Dr. Peter Fedichev is the co-founder and CEO of GERO, a Singapore-based biotech startup that uses physics-informed AI and massive human health datasets to discover drugs aimed at slowing down aging and treating age-related diseases.
In the 2nd part of this episode, Peter explains how Gero builds generative models of human health to predict disease trajectories and identify shared biological drivers across multiple chronic conditions. He discusses why metabolic dysfunction and inflammation represent promising intervention points, how partnerships with pharma companies like Chugai accelerate translation, and why companion dogs could become the proving ground for longevity therapeutics.
Peter also shares his perspective on AI-driven drug discovery, arguing that target identification, not molecule generation, will create the greatest value. He outlines Gero’s open-source approach, fundraising journey, and vision for a future where therapies can meaningfully slow the aging process itself.
🧬 Aging may be best understood through shared disease mechanisms: Gero’s models identify common biological failures across multiple chronic diseases, revealing targets with the potential for broad therapeutic impact.
📊 Longitudinal data enables predictive biology: By modelling how health changes over time, Gero aims to predict future disease trajectories rather than simply classify patients as healthy or sick.
💊 Metabolic health is emerging as the first major longevity market: The success of drugs like Ozempic has demonstrated the enormous commercial potential of therapies that affect multiple age-related conditions simultaneously.
🤝 AI biotechs and pharma are increasingly complementary: Startups can generate novel biological insights and targets, while established pharmaceutical companies provide expertise in drug development and clinical execution.
🔓 Open source can accelerate drug discovery: Gero released its ProtoBind-Diff platform publicly because community-driven improvement can outpace proprietary development in fast-moving areas of AI.
🐕 Companion dogs could become the first proving ground for anti-aging drugs: Dogs age similarly to humans but on shorter timescales, creating an opportunity to demonstrate lifespan and healthspan effects more rapidly.
⚖️ Regulation follows evidence: Aging is not yet recognised as a disease, but strong clinical evidence for lifespan extension could prompt regulators to develop new approval pathways.
🚀 Longevity is becoming a trillion-dollar opportunity: As blockbuster drugs validate the economic value of extending healthy life, aging itself is emerging as one of healthcare’s largest future markets.
“We are looking at biological systems that contribute to the maximum amount of diseases.”
“If you have a drug that extends life by one year, it’s probably worth a trillion dollars in company valuation.”
“The only thing pharma is missing is new biology.”
“We’re building the best models of human health that we can.”
“In five years, there will be a demonstration of a strong lifespan effect in dogs.”
“If somebody generates data showing a massive lifespan effect, regulators will find a way.”
“We want to demonstrate a strong effect on aging, not just another incremental improvement.”
“Aging is such a big problem that it still doesn’t pay to keep many things secret.”
“I cannot imagine a worse enemy of humanity than aging.”
00:00 Gero Roadmap Overview
00:35 Predicting Disease Trajectories
03:03 Super Targets Strategy
05:07 Ozempic Market Wakeup
07:28 Binary Strategy With Pharma
10:11 Gerosense Wearable Aging API
11:58 Crossdisciplinary Lab Workflow
14:09 ProtoBind Diff Explained
17:16 Open Source Business Logic
21:10 Chugai Partnership Begins
23:57 AI Push In Big Pharma
27:09 Funding Gero Early
31:38 Differentiating From Calico and Altos
35:56 Longevity Roadmap Dogs To Humans
42:36 Next Milestones For Gero
43:45 Personal Motivation To Fight Aging
Episode Preview
Dr. Peter Fedichev: If you really want to focus resources on things that can generate maximum effect, you have to have at least some model.
And we believe that having a bad model is better than having no model at all, because a bad model can be discarded by a single experiment, which could be quick. So, that’s the culture that we’re trying to bring into biology of aging, because once again, since our experiments are very long, it’s like flying to Jupiter.
Joson Ng: Hi, thanks for joining us on Nucleate Singapore Pulse Singapore’s premier podcast on the biotech ecosystem. I’m your host Joson Ng, an MD/PhD candidate at Duke NUS medical school. Whether you’re a student thinking about creating your own startup or an industry professional looking for diverse perspectives.
This is a podcast for you. The show notes and transcripts for the episode can be found on nucleatesingapore.substack.com
========================================================
Joson Ng: So maybe we’ll go into Gero’s product roadmap more specifically and the products that your company is developing. So what does Gero’s product roadmap look like today, and are there any specific diseases or pathways that you’re also looking at as first targets for intervention?
Dr. Peter Fedichev: As I said once again, the mission is to produce a non nonsense prevention against human aging. So for that, you have to develop models of human aging and our models, generated models. We take medical data today. We take medical data in a few years. High resolution medical data still doesn’t exist, but once again, for slow processes like diabetes or aging, even two years is a short time. So our systems are taking the medical history prior a certain date and try to predict the next interaction with the medical system for an individual. So it’s like generative models in text, predicting the next token, the next word. And in our case, we’re trying to predict what will be the next disease or the next medical problem for this individual.
If these models are trained correctly. On the inside, you will have the entropy, temperature, and a bunch of other things that together produce the prediction. So now you have what biologists would call phenotypes, and those phenotypes are controlling the prevalence of disease.
And interestingly enough, it’ll happen pretty quickly. If you look at this data, aging it’s not reversible. It’s just time. If you look at diseases, well know that there are certain diseases that are easy to get and there are diseases that are hard to get right.
To get Alzheimer’s disease, you have to be an exceptionally healthy individual. Live up to 80 years first. You have to be healthy. Then you can manifest Alzheimer’s disease. I mean, it’s bad, but not everyone is able to do it. So what is the difference between metabolic disorder and Alzheimer’s disease? What makes Alzheimer’s disease a lot more difficult to get?
