This Ex-Googler Replaced a Whole Drug R&D Team | Javier Tordable (4/4)

Intro - 00:00:06: Welcome to The Biotech Startups Podcast by Excedr. Join us as we speak with first-time founders, serial entrepreneurs, and experienced investors about the challenges and triumphs of running a biotech startup from pre-seed to IPO with your host, Jon Chee.

In our last episode, Javier shared the personal loss that pivoted him from cloud infrastructure into life sciences, how he self-taught his way into the field, and why he believes the most important problems in drug discovery are economic, not scientific. If you missed it, check out part three.

In part four, Javier unpacks why he left Google to solo found Pauling AI and how its agentic platform automates the full computational chemistry workflow from docking to ADMET analysis, collapsing months of work into days. He also walks through what it really means to build a company alone in the age of AI, filing your patents with LLMs, writing your own sales process, and running agents to build the very systems your customers depend on.

Jon Chee - 00:01:29: So, like, when you were at Google, when did you know it was time to start your own company?

Javier Tordable - 00:01:34: Yes. So, basically, a couple of years ago, maybe two and a half years ago, of course, I was kind of on the bleeding edge of AI. Right? You know, working very closely with a lot of folks, seeing what's happening. And at one point, I started thinking very deeply about what in my mind is one of the most transformational technologies, right, which is language models and agents based on language models that can automate tasks. Now today, you may have even seen that obvious. Right? Like, you know, I think everybody has seen, you know, their Codex or Claude or Gemini or Qwen or whatever is your favorite LLM and harness. You go and tell it, you know, read my document and write me some marketing lines that I can put in Google Ads. Right? A lot of us have done things like that and are familiar with the fact that these systems can orchestrate fairly complex sequences of actions. They can make decisions. They can analyze edge cases and so on.

Two and a half years ago, this was not obvious at all. So back then, there were a few cool demos. You could go and tell a language model, you know, give me this, and we'll give you that, or, like, process this text and summarize it, and it would do that. You could ask it to write code. Right? And it would write a test case or whatever. It will write a function, but you would never ever be able to tell it to write a large complex piece of code. If you went to the equivalent of, like, GPT-3 and then tell it, write me an iPhone app that keeps track of my to-do list, there was, like, zero chance, literally, that it would be able to finish that all the way into it. Now that's not the case. You can definitely try that. You can get your favorite harness and language model, and there are many of them around that are very good. I mean, I use all of them. Right? But you can pick whatever is your favorite. And in many cases, it will do that. Right? Like, literally, just let it run for a few hours, and it will complete that task.

So I started thinking very, very deeply about this, and what are the second and third-order consequences of that. And just like a lot of other tasks that are automated, you know, the first kind of consequence is you may need less people to do certain things. Right? Because a system that can perform these things automatically can do it faster and cheaper. Eventually, you can move upscale and think, well, once you have not just something that is faster and cheaper, but a repeatable, well-documented process, you can go and improve it. And you can improve it probably faster than you can improve a human. There's also, you know, one of those things that it's funny to talk about it. Right? But, like, when we talk about how long does it take to train a human to perform a job function, you know, that's twenty years. Right?

Jon Chee - 00:04:14: Yeah. Yeah.

Javier Tordable - 00:04:14: You know, two, three, four-year-olds. Right? And then you put them through school and then college and then this and then grad school and then you teach them on the job. It takes twenty years, right, to train one of these humans. We are in, like, what, year four of training language models. Right? And now they are the equivalent of a PhD in many fields. You know, year five, six, seven. Right? You know, these things are going to be superhuman. Right? There's no question about it.

So, you know, I started thinking very deeply about those things, and then I realized that I wanted to pursue some of these ideas. And it just made more sense to do it outside of Google than inside of Google. So I left, founded Pauling, which is named after Linus Pauling. I think we briefly touched on that, but he is surprisingly unknown, even though he's, like, one of the most celebrated American scientists in history. The only person that ever won two Nobel Prizes without sharing them with anybody. And he was behind a lot of, you know, really, really important scientific discoveries in chemistry. And maybe most importantly for me, the domain name was available. Right? So I ended up naming the company Pauling.ai and started working on these things. Right?

