinical trials. So in all those areas, there is definitely a lot of opportunity to, as I said, accelerate, also reduce the failure rate and really,

make the drugs cheaper to develop. So across the board, that's great. If you think about it, Amar, if you think about it, just to give the one example about the animal models that we try out in the preclinical phases, the reason why we continue to do animal models before we test out a potential drug in humans is because we want to check out in a very humane manner and according to all regulations, but we want to test out whether a drug is going to be safe and efficient in attacking the drug that we would

then attacking the target that we actually want the drug to be able to address. And the reason why we do it in animals is because we haven't thought of a better way to be able to do it before we bring it into humans. So what if, and artificial intelligence is the art of saying what if, so what if we could actually, instead of testing it out in animals, we could test it out in a virtual setting and test it out in virtual reality or however you'd want to call it

based upon simulations of patients that are based on previous data. And that's something that the industry is trying to do, but we don't have that model yet developed. We don't have that technology yet proven. And therefore, regulators have to take and will take it until proven otherwise, a very conservative approach because we have to be conservative about the way we treat disease in our own bodies.

So until we're able to prove that we can do it differently, we'll continue to do so in animals. And it's a very inefficient way of doing things, but we just don't have a better way. And that's what technology can help solve. Absolutely. And the prediction there, and especially as you look at the different therapeutic areas, at least oncology has some decent animal models, some disease models in animals, so you can at least try out. But something like neurology is so difficult to have the right

animal models in neurology that it is very hard. And that's one of the reasons why so many drugs fail in neurology clinical trials is that there are no good animal models. So you're just trying these drugs out in humans and most of the time they fail. So absolutely, I mean, there's a pretty good - And And don't know know that much about the human brain yet and the way it works. There's more unknown than known about the mystery that is the human brain

and how it works and then how diseases of the human brain actually work. So that's where in the understanding of disease mechanism, we actually have a lot of potential that we can do and a lot of untapped potential that we can gain insights into and hopefully we'll be able to use these technologies to discover insights that will help us treat diseases. Absolutely. So tell us more about how did Aion come about. So Aion Labs is a venture studio.

where we create new startups in either a building program or a seeding program. The main difference there is either starting them with a big problem statement that comes from R &D experts from one of our four pharma companies and partners in Aion Labs. Or in our seeding program, we scout for technologies and teams that address big problems that we then need to validate, actually address challenges that our pharma partners are willing to get behind.

But ultimately we create companies where they have three core components. One is a big problem that's an unmet need of the pharma industry. Two, that it's got a top notch team with technological approach that's truly innovative and they can actually lead a new startup. And then third is providing them with the best possible environment for them to be able to co -develop with our pharma partners and...

be able to bring a technology to a place of validation where it can continue to be a track for investment and then continue to grow and graduate from Aion Labs and grow into a growth company. Aion Labs has been around for two and a half years and we were accepted as part of an initiative to really drive technology and to drug discovery and development.

Our four pharma partners decided to come together and actually create startups together here in Israel, but with global talent coming from anywhere in the world. So we're talking about Pfizer and AstraZeneca together with Merck KGaA, AMD Serono, as you call them in the U .S. and Teva Pharmaceuticals is the pride of the Israeli industry here. The four pharma companies came together with Amazon Web Services, with AWS as our cloud and computational partner.

partner and two venture capital firms with the support of the Israeli government and adapted a model for company creation that we brought from BioMed X, a research institute based out of Germany and that now is a campus at Yale. And basically we brought all these components together of talent, drug discovery and development expertise and data from the pharmaceutical industry and the capability to actually

mentor teams from day one and co -develop technologies with them to find the challenges and co -develop the technologies with them, with experts from across the industry, great computational partners and then good investors with the support of government funding to be able to help remove as many hurdles possible from these startup entrepreneurs to be able to bring their technology to life and co -develop a technology, prove that it actually works and then build a great

successful startup and that's what we're doing. We have a very strong mission of trying to accelerate the process of drug discovery and development and bring new drugs to the market and to be able to help build an ecosystem globally as well as here in Israel for tech bio. So now you described the two models right so one model is more about. our Aion building yeah okay building and I Aion seeding. I'm building in our city and

So how does this model differ from, let's say, an accelerator or an incubator? How does that overlap with that? So we're very different from an incubator and accelerator in that what we do is we actually work with our partners. We actually are trying to co -develop with the industry to help disrupt itself. So we work with our pharma partners and we say, give us a big challenge,

a big problem statement and then we can dive deep with them. We don't make it easy on them. We dive deep with them, make sure it addresses criteria that we know that it will be the basis of a good startup. We know how we'll measure it. We know how the ideal outcome will be. And we know what the fundable milestones will be if it's successful. And we dive deep with them and select a big challenge that we make sure is an industry -wide challenge and not just one company wants to get behind, but at least two of them, hopefully more. And then what we do is we actually, instead of,

the process that an accelerator does where it sees deal flow and then helps them or an incubator where it scouts only for technologies and then invests in them, we actually go and launch a challenge very publicly. We say we are open for great talents that can come from the tech side, they can come from the bio side. Ideally, they're a team that has multidisciplinary capabilities that come from both.