This is very important question because we know that all those diseases don’t have a single reason. There’s no single gene. So aging has a lot more important factor than any genetic risk factor for those diseases, and it turns out that the less frequent is the disease, the harder it’s to get, the more failures you have to accumulate in your underlying biology in order to manifest the disease.
So every complex disease is a bunch of failures. Diabetes is a lot of failures. Cancer is also a bunch of failures of oncogenic mutations. It turns out that different diseases have different numbers of failures. And some of those failures are shared across diseases. For example, if you want to get Alzheimer disease, it’s very helpful to get metabolic disorder first. It really helps. Also fibrosis. So there is a like metabolic dimension and there is an immunity dimension, inflammation almost. And this looks almost like the hallmarks of aging.
So what happens is that of course diseases is just a counting tool. All these CD 10 codes like diabetes, all the names of the diseases, it’s just a historical anecdote. Actually, there is more fundamental biology that in combinations produce those diseases. And if every disease requires combination, there are certain failures that are shared across different diseases and they produce target spaces for pharmacological development.
And if you take those that are shared across multiple diseases, like Ozempic for example, that’s something that affects mostly metabolic failure. But at the same time, drugs against it reduce risks of thousands of other diseases. So there is a component of metabolic failure in many other diseases.
What Gero is doing specifically is that we’re looking at these super targets. We are looking at biological systems that we infer from the data that contribute to the maximum amount of diseases. In the hope that if this idea is correct, in each such cluster of features, we can identify a few targets that could be outsized in their therapeutic impacts.
Of course, we believe that metabolic system drugs will provide the largest effect on our diseases. That’s why many longevity companies are now mutating into metabolic health companies.
But other than that, there are other systems that fail, for example, the sterile inflammation is another component in many diseases, and we’re trying to generate targets and eventually drugs there.
So what we’re trying to build is this. We’re building these hopefully best models of human health in the world. These models give you something about aging, which is like super fundamental and this is cool. And then on top of that, they build you reliability theory, representation of human disease.
You look at the most important features that are controlling multiple diseases. You identify targets there in your experimental validation. And that’s where you collaborate with pharma because we believe pharma knows a lot more about diseases than us. We’re just helping them to provide targets with human evidence which is important because all our calculations are in human data. And at the same time we cherish our aging part of the model output. Because that lets us focus on drugs that can affect human aging. And we’re trying to do it at the same time because it gives us reputation and experience.
And when you start talking to pharma companies, we’re saying, okay, but we’re not showing you the aging part. And they say, why not? So I believe that the pharma is increasingly interested in aging as well. And who knows, maybe at some point on this trajectory our interest will meet pharma not only in diseases like it is right now but maybe in aging itself.
Joson Ng: For sure. I think I definitely see it coming, especially with the recent talk. Several people, Brian Johnson for example, the billionaire who’s holding conferences on longevity and how to stop aging. I think pharma companies are definitely gonna start looking at it, especially given how, as you mentioned, Ozempic as well as seems to be interesting to use for a lot of different diseases.
Dr. Peter Fedichev: I think it’s a very interesting point. ‘Cause as you know, the cost of developing drugs are huge. And for a while people were trying to take less technological risk doing orphan diseases, genetic therapies for kids, biomarker driven clinical trials in oncology for very narrow patient populations.
What happened with Ozempic is that once a single drug would generate 50 billions in revenues, suddenly the tech industry was able to see biotech under their microscope. Because at that 50 billion in revenue, the tech guys actually recognized that biotechnology is a business. This is actually a manifestation of the aging population.
So there are also other drugs like Keytruda that are selling for 10 billion, and by the way, all of them are going off patent right now. So we’re three years for patent life now. In August was got a value boost of half a trillion dollars from Ozempic. So now you have a rule of thumb. If you have a drug that extends life by one year, it’s probably worth $1 trillion in company evaluation. So it’s a good business plan. We now have a measured point. It’s not fiction anymore. We have the hard experimental point. So obviously if you have a good idea how to build a company that may extend life by 10 years, that’s quite sexy.
Joson Ng: Wow. I can imagine.
Dr. Peter Fedichev: Well, I actually saw one of the early presentations of Elon Musk. I talked to a person who was invited to invest in, started in SpaceX when it was like very small company. And the first slide was there like a picture of an asteroid. And there was like 140, 14 trillion dollars of palladium there.
That was the total accessible market. And then the next slide was about technology. Now if, if you try to build a longevity company, you’re almost in that situation. You can say that, look, one year of life is $1 trillion.
Joson Ng: Imagine 10 years of life.
Dr. Peter Fedichev: 10 years, let’s go into the details. So what we’re trying to build, actually, like with this SpaceX example, that gentleman is in Texas, wants to fly to Mars and even with his fundraising capacity, and nobody is giving him $150 billion for that enterprise.
So that guy had to think what to do and he looked at there is a business that roughly is worth $150 billion. And that’s like satellite communications. So he decided to build a company that is doing satellite communications, roughly order of magnitude. Try to estimate for the company that is required for the amount of resources that is required to fly to Mars.
I think in the aging field where roughly the same situation, like everyone wants to fly to space, but nobody wants to, to spend money on drugs that don’t cure specific diseases, maybe working or not, who cares?
What I think works is this binary strategy. Why do we do identification of novel targets? Because our mission is to solve aging. If the new targets are against aging, we would be doing drugs. Since we need to find out the targets against human aging, we have to do models of health and we are trying to do targets.