So it feels like a whole lifetime even though it's been basically two years. But left Google, found a few people that were interested in the space, raised a little bit of money. We spent a lot of time just building infrastructure. We have our own agentic system and so on and so forth. But, you know, with that kind of, you know, going through all the details, the key idea is basically to automate the work of a computational chemist. Right? As such, you know, it's one of the first few things that you would do when you're trying to come up with a novel therapy, right, or a new molecule for a novel target. You go and start running some of the simulations on the computer. You may have a structure for your target, and you want to say, well, let me see if I can find a small molecule inhibitor or binder for this thing, then that should have some therapeutic effect. You know, people are gonna spend six months, a year, a couple of years going through that process. You go through many different iterations. You make mistakes. You go back and forth, and you hire chemists that do all this stuff for you, computational chemist or medicinal chemist. You iterate multiple times, and eventually you find your lead candidate. You do your analysis in vitro, cells, and animal studies and so on and so forth, but you still spend a year, right, like, maybe going through all these simulations.

So I thought that makes no sense whatsoever. This is all work that is done inside a computer. I mean, outside of a computer, things take time. Right? If you're trying to give some drug to a mouse and then see does the drug cure the cancer in the mouse, it's going to take months, right, for that biological process to happen. But things that happen inside of a computer can be made much faster. So to me, it never made sense that you would spend six months running an in silico campaign. So I thought, let's see if we can apply first principles to this and do it faster and cheaper and better than people are doing it today, than anybody else in the world can do. And that was basically where we started. Right? Like, we started automating computational chemistry workflows. We use language models to automate those workflows. And the ultimate goal, of course, is to help scientists find new cures, new drugs for diseases.

Jon Chee - 00:06:55: Freaking rad. And, I guess, first off, did you have cofounders, or did you solo found this?

Javier Tordable - 00:06:59: It was mostly solo founding. Yeah. I mean, I have a few people that have been with me since the beginning, but in the sense of given the vision for the company, that's probably just me.

Jon Chee - 00:07:08: Okay. So hard mode. Hard mode.

Javier Tordable - 00:07:11: Yeah. I mean, I remember talking with a good friend of mine about this, and it was like, you chose to do a startup, which is already very hard. In arguably the most complicated field, which is, you know, biotech drug discovery. Right?

Jon Chee - 00:07:22: Yeah. Yeah.

Javier Tordable - 00:07:23: And to do it by yourself. And you're like, this is crazy, but here I am two years later.

Jon Chee - 00:07:28: Yeah. Yeah. Tell me a little bit about just, like, the early days. I mean, I realize it's, like, you know, two plus years, but what was, like, the early days of, like, company building for you? And, like, talk about the early teammates too that are, like, helping you build this up.

Javier Tordable - 00:07:40: Yeah. I mean so you probably made every mistake in the book. Right? I think in spite of the fact that I had experience in a couple of companies, right, and when I joined Google, it was still fairly young. I would say back then, I was still very much a big company kind of person. Right? So the big shift, of course, when you start a company or when you move to a company is that suddenly you don't have all the support systems then you would have it in big companies. You just have to do everything yourself. You have to set up your own payroll and review contracts and do everything.

Even today, like so we started working with customers, you know, a few months ago. We spent a good year and a half basically building infrastructure, working with design partners, but not really doing a proper sales process. Just this morning, I was working with one of my team members on setting up our sales enablement materials to share with people. Right? We have a few other folks that are going to be doing calls with customers who was like, yeah. Here's my deck. Here's my spreadsheet. Here's my frequently asked questions, concerns. Here's the kind of the sales process. I was, like, just writing my sales process. Right? Like, if it was up to me, I would be like, yeah. Let me just write code or, like, benchmark this model, right, to, like, start working on new modalities that we can simulate or whatever. But just have to do everything. Right? You do your servers, your legal, your HR. Right? Like, just have to do everything yourself.

Jon Chee - 00:08:53: Yeah. And I think the timing of this conversation couldn't be better. At least, I don't think it could be because I don't think there's a better time to solo found a company, frankly. Because, like, I virtually solo founded Excedr in 2011. Like, you're talking about sales enablement. The amount of tooling that's available now is insane. Okay. This is what I used to do to, like, get cold calls. I used to go and find, like, deeds and scrape deeds to find people's names and phone numbers. This is before ZoomInfo. Like, I was, like, pillaging I was going to Whois data like, basically, website databases trying to find, like, people and just, like, creating a prospecting list. Now you have, like I have Clay. I have, like, a 120 data sources, and, like, it does waterfalls, and then, like, I can do all these things. Yeah.