And we say, come and pitch us with the idea on how you're going to solve that challenge. And our experts then select the top 15 candidates. And right now, we're in the week -long challenge workshop of one of our challenges like that for how do we target RNA using small molecules. And basically, what we do is we, with a machine learning and AI platform, what we do is we bring in the top 15 candidates that are divided up into five teams.

And we work with them throughout the week with the hands -on expertise of our subject matter experts from the Pharma R &D groups. We work with these teams through the week to help co -develop their concepts up to a place where they can pitch it to our investment committee. And at the end of this week, we will select the winners that will establish our ninth startup at Aion

that will receive a pre -seed investment from us of $1 million at least. It could receive even more if we feel that it's justifiable. But we'll give them at least $1 million to establish a new startup for the next two years to be able to reach a fundable milestone that it can then grow and be attractive for continuous funding. In our third pillar, what we do after we have a great problem statement and have selected a great team, we then...

provide here in Aion Labs the ideal framework for them to be able to reach that validation point and fundable milestone. Okay. And so, the challenges that you're coming up with, are you posting them publicly? Yeah, absolutely. Okay. Absolutely. So you can go into our website. We publish one at a time. We don't publish them all at the same time. We publish one. We receive applications. We go through the whole process.

We've published seven of those up until now. And this is our seventh one out of our cycle. Very excited. We have actually 16 candidates this time in five teams. Excellent candidates, top scientists coming from all the way from Israel, all the way to the West Coast of the United States. They're all here working with us this week in a very intense, packed agenda. And it'll be, you know, ideally our success of such a workshop

is that our investment committee will have a very tough decision on who they're going to choose will be the winners of this startup and the founders of the next startup. Yeah, and so the people who try to address these challenges, are they coming necessarily from the biotech world or from the academia or are they already some companies or are these individuals who are doing something in labs who research? LA? How does that work?

Or like people from Silicon Valley, like how does this work? So our philosophy that we work in accordance with is that talent is dispersed equally throughout the globe, but opportunity is not. So we try to reach out to anyone and to offer them this opportunity to come build a startup. And we're very agnostic. We're not, we're very respectful of someone having a good scientific background.

Ideally, if you have a PhD, you've probably worked hard for it. If you've worked in the academic background or in science, you've probably been able to achieve quite a few things in order to reach that point with a PhD. But we're not snobs. If you're brilliant and you're a rough diamond that we can find you and you don't have your PhD yet, we're not going to turn you away just because you don't have a PhD. So ultimately, to answer your question, Amar, we don't care where they come from. OK. But we do

take a very proactive approach in finding the people that we want to invite to address our challenge. So we have built a talent platform here where we can proactively identify potential candidates based upon their scientific publications, their background, their previous places of employment. We reach out to them very openly and we say, if you ever dreamt about building a startup in the biotech space,

here's a built -in product market fit with a good problem statement in a platform where you don't necessarily need to be a proven entrepreneur with multiple years of experience under your belt and second or third time founder. We'll help you provide, will help provide you with all the background and assistance for you to succeed that you need. Just focus on your science. Come bring your brilliance. Come bring your capabilities and your.

or your brilliant ideas in how to build a new startup, you can focus on the science and we'll help take care of all the rest. So we reach out to as many people that are relevant as possible. And like I said, we have founders that come from a deep biological background, scientific backgrounds, academia that want to move into industry. We have people that have come with deep tech backgrounds that have never worked a day in the bio world. And that's exactly what we want to bring in because if we ask,

if we ask life sciences PhDs to work on AI, that's probably not going to work. But if you ask AI people to do deep biology, that's also not going to work. So you need to be able to bridge between these two scientific disciplines that are very different from each other. Absolutely. And that's what we're trying to do. And so you have this approach then people can come approach your

problems as well as you have the talent pool about who could be the right talent to solve the problem. So both of these approaches that you're using. Yes, we receive, we make sure that our challenges are published quite broadly so that people that have an idea on how to solve solve it, we're very receptive of that, could come and approach us. And then we also make sure that we reach out proactively to individuals and teams that we think could potentially address those challenges.