Once you start doing targets, accidentally, you start generating targets against diseases as well. So we believe that this target identification for diseases is our Comcast in this business because pharma companies are now in a very interesting situation after Ozempic. So all the great drugs are coming off patents.
In order to protect their earnings, they have to generate new biology and that’s precisely what they have forgotten to invest for a while. They are doing drugs better than we. We cannot teach them how to do drugs, so the only thing that they’re missing is the new biology. That’s why we have high hopes, and they have high hopes that with all this, lots of data, with all this new AI coming, there should probably be a way how to generate valuable data on human biology faster.
And look, founded by a bio-age and in-silico medicine, all these AI companies that are now doing deals with the bigger pharma. Why? Because pharma knows how to do everything else. And I think this makes a very interesting situation on the market. I mean, 10 years ago, in order to earn money in biotech, you had to do phase two.
Last year there was a R&D level deal with 100 millions upfront by AstraZeneca. Because for pharma, their pipeline is empty, meaning that they are now scraping everything that looks like a drug because to make a better Ozempic, how many new Ozempic you have to put into clinical trials and hope to get something that can compete.
Joson Ng: It’s not very cost effective to you.
Dr. Peter Fedichev: So basically, A, the people will have to buy a lot of new biology and this is a good time for companies like us, or they will look at this and say that, okay, Ozempic is 50 billion in sales per year. What should we do next? And everything else is less than that, or we should start doing aging because aging is now the only way to escalate from the position where these companies are. And if you look at this venture arm of Eli Lilly, they are telling we want to do aging. So I think this is a sign of the future to come.
Of course everyone else has to run and catch up. So this will keep people busy for a few years, but then people will start thinking what to do next. And the only next thing that can match Ozempic and affect earnings is aging. That’s why I think that the future is good. That’s why we developed these two arms, we want to talk about diseases with pharma. We want to talk about targets that has maximum effect on multiple disease as a big target. And once we make people think about it, they almost immediately start thinking about aging.
Joson Ng: So Gero also recently launched an API Gerosense, and that measures biological age and resilience from wearable data. So how does it fit with your overall R&D strategy?
Dr. Peter Fedichev: So, we did Gerosense as a research product. So although through API, a few wellness companies are using it, the goal of that was to collect what is called longitudinal data at high frequency. Longitudinal means that every person is measured multiple times.
As you know, the problem with clinical data is that, in the good senses, how often a healthy individual would go and do their blood work. Once a year. I mean, that’s already kind of borderline for many people, especially for young people, meaning that whichever ideas we have about the nature of biological noise effects, of biological noise on, on incidents of diseases, you cannot test it reliably with medical data because medical data is overrepresented with sick people, right?
You only get high time resolution when people are almost dead, like in emergency care department, they do have high quality and longitudinal measurements. But you don’t have them on healthy people. So that’s why we built this API to collect, of course, much less quality data. Of course verbal signal is not as good as clinical blood analysis.
But the power of it is that it’s very high frequency. So you have measurements like once in a few minutes. And with that, we published the work where we confirmed this measurement of the maximum human lifespan. So the primary goal of this application was to collect scientific data. So we collected more than 100,000 data streams from people.
We published two papers with them. So we kind of got confident that what we believe about aging is right. And yes, we’re serving it through API to wellness companies because that’s precisely targeting healthy people. But that’s not like our huge commercial product. It’s almost like we sent a cheap satellite into space to see how aging occurs in healthy people. And we collected very interesting data with that.
Joson Ng: I see. But also tangential to that probably is how does Gero approach experiments in R&D in the lab? Like how do you integrate computational?
Dr. Peter Fedichev: That’s very important question because we’re trying to keep Gero as a very crossdisciplinary team. So we believe that, if there is no data science and physics, then we are no different than any other startup. We’re trying to keep inside people with these strange competencies, our skills. At the same time, to do high level biology, you have to have ultraskilled people. And our current mode of operation is that research heavy issues are done in collaboration with top academics.
So we’re working with Longevity Center in NUS, for example, with people like Brian Kennedy who are like world-travelled people doing the biology. So whenever something is strange on you or coming out of our twisted physicists heads, we’re going to great biologists trying to convince or maybe sometimes we’re hit back.
But we look at this as a highly collaborative work. So we are suggesting what we hope are somewhat common ideas in cooperation with them. We are building ways to establish if these ideas are true or not experimentally. And then of course, once we start doing things more commoditized things, we are working with contract research organizations like many biotechs do.
So we work with people in India, in China, in Europe. So whenever things get standard and we are trying to do it in contract research. And also over years, we were able to build an exceptional team of advisors. Some of them retired big pharma people of very high level. Whenever we have a specific chemistry issue, we still do it in clinical research organizations, but we’re able to talk to very interesting people to help us do the actual planning and execution of discovering development programs.
Joson Ng: Right. You also mentioned a while ago that a lot of your research findings, you also try and publish papers with it?
Dr. Peter Fedichev: Yes. Once again, aging is so huge that you have to be influenced by the community, but you also have to influence the community. Because if what you’re saying is too orthogonal to the consensus, it’s also hard emotionally, but it’s hard to get attention of investors, pharma. So you don’t need to do what everyone else is doing. We don’t have problems with that. We really do stuff, which is very orthogonal, but we try as best as we can to convince the community that what we’re doing is at least a good idea to try.
Joson Ng: The other product that Gero also has is a ProtoBind-Diff, which you’ve described it as a mask diffusion language model that uses pre-trained protein sequence embeddings. I think one of them was from Meta’s ESM-2?
Dr. Peter Fedichev: Yes from Meta.