Javier Tordable - 00:09:41: Yeah. Or Apollo. Right? I think we use Apollo. I don't think we use Clay, but they're all the same, I think.

Jon Chee - 00:09:46: But you know what I'm saying. Right? It's kinda like this crazy thing.

Javier Tordable - 00:09:48: You select biotech founders in this geo area. Here's a thousand people. Right?

Jon Chee - 00:09:52: Here's the you're like, what the heck? And are you building, like, agentic systems for yourself? Like, not just for your customers, but, like, core infrastructure. Can you talk a little bit about that? I'm fascinated at how people are company building in this day and age because, like, you're founding a company when this is all, like like for me, I wouldn't say I'm, like, ripping things out because, like, I wasn't building on, like, SAP or anything of that sort. So I didn't have to rip things out. But how are you building your company with everything that's now available to you, and how are you thinking about it?

Javier Tordable - 00:10:22: Some things are definitely easier. Right? Most of our website is generated by an LLM. Right? I mean, it's like, you read it then. If you're very familiar with this, if you do this every day, you can kind of recognize a few things here and there. Right? I would say it's pretty good by now, but, like, yeah, I mean, like, we just go and point the LLM. So we have, for example, our list of the protocols that we run, right, computational chemistry protocols. I did not sit down and spend three days writing that. Right? I pointed the LLM to the code, and it said, here's the actual code of the pipeline. Go and generate a website where you explain what are the different steps and how they interact with each other. And then it generated the whole thing. Right? I mean, literally, that was, like, half an hour, right, maybe reviewing that. Whereas before, if you had asked an expert computational chemist, write down your docking protocol in enough detail that somebody could reproduce in a paper. You're like, yeah. I mean, that's a week of work. Right?

So a lot of these things are, of course, you know, marginal for use for reviewing contracts. Right? I mean, I don't know if this is one of those things that, you know, may end up biting later. Right? But, like, I do a lot of ChatGPT, Claude contracting. Right? I filed a patent entirely by myself, the whole thing. We did a non-provisional patent back in the day and then maybe about a year ago. So we did the provisional one first, then we did the non-provisional one, and that was literally just going back and forth. It was still quite a bit of work, right, to go through every single thing, review every single claim, the drawings, the formatting, all that kind of stuff. But that entire process, right, like, normally would be $15,000, $20,000. Right? I think software patents are a little bit easier, right, than bio patents. But we did an entire process entirely, you know, by myself with LLMs. So a lot of those things are becoming possible. Right? Like, of course, two years ago, that would have been completely impossible. Right? Like, you would need to know enough to at least not make mistakes. Maybe you would have saved a little bit of time, but it wouldn't be so special.

I mean, apart from that, like, of course, we're just seeing agents for coding. Right? Like, at the core, we build software. Right? So we use agents. Not going to talk much about that. I think that's kind of our part of our secret sauce. Right?

Jon Chee - 00:12:14: Yeah. No worries.

Javier Tordable - 00:12:14: We're very, very good at writing code efficiently, and we have set up a lot of infrastructure in our systems to be able to do that properly. So we not only run agents that do the computational chemistry for our customers. Right? But we run agents for building our own systems. And, you know, there are many other things. Right? I think in the marketing space, it's interesting, right, because it's become much easier, but it's become much easier for everybody. So the net result is that now it's actually harder to reach a customer than it was a few years ago. Everybody has a 100 emails from people. I don't know about everybody else, but, like, there's not a single day when I don't get an email from somebody offering to reach customers on my behalf on a commission level. Right? Literally, it's every single day. It's become so easy to reach everybody. Right? Then everybody is just, like, going through your favorite prospecting and an email tool. You know, 10,000 people. You send 10,000 emails, right, like, before you even had the chance to finish your morning coffee. Right?

So the end result is, like, we all have, like, literally hundreds of messages per day. Right? And I've been, personally, I've been very aggressive just marking everything as spam because there's just there's no way I can deal with all this stuff, and I'm just not interested. But I can imagine when we reach out to people and tell them, hey, we can save you $20,000 in your next computational chemistry program. Right? For some people, maybe that's just it goes into spam. Right? So I think email is definitely struggling as a way to reach customers. We use a variety of other things, like in person, right, or other ways to reach people. We haven't been doing a lot of cold calling. It's kind of hard to sell computational chemistry when you're cold calling somebody, but I think that's definitely one of the areas I wanted to explore a little bit. It's still hard to do that with an agent. Right? And you can definitely point an LLM.