And ultimately, we're very happy with the fact that we have 15 excellent candidates here this week with us in order to choose one person or one team that we're going to build a startup with. And how do you then, it's kind of like matchmaking where you're getting some really bright people from AI, some from life sciences. Are you the one who then puts the team together and now you're the one that then says, okay, well,

you're the right team. You should be forming a company. We'll be forming the company with you. How does that work? Yeah, so we don't try to play God in that sense or matchmaker, or however you want to call it. What we say is we're offering them the opportunity to build their own teams. So they can either come as a team. We've seen that. Or we can offer them opportunities to match up as a team before applying

and offer them, but they choose who they want to match with. We just give them the opportunities. And then third option is if they haven't chosen a team member to join up with before the workshop. And then on day one of the workshop, we call it the team building day. Everyone gets to know each other. Everyone presents about themselves. And then they decide who they want to work with by the end of the day. And then on day two in their central hypothesis day, they receive their science and start to work as a team.

Okay. And then how do they choose their team members? They choose that. And how do you then like form this company? Because then you are really driving formation of that company. Right. We see ourselves as co -founders of the company as an outfit. So obviously the scientific founders are the ones that are putting their sweat equity into the startup.

But we're coming with a problem statement and we're coming with the investment and all the framework that will help them succeed and co -develop the technologies with them. So we really see ourselves as true partners with them. And once we select the founders, we give them a term sheet, we reach an agreement, we sign it, we receive the funding from the Israel Innovation Authority that's very supportive of our efforts and very integral to our efforts. And they're...

They're an outstanding example of how a government can actually incentivize innovation in that sense and be very effective in doing that. And then we set them up as a new startup and we offer them all of our access to all of our technology, all of our infrastructure. We have a whole team that works with the startup that helps to make sure that we have a very clear pathway for them to reach a fundable milestone and unparalleled access to the experts

in the pharma industry in order to do so. Just to give you one example, we have a startup here in antibody optimization, building a machine learning platform, an end -to -end machine learning platform, machine learning based platform, to be able to optimize an antibody. So basically, if we have a potential drug as an antibody, to make sure it's efficient and safe for usage, and to be able to run it through this machine learning platform to identify how we need to optimize in order to make it a better

drug. And that's a team that's made up of an excellent structural biologist together with a machine learning expert. And then two excellent team members that they've added to them. And they meet on a monthly basis with the heads of antibody engineering from AstraZeneca, from Merck, from Teva, with input from AWS and our venture capital firms. And that's unparalleled access to a pre -seed startup that's only been around for six months. That doesn't exist anywhere in the world.

And our partners thankfully see the value in this to be able to help partner with these types of startups like CombinAble.AI AI that I mentioned right now and others to help them develop these technologies because they've seen the other option. I mean, what's the alternative for the way this work is you come as a startup and you say to a pharma R &D expert, just give me all your data. I'll bring you the moon and the stars and I can promise you anything in between.

And our pharma partners are a little skeptical to that approach because so far they've been underwhelmed by the results of those promises. So they really want to get under the hood and understand what's behind these technologies and help bring these technologies in an effective manner into the drug discovery and development process that they are experts at and want to be able to do that. So tell me how your four...

how do pharma partners work with each other because something new, exciting is getting discovered, they all want their hands on that. So how do you work with the four partners? Yes, that's an excellent question. And I would say there's three things that go beyond that. Number one is we have very strict antitrust guidelines. That was the first decision our company took because I'm a lawyer personally and we have a very strict

respect for the rules of engagement in that sense, to make sure that we cooperate on things that are for the good of humanity and the good of scientific breakthrough, and do not cooperate on things that we should not, by law, be cooperating on. And we very strictly adhere to that and remind ourselves of that on a regular basis. So antitrust guidelines is number one. Number two is, as a result of such,

I Aion Labs build startups that are in the pre competitive space. So we build technologies that can serve the whole industry. Okay. And then number three is the startup is the owner of its IP outright. So the investment that we give to a startup is an equity investment that Pfizer, AstraZeneca and so on are providing without any IP rights. So that'll take any first right of whatever,

in any sense, as a result of their equity investment. And if they want to reach a business collaboration with a startup to acquire rights to a potential breakthrough innovation, to a potential pipeline asset, then they have to negotiate for that on commercial basis, just like any startup would in accordance with any company. And it could be Novartis that's not in our partnership, or it could be Roche or any of the other competitors out there.

in that sense. So we make sure that we're building startups that have no strings attached to them and that can compete freely in the open market and our partners receive equity in exchange for their investments but no IP rights. Okay, so what would be the incentive for these four companies to spend their energy in developing these? It's a great question. So they want to be the first to know Okay.

of a great technology coming in, be able to test it as well and have gained confidence in it. It's kind of like a sandbox that they're collectively deciding to play in and test out technologies in and have access to and the knowledge of what actually works in that sense. And imagine there's all these technologies out there and here you have an effective risk sharing platform and how to test them out together and really help develop these technologies together in a manner that will be able to serve their purposes.