Joson Ng: Yeah. And you also use the noising diffusion frameworks that generate small molecule ligands, which are then tested by your partners as well?
Dr. Peter Fedichev: Yes, well, of course as usual, we do it for ourselves. But, first of all, we open source it. There’s an interface on Huggingface, so everyone who has a faster sequence of a protein can actually upload it and the algorithm will generate molecules that we hope will bind to the protein.
The reason why we did it is that first of all, we do novel targets and for novel targets, it’s often important to generate something that looks like a chemical probe so that we can do cell assay, sometimes in vivo work. It’s also important to get some IP. So as you know, targets are not patentable.
But if you have a new target and a molecule and a bunch of experiments that this molecule affects the target and produce the biological effect. It’s easier to file a patent. It’s easier to do collaborations with pharma when you have something tangible. So that’s why we’re always looking for ways how to generate this chemical matter quicker.
We decided to do a generative model for that is because people have been trying to score molecules against proteins. And as you may understand, this is very hard problem because proteins are changing all the time. They’re moving even with AlphaFold, you have just one conformation and conformations change. Sometimes there is no pocket for your molecule and they predicted information.
So, I think methods that rely on 3D structure may not work very well. So we were thinking that maybe nature has already solved this problem. Not by building a quantum computer, but just by selecting. Maybe nature operates with molecules that feed proteins and have some structural relation, which is stable under the mutations in proteins, which means that in all the biological active molecules should be written in our genome in some way. Because whenever you put DNA into a cell, from these DNA, you produce everything, including proteins and metabolites of all the active molecules. So technically everything is written in the sequences. We try to escalate it as a thought exercise. If everything is written in the sequence, then there should be a generative model that generates active molecules from the sequence.
And we were delighted to see that if you build such a model and then you try to score these molecules with AlphaFold or Boltz, then you see that suddenly goes Boltz and Alphafold like those small molecules, which means that these molecules are non-central. And if this is true, then it’s very good because obviously structure based methods can be trained only on complexes of protein and molecules that have 3D structure. And less than hundred thousands of them. But this kind of molecules could be trained with anything that has activity. And there are millions of activity points.
So, we release that as a proof of concept that this can be done. And we are now asking our operators doing that lab experiments ourselves in order to see how it worked for novel proteins. Now it’s a preprint, once the experimental validation comes in, we’ll pack it and try to publish.
Joson Ng: I see. You’ve also released this on GitHub as an open source, right? And you’re also eventually releasing this for a public web demo?
Dr. Peter Fedichev: Yes, the demo is already on Huggingface right now, so everyone can use it.
The reason is very simple. This is our understanding of the business model. So I think everyone understands that for a few millions of dollars and for a number of years, you can get a molecule that hits almost any target. There are ridiculous exceptions to that, but most of the time it’s true.
Which means that if you have an algorithm that outputs immediately, not just an active molecule, but the molecule that can be denominated, at most you can save a few million dollars. Which means that kind of the upper bound of what you can charge for this million dollars. Remember I told you that this algorithm has to output the molecule that is ready for clinical studies. We’re not any close to that. Of course there is this hype there are these deals like Isomorphic with Google to big pharma, with billion dollars of buybacks there.
But we believe that when the dust settles on the efficient market, this will be a zero job that will at most charge hundreds of thousands of dollars per job. And this is like already much, you cannot make a trillion dollar company. This is not a venture, that’s not something that a venture-backed company will do. So I’m predicting that most of the companies that are using AI and biology. They can do AI for target identification. They can do AI for molecules. They can do AI for patient certification. They can do AI for actual medical applications.
I believe that the particular application of AI for molecular generation is overpriced right now by the market. Because this is not the most expensive part. If you do a mistake with a target, you learn about it in $50 million and five to six years from the solution. So if somebody can double the chances of successful phase two, this job costs tens of millions of dollars. But if you just generate a molecule. People can just screen sometimes, you are competing with people who are screening and they’re charging millions. That’s why a small company cannot do everything commercially.
Scientifically, we generated an interesting proof. We allow letting it out into the wild. We hope that there will be a community built around that. We are sponsoring some hackathons right now. In September, there will be a hackathon in San Francisco where we participate. This will be for longevity. And on that hackathon, one of the competitions make AI agents that would scrap patents and journals for more data for ProtoBind-Diff. I still don’t know how things will unfold. But maybe if we just let it out, people will benchmark against it, people will improve it. Some people will scrape more data, and then maybe in one year we will download a better version of ProtoBind-Diff at no cost to use in our products.
Joson Ng: I see. So you’re just letting the community help iterate and see how they can improve it further?
Dr. Peter Fedichev: Yes, because once again, we believe that this is the wrong stage of drug discovery to optimize with AI. And at the same time, many people need it, right? So many people need chemical probes for research. Some people can have more data than we have. Of course, we try to make sure that people who are developing on top of it have to give it back with the license.
I don’t know how this will work, but really we need it. It works. It may work a lot better, so if this picks up, I mean if other people reproduce it and see that it does work well on benchmarks? Maybe, I don’t know? there will be a team that will make it 10 x better. But still on GitHub. And then next year will just download it not from our repository, but from there, right? So as soon as we see a paper that somebody makes our algorithm benchmark against their version of generative algorithm and their version is better.
Joson Ng: And you just find where it got better?
Dr. Peter Fedichev: We’ll use it there. So I think that’s how I, I’m a true believer in open source approach everywhere because if somebody is publishing that they have an AI mamba jamba that is 10x better than everything else, and I cannot test it. It doesn’t bring a lot of value. Mm-hmm. I mean, it’s good for, for a press release, but most of the time it’s not, it doesn’t generate the value. I think that if in some small way, these things will help build better methods and we’ll all get you know, the value in the form of better drugs eventually.