Jon Chee - 00:13:49: I'm not comfortable yet with the voice agents. Well, I did see something creepy on Twitter. I was like, oh, shit. Like, that is creepy good, but I'm, like, not there yet. Definitely, on the written word, it's getting a lot better, but I have a similar experience too. Like, I think but it's kinda like this thing where, like, you're just like, alright. There's, like, phone calls and meeting people in person. Again, maybe voice AI will become so good that, like, it can do

Javier Tordable - 00:14:15: It's just a matter of time. Right? The last 5% there, right, like, you're getting the latency. Right? Normal people would tend to interrupt a little bit. Right? Like, go back and forth. Right? So that is still going to be, I think, a little bit harder. Right? Like, language models still take a hundred milliseconds to hundred milliseconds or whatever. Right? Even, like, highly optimized systems takes a while to load your entire context of memory and so on. So it's hard to get into that fifty millisecond kind of latency of, like, real human interaction. And I think it's probably going to be for a few years. So you can definitely generate a 100% realistic sounding voice. You can generate an audiobook using a voice model indistinguishable from a human. But having a live conversation is a little bit harder.

Jon Chee - 00:14:54: That's been a huge unlock for me. I'm a slow reader, but I'm a really quick listener. I'm one of those, like, sociopaths that, like, will do, like, two, two and a half, sometimes three, and just go.

Javier Tordable - 00:15:07: I find it very obnoxious. I can't do that. Really? I know. It annoys me so much. More than 1.5x, I just find it very, very obnoxious.

Jon Chee - 00:15:15: That's so funny. Like, it's my preferred method of ingesting. Like, at one point in time, I wanted to be a lawyer, and I was like, how I read so slow. Like like, this is, like, impossible, but I'm, like, a great like, I'm great at this auditory, like, but, anyways, now it's been an interesting thing is kind of hearing the kind of, like, 11 Labs of the world doing that. And I was, like, oh, this is crazy. Like, this is actually crazy. It's interesting too as you're building the company is that you go back to meeting people in person and stuff. It's just like the more digital we get, the value of, like, analog, it's just like, oh, yeah. It's like

Javier Tordable - 00:15:52: You appreciate things that are done by hand. Right? Honestly, like, this is also maybe one of those second-order effects, right, that is hard to distinguish. But in the age of industrialization, right, like, people still appreciated handmade things. Right? Like, you get a handmade leather bag that is more expensive than a leather bag that is made by a machine, even though it is, objectively speaking, worse. Right? The stitches are going to be worse. The finishing is going to be worse, but it is handmade. Right? And you pay extra for it. I mean, I definitely do not pay extra for it, but some people do. Right?

Jon Chee - 00:16:35: Yeah.

Javier Tordable - 00:16:35: So I think there's going to be a premium in that. Right? Like, being able to do in-person interactions, being good at selling in person is definitely going to become harder, and it's going to become more and more important over time.

Jon Chee - 00:16:35: 100%. I think too is because, like, eventually, you're just not gonna know if it's a person on the other side. You won't. Like, you know, if the voice gets so good, there's I used to laugh at, like, you got got by a deep fake. Like, I'm getting got now.

Javier Tordable - 00:16:51: Oh, yeah. Yeah. No. No. It is. It is. So I think voice is literally solved. Right? Then conversation, maybe we'll need a couple of years, but voice by itself is solved. Video is very, very close to being solved. Like, there are some videos I mean, don't know. Like, if you open Instagram or whatever, like, I would say AI accounts are, like, I don't know, like half for what I get in my Instagram feed. Many of them are really good. I sometimes would go and, like, literally just open an account, and I was like, oh, is this AI or is this real? And I would have to scroll all the way until it's like, oh, here is an artifact in the AI generation. Right? That'll be coming in really, really good. I stopped correcting every single spelling mistake in my emails. Right? Like, back in the day, they were like, oh, yeah. You need to correct all your spelling mistakes. And I'm like, well, like, if you have spelling mistakes, obviously, you're not an AI. So I'm like, yeah. Screw it. Right? I'm just not going to click on the red line. Whatever. I send the message just like that so you know that I'm a human being sending you this email.