So just having the first... And address their unmet needs. And address their unmet needs. I mean, we're coming all the time, think about it, we're coming with a real pain point for the industry where we said this is an issue that we want to solve because it can help make our work more effective. And it could help us bring breakthroughs and then develop drugs that will go into our pipeline, etc. etc. Help our pipeline assets make it through the process

successfully. So these are coming and these technologies will address pain points that have come from within the industry and that's a huge value. So it's a pain point that they have and now they've seen firsthand a potential solution for those and they're getting the first knowledge of that and that's the advantage that they're getting. Yes and they have some financial upside into this as an equity investment as well. Yes absolutely. Okay. But that's the upside. That's not the core driver. Okay. Okay. And what about AWS?

So AWS, it's a great question. AWS, we ran a lengthy process to choose the right cloud and computational partner. And we're very, I mean, AWS have been tremendous partners. They have deep computational biology knowledge within AWS to be able to support our startups. Ultimately, what I believe that they want, and you'd have to ask them to get a definitive answer, but I think they really want the insights for how the pharmaceutical industry works,

what the pharmaceutical industry is interested in developing and imagine that they as AWS get unparalleled access to be able to sit around the table, not just with the CIOs that they have regular access with or with the CEOs that they get to have an up with, but actually with the VP R &Ds and subject matter experts from within the R &D groups, the directors, the associate directors, to actually sit down with them and really understand in depth

the drug discovery and development process and where cloud computational technologies can play a role to accelerate their processes and build those relationships and understand their needs. And then all these startups will understand the potential of building their technologies on the cloud from day one. And that's, I believe, a huge incentive for a cloud computational partner to be able to help build these technologies the right way on the cloud to accelerate their technologies from day one. Okay. Okay.

Now you have, as you said, you're on your ninth question, right, at this point. On our ninth startup, from the two tracks, from both our building track and our seeding track we have where this will be the ninth startup that we've established that will come out of this week. OK. So do you want to talk about maybe a couple of the startup companies? Yeah, sure. I'd be happy to. So we have five startups that we've actually unveiled

and announced up until now. I'll just give you one example, which is DenovAI Biotech, which is a startup that addresses a challenge that we call DenovAI Design of Therapeutic Antibodies. So basically, how do we take that process of discovering a new antibody on a computer and do it through computational technology? So that was the first challenge we actually issued globally. And it came up as an unmet need

from all four pharma partners in a bottom -up exercise where we asked the R &D partners, what's your biggest pain points? And that issue came up from all four of them, because all four of them want to be able to do this computationally, both for the existing targets and drugs that they know, as well as for undruggable targets that they believe biologics have the potential to be able to address.

And we launched this challenge globally. There's a lot of startups that are in this space, but our pharma partners and AWS, we're all in our venture capital partners. We're all very convinced that there's no technology accessible to them out there that currently addresses that change, that challenge. And that still remains to be true. And so we went and sought out potential solutions for this challenge. And the winner had never been to Israel before, had never thought about building a startup in Israel before.

He's a startup CEO called Dr. Kashif Sadiq, who was a senior scientist at the European Molecular Biology Labs in EMBL in Germany, in Heidelberg, Germany. And he came, won the challenge with the support of his co -founder and hopefully future CTO from the EMBL Labs and PhD candidate from EMBL.