Joson Ng: Mm-hmm.
Dr. Peter Fedichev: I think this, I mean, these algorithms will work in some way and I think they will be commodity, so I think they will be eventually free.
Joson Ng: I see. Yeah, that definitely makes sense. But now perhaps to move on to one of the big news that you guys have recently, now you’ve partnered with Chugai. What stood, what stood out to you about Chugai as a partner?
Was it their platform for or expertise in antibody engineering? Was it their reach or their expertise as you mentioned, with big pharma companies?
Dr. Peter Fedichev: Oh, first of all the team we partnered with, I mean, we partnered with the company, but of course it’s a relationship with a specific team. Mm-hmm. And this is the team from Singapore, by the way, which could be important for
Joson Ng: specifically in Singapore?
Dr. Peter Fedichev: Chugai Pharmabody research which is just a few hundred meters from here, I think. Right. So I mean, this is what, this is one interesting point.
The second point, I mean, on all the levels of communications we had such a, you know, wonderful, you know, ethical, respectful communication all the time, even by scientific interest, which made the whole communication, I hope, the huge corporation just very interesting.
So these people have been doing best in class therapeutics all their lives. I mean, they’re probably the best in antibody development. And they are driven by scientific curiosity a lot. So they are looking for ways how to do now, maybe for the first time, first in class, which is than best in class at the same time.
Right? Right. So they, they’re trying to, they’re challenged with this idea that probably they need to suck in more new biology. They know how to do drugs and they, this makes them a perfect partner for us. Because once again, we’re small. In no way. I think, believe me, there is no way that I can build a team that will develop an antibody better than Chugai. I mean not in 10 years, as I’m saying, maybe never say never, but not like in a short time.
So I think that the deal we are having with Chugai, or like BioAge is doing with Novartis or Insilico is doing with Sanofi is indeed the recognition of this mega trend. Pharma companies do everything in drug discovery much better than every small team, but the only piece that they do not yet get, and they will eventually get it as well, is that as we started from biology is complex.
So they need somebody with either super advanced AI or maybe people with interesting biological ideas. That would help them to identify relevant human biology with exceptional effect size because to get a pharma company interested, you really have to have a strong effect size. And I say we’re very happy about this cooperation because it also solves a problem for us because no matter how fun my stories about entropy and temperature imagine before it is proven in clinical trials, it doesn’t cost much.
Of course, we’re trying to do it in dogs and animals, but who cares about that? So I think that in no world we can do, we can reach clinical validation faster than we can do with a major pharma company. So that’s why we’re so much inclined to do collaborations with pharma companies because only once there is first, and some people will still say first doesn’t matter.
Few clinical validation. Everything that I have just told you so far, I mean it’s, it’s a fun story, but not yet a validated thing.
Joson Ng: I see. You mentioned that Chugai definitely has a lot of experience working on best in class therapeutics, and now that they’re partnering with Gero that they’re finally starting to look at first in class.
Dr. Peter Fedichev: Yeah, I think it’s, it’s a trend. You can look at other sources, what’s going on there. I think they are now making a big push in AI.
Joson Ng: Mm-hmm.
Dr. Peter Fedichev: There was recently a deal noted with SoftBank, so together with SoftBank, they are training agentic AI. Because look how, I think, how much of experience company like Chugai has, I think if they could train agents to help with every piece of drug discovery process.
I mean, they are probably one of the best teams in the world to train AI how to do drugs. Yeah, so it’s interesting to see how this company is trying to encompass, like also Sanofi, like also Novartis, all these companies, I think all of them understand that when Marc Andreessen starts managing half a billion venture fund, it could be eventually a company that will be bigger than any of them if it’s done correctly. So that’s why pharma companies have infinite amount of money on a kind of personal level, on personal scale of all things. And I think they’re trying to embrace, each of the companies is doing it in a different way. And I believe that once again, the combination of AI and data drug development capabilities is unique.
You can never have a venture funded AI driven biotech that at the same time can do drugs at the same rate. I think, as I said, this is a mega trend. They need, they know how to do everything other than to identify new biology, new human biology, and they will learn that eventually, not even eventually, they will do it pretty quickly.
Joson Ng: Right. But then comes a problem of how do you then go and manage IP strategy when you’re working with someone with that much expertise on the other end?
Dr. Peter Fedichev: Well, it’s of course that question because you can only know that later on. Yeah. But once again, the team over there showed, I think, very high ethical standards, respect, understanding of our needs as a small team. I don’t think they want to exploit small teams.
I think they, look everyone who’s doing drug discovery, all of them are for a long game, right? There are no people, there are no half-hearted people who are doing drug discovery, right? So I think that they’re very science driven and they don’t want to starve small teams that can contribute to the AI driven future, right? I think their best strategy is to embrace teams like us in the most, let’s say, friendly way. And of course it’s still very competitive because I think obviously Gero is not the only team that is helping them there.
So I think by showing respect, good business practices, they eventually will be able to attract the best companies and select for the best of the best. So I don’t feel like there is a, of course everything has to be, you know, formally arranged and everything else. But I think everyone understands that we’re living in a very interesting moment in time where this new technology will empower those few big pharma companies that will do the transition correctly.
And I think it doesn’t pay, it will pay off to do it, you know, properly.
Joson Ng: Right. And before they get to the point where they can do it themselves, it’s still good to have that proper cooperation.
Dr. Peter Fedichev: Yes. Because this also validates us. And the terms are also very good. So I don’t see really, I think, yeah, for those people who are in biotech, I think we all understand that.