Jon Chee - 00:17:43: By the way, I'm doing the same thing. Like, I'm, like, not even gonna capitalize it. I'm almost sending my emails like I would text my friends. Just, like, rapid fire texting, like, just, like, wrap like, almost, like, AIM. Just, like, like, let it rip. Because at this point, like, you know, you can just, like, click the polish, and then it's just, like, boom. It's like, oh, perfect email. Of course, robot. Like but now I'm in the same exact way. I'll just let it fly. I don't care. Like, we're just gonna, like it's actually valuable in this sense.

Javier Tordable - 00:18:13: And I think it is going to be more valuable, honestly. Right? It's like human-written communication is going to be at a premium. It's going to be more effective than agentic communication. I mean, we'll see. Right? All these are kind of hypothesis. Right? But, you know, if you look at, you know, what happened in other places, right, it's just going to be a little bit of a premium on, like, doing things by hand the old-fashioned way.

Jon Chee - 00:18:35: For sure. And I think too is just kind of doing it like the artisan way. I always like going to Japan and just seeing how, like, they handcraft, like, a lot of things. And I think we're gonna just, like, see is just, like yeah. Like, to use food as an example, like, I wanna, like there's this person who makes one dish by hand. That's all they do. I will pay a premium for that. And they dedicated their life to this thing, and it's kind of like that artisan craftsmanship that I think with everything, the proliferation of, like, being able to just so quickly just, like, industrialize almost, like, everything. Exactly what you said. It's just like and then, you know, you get back into, again, meeting people in person, getting a meal, getting coffee, whatever it may be, having, like, a real authentic, like, real relationship with someone. And, ultimately, at the end of the day too, even with all these, like, scientific endeavors, at the end of the day, it's still, like, human to human. Like, people do business with people, like, at the end of the day. And I always think too is, like, is this you know, there's, like, always, like, people will be like, yeah. Well, you know, the clinkers are gonna take over everything. It's like, no. Like, at the end of the day, there's someone that needs to be responsible, like, at the end of this. Like, some human's gonna have to sign off on this and, like, be like, I bear the responsibility of this output.

Javier Tordable - 00:19:47: Yeah. Yeah. In many areas, for sure.

Jon Chee - 00:19:49: Yeah. Yeah. In medicine, for sure. Like, I don't think the regulator is gonna be like, yeah. Fuck it.

Javier Tordable - 00:19:53: Here's the funny thing. Right? I think in The US, the AMA is always going to lobby for a human doctor to review every single prescription, to approve every single treatment. Right? They're a very powerful lobby. They're like, no. It's going to go and fight them. What you are going to see is that in other places, you will get automated systems that are just maybe just as good. Right? So, like, maybe in China or in Japan, you will have an automated system that goes, does an interview in a fully automated way, does whatever blood test you need to do, and then will give you your prescription for your antibiotics. And it will cost $5 for the appointment.

So what you will have is that the technology is available worldwide, but in some countries, your health care costs actually start to go down because they have the right incentives. Right? Whereas in The US, we're gonna say, nope. You go to the doctor and you want to get your antibiotic prescription, and that doctor is going to charge $750 for your appointment, and your insurance company is going to pay 215, right, for your twelve-minute appointment. And then you go to the pharmacy, and it's going to cost $135 with insurance out of $800 to get your antibiotic. Whereas in Japan, it would be like, yeah. Here's your receipt. 200 yen. Here's the exact same level quality of care. Right? But it's going to be much, much, much cheaper. It is sad, but, again, like, I think we live in a world where the incentives, right, within the care space are just so misaligned. That's something like that. I mean, I would say probably, like, 80% of something like that actually happening.

Jon Chee - 00:21:22: Yeah. I think I mean, we're kind of, like, as an industry grappling with it. All you can see in, like, Endpoints and STAT, they're just like, China's overtaking us. It's just organized differently. Like, the barrier to getting building things, like, you have a very engineered focused it's an engineering culture, whereas, like, we have a very legal culture. It's a very litigious and, like, regulatory kind of, like, framework. Each has their pros and cons. But, you know, what we're seeing now is, like, I mean, friends who go to China come back, they're like, holy shit. Like, talk about the preclinical, like, research that they're doing out there. They're just, like, cranking, like, the robotics and, like it's just like these, like, just like these loops. Like, it was just like lab loops that they just have going and just being able to get clinical trials done without absolutely obliterating. Like, it's just like quick. I've always, like, thought about life science. If we can somehow bring that, we need more of that, like, software feedback loops that you can get. It's just really hard, like, in the the framework that we operate in in The United States. But then you just, like, see how China's doing it. They're kind of, like, mimicking feedback loop, like, almost like software feedback loops.