And they won this challenge and Kashif is now the CEO of DenovAI Biotech. And he's been able to build an excellent team here in Israel with an additional 14 members, a head of bioinformatics, a COO who has a vast experience in drug discovery, as well as two computational biologists with deep structural biology expertise. They're building modules to be able to address both the design of proteins as well as,

as well as antibodies de novo. Basically, they're starting now to work on partnerships with our pharma partners to be able to validate the technologies that it actually works. And within the year, we hope to be ready to raise significant funds to help the startup grow. They've already been able to reach a seed funding beyond the $1 million that we provided. We provided them with $1 million, they've been able to reach additional funding and seed funding. And we're hoping to be ready for...

much larger realm based upon validated technology by the beginning of next year. That's a pretty nice story. Any other startup that you want to highlight? Sure, so I can actually talk about the opposite view, where part of doing a venture studio right is also understanding what doesn't work and being able to test out technologies,

but taking quick decisions. And if you read the case studies from Flagship Pioneering, or any other venture studio, you know that one of the keys to success is taking tough decisions early on and taking quick decisions and decisive decisions to shut down projects that aren't working. So we established a startup called Omic .ai, which was a great team, still is a great team, of two entrepreneurs

that didn't have previous experience in the life sciences space. They had vast AI experiences in the auto tech space as well as in the agrotech space and generative AI before we actually called it that. And they were able to come with a technological approach to how to do preclinical assessment of clinical readiness of a drug candidate before it goes into clinical studies.

And we worked with them for a year and in a very mutual decision understood that there wasn't a product there. And the problem statement wasn't clearly enough defined. And the expertise that they brought was not sufficient. We needed to have deep biological expertise as well in the approach. We just weren't able to prove it within a year. We weren't able to prove that we were able to, that we were going to be able to prove it within the next year. So in a very mature decision, the team,

and we decided to shut down the project. And we're now working with that team in support of one of our other startups to help them develop their technology. They've gone on to do amazing things, they're good friends. And they've gone on to help one of our startups actually on a project base, develop breakthrough technologies for a collaboration that one of our startups is doing for one of our pharma partners. But we learned and you learn from your failures and part of the value I hope that our pharma partners are getting out of here

is that they're able to test out technologies that are high risk and some of them won't work, but then they'll have the confidence that a specific technological approach actually doesn't work. So when they're pitched by the next startup promising that they can do that, they'll have the experience understanding hands on that, man, this one probably didn't work. So as you choose these questions, right, there is the value part. And for a lot of those, the value part, yeah, I mean, there can be huge value,

but there's also the feasibility aspect, right? So those are two axes I can see. How much can you assess about the feasibility and what are some of the other aspects that you also take into consideration when choosing a question? Yeah, so we have a very strict set of criteria that we call a deep dive that process basically into what makes a good problem state

before we launch it. And as a result of our lessons learned from the process with Omic AI, we're very strict with our pharma partners. We don't cut corners in that process. We make sure that all of our process and all of our criteria are met. So that's first and foremost, what is the exact problem statement we're trying to solve? What will success look like? What will the ideal solution look like? How will we measure it?

If we say we'll measure it only in clinical trials and we really haven't done anything because we haven't saved any process. So how do we, how do we have, how do we have a measurable milestone that's not a clinical trial that our pharma partners will actually trust and then be willing to test it out and validate it on and be able to use this as a commercial product in that sense. And then we have to get the VC input of, you know, is the

competitive landscape open enough for us to actually build a technology that will be the basis of a good startup. And so on and so forth. And we put all these criteria together to make sure that ultimately when we sign off on a problem statement that we're going to publish, that this is something that will be the basis of a great startup that's reflecting true market insight and a product market fit that's built in to be the basis of a next great startup. And that's something that we're

very strict on. And when we publish a startup, it's a long... When we propose a challenge, it's a lengthy process that goes into actually publishing that challenge. It takes months, if not close to a year to develop a problem statement and approve it and make sure that we have the endorsement of at least two of our partners, sometimes up to six of them. But ultimately, when we publish a challenge, it's gone through all that process so that entrepreneurs can feel very confident that if they solve it,

and build a startup to address it, that it will receive the support of our partners. And if they're able to solve it, it'll be attractive for a lot of funding to be able to go do some good in the world. So now for these kind of startups, right? Like for, I'm looking at like what could be the causes of failure. And so of course, I mean, defining the problem statement well, or having the right team, or just the technology, I mean,

is just not good enough. Or I wouldn't say good enough, but the technology is not able to solve the problem. So, when you're looking at, as you have been involved in this area for so many years, what are some of the biggest reasons you see these companies failing and are there things that companies can do to succeed?

So you know Amar, it's not a secret that most startups actually don't succeed ultimately. And then you can ask a question of what success look like. So ultimately what we feel is the core components are have the right people that are working on the right problem statement and with sufficient enough funding to be able to actually develop their technology in the right way with the right plan. So if you choose the wrong people,

or if you choose the right people, but then they choose the wrong people. That's a key element of failure. If you're not able to take decisions and you don't have the right leadership skills in order to lead a company the right way, that's a potential fail. If you don't have access to data, since we are in the AI space, that's a potential failure point. And if you don't have access to funding at a sufficient rate, or if you take up, if you take too much funding at too high of a...

price too early, then you reach down rounds and then your investors might get a little bit anxious about the potential of the startup going forward. Those are all elements that can contribute to a failure. And sometimes, and this is something that you see in the tech for bio world, is that no matter how great your engineering plan might work, biology also works different.