Of course there are, you know, business issues, challenges, and everything else. But the foremost challenge is the complexity of the beast. We are, we’re trying to, to fight with. Right? Compared to that, everything else is a minor point. Right. Really the complexity of underlying biology is the major challenge. Everything else can be arranged just by conversation.
Joson Ng: I see. Now I want to talk more about Gero’s financing journey. So it’s quite interesting because you straddle both biotech and AI investors, so sometimes these are separate sorts of groups of people, but now in this case, you’re trying to attract both.
And to recap, Gero raised about $5 million in seed funding in two trenches very early on. And you recently also closed a $6 million series A extension round.
How did you secure those early investors back then, and how was it like to pitch an AI driven longevity startup back in 2015 when AI wasn’t as common knowledge right now because of things like ChatGPT and so on?
Dr. Peter Fedichev: Well, of course also even the language was different of course because at that time it was more like machine learning and data driven discovery and it was physics based and longevity focused. So that’s what we were pitching, the combination of thankfully our investors are high network individuals who are either from personal biotech experience or there are some NASDAQ level IT companies or people who earn money by selling their AI startups to Google, Facebook, and others.
I think aging/longevity is a very interesting business model. I don’t think there is any business model that failed more investors than the promise to develop a drug against aging. Right.
In certain ways it’s hard to pitch because whenever you are saying that you’re trying to achieve a large effect, you look strange. You look nuts to many people. So that’s why some biotech funds have high level of allergy to aging and to longevity biotech. So, I mean, up until very recently, longevity biotech was very much on the fringe.
At the same time, people who are coming from AI and IT, they have a different mindset. So their mindset, I mean these are people with lots of newly acquired resources, mostly with technical engineering, mathematical backgrounds, so they understand this data driven and machine learning part very well. For them it’s very easy to explain that. All those things that we started to talk about first, like that we have to have a certain level of modeling before actual trying and so on.
I think what happens right now is that people understand that AI is a transformative tool. And they believe that many business models of today should include AI in some ways. But of course, AI works somewhere, doesn’t work somewhere. So these kind of people are trying to invest in companies that are built on the intersection of AI and something else.
So we have people who do AI and weather, AI and agriculture, AI and longevity. Why not? Right? So I think that for many investors now, this is a part of their personal journey where they believe in AI as a technology, but they also want, just by trial and error to see where it works, and what it doesn’t. I think they are training their neural networks to understand where these tools are useful.
By deploying investments, working with the teams, and since they have kind of extended experience and application of heavily math based AI tools in different industries, it’s also very helpful for startups to get to talking to them because they can share lots of business experience. So I think that, well, it’s like try or die for biotech funds.
I think many biotech funds like professional biotech funds, and that underestimated that have been underestimating, aging will fail in many ways. And yeah, AI is a force multiplier right now.
Joson Ng: I see. And I guess everyone kind of knows that it was only until recently that the conversations started to shift about with on, on how attractive longevity is with successes like.
Dr. Peter Fedichev: Yeah, we were teaching longevity from the start. I mean, that’s why our trajectory is a bit longer than usual, so we didn’t pivot to AI and drug discovery and then we, it was very tempting at some point, but we were feeling that there is infinite amount of, number of AI and drug discovery startups. It’s very crowded there and it’s very, very hard to achieve any differentiation. And now differentiation from, from the day one was that we want to go against aging in a strong way.
Joson Ng: And how challenging was that when you were first starting on, when, when you were talking about aging, but you know, that conversation just hasn’t started yet overall.
Dr. Peter Fedichev: Oh, it was challenging. I mean, it’s still challenging with professional funds. I think it’s very interesting so when you talk about the novel technologies, not all money that are coming early in startups are what is called smart money because you have to be, let’s put it adventurous.
I think what helped me personally and for the company is that I had very good science pedigree at that point. So my reputation in physics was quite strong. So people were able to get references. So I was telling strange things even on a scale of strangeness for me now. But I had good references and people were listening, so I think it helps.
Joson Ng: Definitely. I see. Okay. I think some of these things we talked about a while ago about Calico from Google. We have now Altos labs as well. Getting billions in funding for longevity and AI discovery. How does Gero differentiate yourself from these companies?
Dr. Peter Fedichev: Well, I mean, first, first of all, they are a lot more, a lot better funded than us at the moment.
At the same time, I mean these are very different companies here, so, mm-hmm. For example, Calico spent lots of money and lots of years to study biology of aging. Calico was the company that funded the experimental validation of non aging in mammals, right? So with their budgets, they’re able to do essentially fundamental science.
And then recently they have acquired a startup that tested an antibody against aging, which is roughly the same stuff that we out-licensed to Chugai. So to me that means that both Calico and Chugai got product from a specialized small biotech. Right. So it’s probably an interesting signal. We know that 75% of drugs originate in small biotechs.
So these guys, I look at these guys as big biotechs, especially in longevity. They have financial muscles to do clinical trials eventually, but still, for some reason, discovery is done in smaller teams. And I think there is now almost no exception to that. If you look at the Insillico Medicine, for example, they became a machine for producing computer generated drugs in clinical trials, but they’re not focused on aging specifically by age mutated into metabolic health setup.
As I said, it’s a decent strategy because, and by combating diabetes type two, you can extend life. So technically it’s longevity company, but I also provided your arguments that this doesn’t actually intercept functional decline. So this is not like a true longevity drug. So I think what we are trying to achieve here, we’re trying to have focus. We don’t want to drop the idea that the company operates to discover a drug with a lot better effect than this.