So sometimes you might have a great plan for a great drug that you've researched well enough and ultimately the science just doesn't work. Or it does work but it's not differentiated enough from the standard of care. And these are all things that can contribute to failures and we fail more often in startups than we are successful. And that's true especially in the biotech world. So what we're trying to do at Aion Labs is take as much of those elements that we actually know

that contribute to failures. There's a lot of those that we don't know. But take as many as we can that we can anticipate and that we know are contributing to failures. And try to remove as many hurdles as possible so that our startups can be as successful as possible by removing as many of the hurdles that we know of so that they'll make their own mistakes. But let's try to remove the ones that we're able to remove by anticipating what contributes to their failures and providing them

with solutions from that. Now, in the four areas that we talked about, where do you see, I mean, I'm just like trying to see if you have a crystal ball, right? And what are some of the areas where you see AI making a lot of difference in the near future or even in, let's say, the medium term?

So I think where we're seeing a lot of movement on is clinical trial space because there's an understanding there that that 90 % attrition rates and failure rates is just something that's unsustainable and there's no real reason behind that. We should be able to design trials better. We should be able to help pharma companies reach ultimately the end goal of FDA approval with a good drug at a much higher rate than a 10 % success rate

in clinical trials ultimately. So we're seeing a lot of technologies out there. It's a very saturated space. What's needed for these companies to succeed is to have sufficient funding, access to data, and a really good team with a very smart approach. And even here in Israel, we have companies like Phase B that's gone out of stealth and raised a massive seed round and been able to actually publish some exciting results. Working with pharma partners to be able to help with adaptive

clinical trial technologies as well as address the challenge of clinical trial outcomes is something we're very excited about. We're seeing many startups out there in that space and I think that's something that we should be able to do much better and that AI really has the capability to influence that impact. And in the last year and a half, you've seen this huge surge of generative AI. How do you see generative AI now coming into the biotech startup world?

I think just like anywhere else here, it's very expensive to develop a technology like that. And everyone says, let's build a chat GPT to be able to, if you just type in, listen, I have this disease. Can you develop it? Can you give me the right molecule to be able to address it? That we should be able to have that type of technology. Well, we all know drug discovery and development doesn't exactly work that way in that sense, but we are seeing what we need here is really good

tech entrepreneurs that will bring their expertise into the biotech world. And it's scary for these tech entrepreneurs. They're much more attracted to the cyber tech, to fintech, to other industries that are non -life sciences applicable. But in pharma, when you actually do get great generative AI and tech experts to come into pharma throughout discovery, you're able to offer these excellent entrepreneurs the opportunity to work on two things:

a very tough challenge that's hard to solve, and a good entrepreneur likes those type of things. And secondly is that by definition, if they're successful, they're going to do huge things for humanity. There's no way you can be successful in the biotech world without doing good for humanity. It's just impossible because if you develop a good scientific breakthrough, whether it's AI driven or biology driven,

you're bringing hope for patients. So what's better than that? And you can do very well financially by doing so. And there's no other industry in the world that by definition you could say that. So what we're trying to do at Aion Labs here is say, we've built this platform with a lot of great biotech expertise, a lot of domain expertise in the drug discovery and development world. Just bring us your AI skills. Bring us your gen AI skills.

Come tackle challenges with us instead of developing the next technology that can protect our computers from cyber attacks or that can help us exchange money and do cryptocurrency in a much faster manner. Come solve humanity's problems of developing drugs for previously undruggable targets to be able to bring new drugs to the market or even make the drugs that already exist much more accessible by

reducing the costs of this process dramatically and then make them more accessible throughout the world. And if you are an expert in generative AI, bring those capabilities into pharma. There's a huge opportunity to have positive impact as well as to make a lot of money and to bring tremendous value financially to yourselves and to the companies that you'll build. So that's what we're trying to sell and we want these generative talents to come in. It's scary for them, I'd have to say,

because they have to bring their technologies into a domain that's somewhat foreign to them and not their comfort zone. But the good entrepreneurs want to go outside their comfort zone. So maybe it won't be in the first startup that they build, but maybe in the second or the third one. And to bring that knowledge in, what's encouraging is that the pharma R &D experts that normally come from a bio background,