Altos Labs is built mostly around the epigenetic rejuvenation. This is a super interesting technology, which we don’t believe will produce a strong effect on lifespan. To me, this is a new therapeutic modality, so they will earn a lot of money, reputation, save millions of people by producing novel types of drugs for people. But once again, we don’t see the intellectual power there working precisely on strong effect on aging.
I think it’s interesting, to put it this way if you look at all these companies, it was very hard to pitch aging in a strong way to investors. So most of these guys pivoted to the idea, okay, we’re not like really aging, we are just extending life by 10 years. And investors understood that 10 years is a lot, but not like crazy a lot. And of course if you hope for 10 years, you will get two years eventually if everything works out. So that’s why most of these companies are now mutating into biotech or pharma companies.
So to me it’s the scale of ambition. If you limit yourself by something that your investors are ready to believe and you will get quarter of that. I’m not saying that it is bad business, some of them will produce terrific drugs. But I’m saying that almost all of that still falls short to the original dream. We hope for another way for companies like us, we’re kind of delayed in development because of this position on aging. But I think that a new way for companies is now taking off and people are thinking about organ replacements, about stem cells therapies for the next generation.
I think in a few years from now, there will be new consensus about the next generation of longevity companies. I think Gero will play quite a role.
Joson Ng: I see. I also kind of wanted to explore how you interact with other longevity companies. Is it mainly competition or is there a lot of collaboration? Because we also talked about how you put ProtoDiff as an open source. Is it a lot of knowledge sharing that’s happening through publications? Through conferences, or is there like something you hide behind like a corner?
Dr. Peter Fedichev: Well look, the biotechnology is interesting because like Peter Thiel says that businesses without monopoly are for wings and biotechnology have very strong patent protection, meaning that most of the time you can talk about your invention or even publish your invention or with a quite high degree of disclosure. That’s why I believe aging is such a big problem that people find important to talk.
We know most of the people working in other companies. We go to conferences, we explain our position on aging, we show specific results. I think there is quite a level of disclosure because once again, the problem is so huge that at this stage, it still doesn’t pay off to hold many things secret.
Joson Ng: I see. The other thing I wanted to ask you is your perspective on the whole sector, or the field in general.
Dr. Peter Fedichev: Yeah, I can tell you. I often pitch that medicine is boring. It’s very easy to do 10 years prediction. Cause given the timescale of the development, things that will be hot in 10 years in patients are those which are now published in Nature papers. So essentially, you know the future.
And the future looks like this is that we have now, over the last 20 years, $20 billion went to companies tackling different hallmarks of aging. As I tried to explain to you hallmarks of aging, actually disease. So those companies will take perish, or some of them will be companies holding drugs against specific case related disease.
Some of them are in metabolic space, others are in neurodegenerative space. So we will have the whole generation of companies that exploited features of aging biology to cure specific age related diseases. In terms of lifespan extension, that would be a few years, maybe together five years. Well it’s actually interesting ‘cause if our life spans are now increased by five years, less than five years, that are required for them to complete clinical trials. Technically, we’re not aging right now, but for a moment. Imagine that now our conversation occurs in some form of suspense. The time is not ticking, because at this time, biotechnology extends our life at the rate that is faster than we’re aging.
But since this is just the final number of ideas, once this effect is exploited, we start aging again. The suspense will stop and will keep dropping. What happens next is that we believe that there may be a new level of therapeutics that could try to either reduce the difference between average and maximum lifespan. So bring us to the maximum lifespan.
That’s what we’re trying to do. And there would be therapeutics that will possibly stop aging, maybe some kind of replacement strategy. We believe that developing a therapeutics that brings us to the maximum lifespan is technologically super feasible, can be done. So we hope that it’ll be demonstrated first in dogs because dogs still age much faster than humans. But in the same way, biologically they are not like mice. This is kind of our contribution. So I focused that within five years there will be what we call level two, that is there would be a demonstration of a very strong lifespan effect in dogs.
And by the way, regulators in the States now allow (companies) to sell drugs conditionally to dogs if there is a reasonable expectation of efficacy. Not a really successful clinical trial, it’s limited in time, so you have to collect the data to convince the regulator that the thing works in a few years. But still, it lets you immediately sell, generate very large cohort, which is important if you measure aging.
So I think that let’s say in five years in pets, there will be a demonstration of a strong effect. And biohackers like Brian Johnson will take it, and we’ll see.
Joson Ng: With all his money. Sure. I think there’s something that will come out of it.
Dr. Peter Fedichev: And then I believe that closer to 10 years there would be a breakthrough in terms of reducing the rate of human aging because there’s interventions that bring us to the maximum lifespan. They don’t understand function decline, so I believe that with high probability in 10 years we will have something that works toward the maximum lifespan, like therapeutics, not really scientific demonstration. And I hope that within 10 years we’ll have a demonstration of something that reduces the rate of aging in humans.
Joson Ng: And that’s how many years into the future is what you’re expecting?
Dr. Peter Fedichev: I think that if in five years we’ll see something in dogs, it means that by the end of 10 years, we’ll have something that works towards the maximum life span in humans in clinical trials. So I would say that we have like three years and three years we learned with things like we have Ozempic is just one example.
Of that I believe that in 10 years we’ll have the final stages of clinical trials. Something that affects multiple diseases at the same time, brings us to the maximum lifespan. And in 10 years there will be scientific demonstration of a drug that you can reduce the rate of aging in.
Joson Ng: Can’t wait for that to happen.
But also at the same time, I realized that currently aging itself isn’t really recognized as a disease per se by regulators and all of the drugs that are out there on the market have to target a specific disease as per FDA regulations. Do you foresee that sort of changing in the near future where sort of regulatory agencies start to create a framework for targeting aging?