and from the other scientific disciplines of chemistry and biology know that they want to work with these type of people that come with generative AI capabilities. So they have the humility to say, we want these technological capabilities that are very different from the ones we bring to the table. We actually want to work with them. So they're providing, especially the partners that are coming to a place like Aion Labs from AstraZeneca, Pfizer, Merck, Teva, they're coming in and saying, we want to work with people

that would probably never go to work for a company like Pfizer -AstraZeneca because they like to work in flip -flops and wear shorts all day and solve big problems out of their garage. We want to work with them. Bring your technology to us. Bring your capabilities to us. We want to open our doors to them and actually work with them to bring new drugs to the market. And that's something that I think is very exciting. There's a lot of potential to do it. Absolutely. Mati Gill, CEO of Aion Labs. Mati, thanks for your time today. Thank you, Amar.

Well, Amar, what did you think? It's a very interesting model that he described, right, with the Venture Studio and bringing up not only the pharma, like the pharma leaders in terms of defining the questions, but also validating those and giving really like a very structured approach to forming the new startups and making making sure that

their failure is as less as possible and also taking really like the mundane parts of that out of the main, you know, talents for each, right? I mean, I can imagine that like a lot of people don't want to start a startup because of a lot of the headache that they have to deal with. They just want to solve the problem. So what they're doing is they're focusing the talent to just solve the problem, with them taking care of everything. They're also providing a structure to solve the problem

and also making them meet with other talent who they by themselves will not be able to meet by forming a startup. I am just helping getting the right people together to solve the problem. I think that's a very interesting approach. Well, these are guys who are basing a big bet on the promise of AI to improve drug development. The high cost, as he notes, is tied to the failure rate of drugs in the clinic.

Will that be the big payoff from AI ultimately to reduce failure rates? I would say the payoff is going to be across the board. See, the reason there is this 90 % failure rate of the potential drugs is that they were not validated well enough. The reason they were not validated well enough is that the systems to validate them didn't do a great job. And it's because

there just aren't that many good systems to do that. And as we talked earlier, right? So the animal models, yeah, you can use animal models, but a mouse is a mouse, a human is a human. It's not the same, right? There's a lot of differences. So you can validate something in a mouse, but then it's not going to translate fully into humans. So there are these challenges that they are trying to solve with AI. Now, of course, I mean, the challenges here are, as you know, with AI, if ...

AI is trained and the right amount of data it will provide an output around that. But then this is where we are actually asking AI to develop something really new, do predictions outside of what it may be even trained with. So, that's a challenge that there is to solve. And I would say, at some point, the failure rate will come down in the clinical aspect, but also we will be, hopefully, even earlier on in the drug discovery, we'll be able to

develop better drugs, right? So the failure rate is not just about the clinical side where the failure rate can come down, but the failure rate can also come down because it was a better drug from the beginning. So there are multiple ways in which the failure rate can come down. Mati offered a segmentation of Aion's portfolio, new drugs, drug discovery, improving precision in preclinical studies and clinical study attrition rates.

Is this a practical way to segment the world from an AI point of view? I would say so. So the first approach is more like identifying the right target. I mean, this is the standard way we think about pharma, right? So you have like discovering, so what we call target identification, which is more on the biological side. And then there's the drug identification, which could be chemistry or it could be a large molecules. And that's the bulk of research.

And then when you go into development, there's a lot of preclinical studies that happen and then the clinical. So, this is how we think about the pharma value chain, the research and discovery. So, it makes sense that even from an AI point of view. The techniques of AI could be applied anywhere and then the same techniques could be applied anywhere. But the way they're thinking about it is

where the problem lies, right? So like, and that a problem lies in these areas. And so then focusing the eye on those. So it makes complete sense to me. What do you think of Aion's approach with regards to how it's using its pharma partners? The phrase that stuck out to me was that they co -develop with industry to disrupt itself. Does this give them insight to make sure they're addressing the right problems?

Yes, and see a lot of these problems, the pain points, I mean, the pharma have the same pain points. They're all trying to solve the same pain points and solving a specific pain point will be tremendously beneficial for these companies. So I probed them on what is the incentive that the pharmas have, right, in solving these problems because they will be spending their energy

in working with these companies. So, and what he said is that, of course, I mean, then they get like the first knowledge if something is being solved. And also, I've been in R &D and there are a lot of startup companies that come to you and say, hey, we have this technology, that technology that's going to solve your problem. It's very hard for someone in a pharma company to know if the technology

is any good, if the technology is actually going to solve their problem. It's very hard to know until you actually go in detail, right, to really understand that. So here's an interesting way for the pharma companies, the benefit that they're getting is, of course, there is the financial aspect of the company's success, but more than that, you know, for the actual scientists, right, what's beneficial for them is…

as Mati described, it's a sandbox environment where they're actually seeing some of these technologies working on their problem and they get a sense of whether it's going to work or not, what is needed to be successful for those to solve those problems. So, I think that's a lot of value inside that the scientists are getting for those, which is very beneficial. And of course,

these four companies, because they are involved with these, if there is a breakthrough, they are of course going to be the ones who are going to know that the other companies are not going to have an idea. I mean, this is a huge area. You don't know exactly what's going on in the competitive area right away for those problems, unless you're reading the papers all the time. So, there's definitely the advantage in knowing that technology can solve a specific problem.

What did you think about the challenges they run and their approach to company creation? It's, yeah, I asked him about matchmaking, right? Because this guy, like, is bringing different people together. So one of the, I mean, first of all, definitely there is a challenge in terms of, in terms of setting up, getting the right question, right? And I asked him about like the criteria about that, right? Because there is the value. And of course, I mean,

We know what is the value of a lot of these questions, but the feasibility is a really big question, right? And then, as I said, a lot of the companies do fail, right? Only a few will be successful. So, how do you choose a problem where there's a higher likelihood of success or not? That's a very difficult problem in drug discovery. Then also, getting all these different people, putting them together as teams, creating this brand new

company is, I would say that's definitely a challenge because, I mean, we say in pharma, right, you can be a genius, but if you can't work with other people, you can't really add much value. And I'm sure they're working with a lot of very talented people, but those talented people need to work with other people. And that's a big challenge that I see. So that's something that they're dealing with. And then also making sure that everything else is working around them,

having the setting, really smart criteria about success because always when you have a new technology there are some successes there are some failures when do you make a decision that this is not something that you want to go forward with it. That's a very hard decision to make because there's always a technology is going to say hey yeah we're able to solve like 90 % of the-- well you know we're able to solve 10 % of this--

give me more money, give me more time and I'll do better. That's how they work. So you have to, at some point you have to say, well, this is where I have to make a decision about stopping this. That's a tough decision to make. Well, failing early in company creation is just as important as failing early in drug development. I thought it was interesting, his willingness to talk about Omic .ai and giving that example of Aion's decision to kill projects that don't work out. What did you make of that?

So that's leadership, knowing when to actually stop it, right? And then of course, when you make that decision, it is hard on the people who are working on it. And a lot of those people really believe that it's going to work, right? So then working with those people, as he said, he then, they're now working on a different projects with another company, right? So how do you, because the failure of that company is not necessarily related to the talent. It's because the technology was not the right fit

at this time to the problem. So, you have to really know that and you have to make those decisions and see, it's kind of like what we described is very much true actually in clinical trials where there is the 90 % failure rate in clinical trials. But again, there the question is, how can we know which drug product to stop early and how can we fail early rather than later?

It's exactly the same problem that you actually have in the clinical trials themselves. So, yeah, but it's a tough decision. Even in clinical trial, I've been in clinical development, stopping a drug and when to stop the drug is a very difficult problem to have. And how do you trim the portfolio is, of course, one of the biggest questions in pharma. Well, it was an exciting conversation to see what Aion's doing and

I look forward to our next discussion. Absolutely. Thank you, Danny.

Thanks again to our sponsor, Agilisium Labs. Life Sciences DNA is a bi -monthly podcast produced by the Levine Media Group with production support from Fullview Media. Be sure to follow us on your preferred podcast platform. Music for this podcast is provided courtesy of the Jonah Levine Collective. We'd love to hear from you. Pop us a note at Danny at levinemediagroup .com.

For Life Sciences DNA and Dr. Amar Drawid, I'm Daniel Levine. Thanks for joining us

Our Host

Dr. Amar Drawid, an industry veteran who has worked in data science leadership with top biopharmaceutical companies. He explores the evolving use of AI and data science with innovators working to reshape all aspects of the biopharmaceutical industry from the way new therapeutics are discovered to how they are marketed.

Our Speaker

Mati Gill, CEO of AION Labs, brings extensive experience in the pharmaceutical and biotech industries. Prior to leading AlON Labs, Mati spent 11 years at Teva Pharmaceuticals, where he served as the Head of Government Affairs, Corporate and International Markets. During his tenure at Teva, he developed the company's innovative strategy and built a robust network of partnerships. Mati holds an MBA and a BA in Law from Reichman University, underscoring his expertise in both business and legal aspects of the industry.