Dr. Peter Fedichev: This is a very important question because as I try to convince you, focusing on diseases kills aging research, unfortunately just for fundamental biological reasons.
I see the signs of change in the regulator’s mind. So this example with the dogs is a strong example. So first of all, they allow to sell drugs before the clinical trial ends. And we have this now this legislation in the States that extends federal right to try to non lethal diseases. So we have more and more evolution in the regulatory space where people are getting access to experimental drugs earlier and earlier. But in dogs, you can do antiaging drugs right now because pet owners would buy lifespan as a product.
In humans, it’s still not possible, even with regulation, you still have to cure a disease. I think that many people complain about that and I even hear sometimes investors are saying that since aging is not a disease, you don’t have an available business plan. Now I’m saying that, okay, I’ll do it in dogs and we’ll see. But I think what is important here is that the problem is maybe in reverse.
There is no regulatory change because there is no strong enough effect to think about. So if somebody was able to generate the data that there could be a massive lifespan effect of a certain drug, not like on the level of diet and exercise, but let’s say twice more than that. I think that regulators would start thinking about it.
I think medicine is, in a good way, conservative meaning that if there is no problem, they’re not solving and they have enough problems. So we really have to show them that something may exist first, like this reasonable expectation of efficacy. And believe me, given the scale of demographic transition, the effect of the elderly population on the economy, medical services, and everything else. If somebody comes with reasonable expectation of efficacy there, I think they will scratch their heads. And for that, they will find a way, like with vaccines during COVID times, if there is a problem and the problem is there and if there is a solution. I mean, to find the regulatory solution, I think this will be not a big deal.
I think Loyal is helping tracing the path, establishing the path right now in pet animals, longevity drugs. This may be a way to go for many companies, including ours for a while. If a strong effect is demonstrated there, bio-hackers will use it anyway and somebody will find the regulatory path.
Joson Ng: So fingers crossed, once we settled that missing piece, that strong convincing piece of data.
Dr. Peter Fedichev: It has to be a strong effect. Nobody cares if you can do another Ozempic. If there’s another Ozempic, we can do a disease education and do market research and prove that it works.
Since a strong drug against aging has never been demonstrated in mammals. So I really believe that we’re kind of moving too fast here. First, it has to be demonstrated that aging is amenable first. Then I do believe that the world is wise enough to start thinking about this problem.
Joson Ng: Gotcha. Looking ahead, what kind of milestones do you see for Gero in the next few years? Like what are your next goals?
Dr. Peter Fedichev: Our goals, of course, we want to progress in our collaboration with Chugai. We hope to get more collaboration with them, maybe with somebody else. That’s very important because that establishes our business. We want to generate money from this collaboration, also experience, recognition, on a scale sufficient to progress our internal discovery program in aging.
We do believe that we can be the first company to demonstrate the strong effect in dogs. So we will be focusing on dogs quite heavily in the coming years. So, expect more collaborations. Hopefully, milestones in those collaborations.
And the company was made to make a strong drug. We established now collaborations with people who do dogs. We’re publishing works on the features of dogs aging. And now this theory has to play out. And we have to do a prospective demonstration in dogs, that what we believe actually affects aging on the scale that we want. I think that after that demonstration, things will look a lot better for our company. That’s the rocket coming on, but demonstration that is required.
Joson Ng: Looking forward to it, but more on a personal level. What keeps you motivated in this quest? Like why is aging for you such a deep-seated problem that you really want, are motivated to solve?
Dr. Peter Fedichev: Well, as you know that aging is one of these components of the transhumanism agenda.
Some people from transhumanism circles complain that I’m not true transhumanist. They call me a Latin transhumanist because I’m coming from physics. And in physics they teach you to solve big problems. This is part of the culture there. You’re not respected if you are doing something small. It has really to be universal and so on.
I hope that I identified aging as the right problem to be solved within my lifetime. This is probably one of the largest impact that I can create if I can solve it and I hope that I have the tools. Personally, I’m driven by ambition as many physicists. Physics is very competitive. So I hope that I can be a significant part of the developing solution in the end stage. With the skillset that I acquired, there is no bigger problem that is solvable.
And of course, on some personal level, I do believe that I’m already at the age when many people were dead. I mean, for the last thousands of years. I’m kind of already over living my typical life expectancy for a human being in the world. So I am, I’m living on borrowed time and I’m still having fun. And this means that so much fun, societal personal value was destroyed by aging over the lifetime of our civilization that I cannot imagine a worse enemy of humanity. It’s so unjust that humans are aging, especially now when we live in relative abundance, that I don’t believe that this problem cannot be solved and that it must be solved.
Joson Ng: Gotcha. Thank you so much, Peter. It was such a pleasure talking to you.
Outro
Joson Ng: Stay tuned for monthly podcasts with key stakeholders of the biotech ecosystem, including founders, investors, and policymakers. If you have suggestions for the podcast or who you’d like to hear from, feel free to send us an email in the episode description.
Subscribe to our newsletter, the Nucleate Artery on Substack. Stay engaged with Singapore’s biotech ecosystem. Join the Singapore Life Sciences community Slack channel, powered by Nucleate Singapore, where we are building an open community to enable conversations, the life science ecosystem of Singapore.
Longevity startup Gero AI has a mobile API for quantifying health changes | TechCrunch
Physics-Powered GenAI Biotech Gero Raises $6M to Find Root Causes of Aging and Age-Related Diseases
Drugs That Modulate Aging: The Promising yet Difficult Path Ahead - PMC
Humans Could Live up to 150 Years, New Research Suggests | Scientific American
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.