Javier Tordable - 00:22:34: I mean, I don't think the park is that hard. Right? So, like, you can build robots here assuming you can build them there. I think the competitive advantage is a little bit more settled on that. Right? So, you know, people will work longer hours, right, and they will do multiple shifts. Right? You know, for example. Right? So if you go and you have a research lab and some people work in the morning and some people work at night, right, and it's the exact same amount of work, right, but you do two shifts a day, you're moving twice as fast. Right? Now here in The US, somebody from your state, your department of revenue or whatever is going to go and complain, you cannot have people work for twelve hours a day or whatever. Right? You will find all sorts of quality, and people don't wanna do it either. Right? Like, now what? He wants to wake up at five in the morning and then work for eight hours, and then somebody else starts at 4 PM, right, and then works until one in the morning. We don't wanna do that. Right? As founders, we probably do. But, like, most people don't.

Jon Chee - 00:23:25: Most people don't. Most people don't. Like, we're kind of, like, operating on, like, crazy time, like, demon hours. The one thing I will say, though, is, like, we can create these, like, automatic, like, lab loops, like, preclinical stuff, but, like, they actually get it into the clinic. And then you see all these pharma just, like, snapping up, like, anything just, like, for like, the boom. First clinic. Oh, good readout. Okay. And it that part, we don't do that. Like, we just I can't get into the clinic and actually dose and then actually get a readout as quickly as that, and I'm like, we need to do something.

Javier Tordable - 00:23:57: It's probably not even legal. Right? Like, in The US, you need some protocols that you need to use to set up your trials and you set up your controls, how you give medicine to people and so on and so forth. Yeah. It's probably not even legal. No. No. A 100%. Right? My point is, like, you know, some parts are probably easier to replicate. Right? The robotics part? For sure. Like, you can build it here the same way that you would build over there. You know, other parts are definitely more subtle. Right? The schedules, the ability to, you know, operate within a legal framework where you can, you know, maybe take more risks or do things in a more automated way. Right? You know, I think a lot of those things are very significant competitive advantages. Right? And now you're seeing, like, what is it, 40% of all new molecules are actually coming from China. To be fair, a lot of those molecules are also not truly novel. Right? They are patent busting kind of molecules. Right?

Jon Chee - 00:24:40: Yeah. Yeah. Yeah. It's like kind of like follow-on.

Javier Tordable - 00:24:42: Yeah. Exactly. Right? But, again, it comes down to an economics question. Right? You know? So if you go and say, you may be a Cambridge graduate, incredibly smart person. You did your postdoc at the Sanger Institute. You discovered truly novel biology, and then you go and publish your paper, and then you patent your molecule, and then you're like, this is great. I'm going to go raise $50,000,000 and do a phase one, and so this is x. And then before the time you get the chance to finish your race, somebody from China has already done that and given it to a patient in the clinic. Right?

So now big pharma comes in and says, well, why would I give you this amount of money for you to set up your office in Cambridge and do all the stuff? I mean, you may be the smartest person in the world. Right? But why would I give you this money where I can just literally buy the exact same thing from this company in China that took your invention, and then three months later, it's giving it to a patient in the clinic. Right? It makes no sense. Right? So as a pharma company, which is a rational economic buyer, right, you go and say, I'm just going to put my money where I have the highest chance of making a return. Right? And, again, like, the people that are coming up with this, like, they may be smarter. They may be working harder, but the economics just don't work. Right? So, unfortunately, a lot of innovation is just going to be entirely disrupted in the West because there's no way to pay for it.

Jon Chee - 00:25:58: Yep. Yep. And this is kinda what I was alluding to before where I have this, like, urge to, like, fuck it burn it down, and, like, we need to rebuild the framework because, like, clearly, it's not working. And it's kind of this thing where we're seeing it happen before our eyes. And I wonder again, this is, like, the optimist in me that, like like, okay. Maybe we can get, like, statutory changes and regulatory changes.

Javier Tordable - 00:26:24: Good luck with that. You have all my support. Thank you very much. I have one for you.

Jon Chee - 00:26:27: Exactly. I'm just like but then when I say that, it was like, do I really believe it? Like, I don't know. But you're right. It's an economic problem. And also the pro and con, when it comes down and saying, like, this is an initiative for the nation, we're gonna make it such that this is a thing. Of course, it's gonna work. Like but we just don't have that here. Like, we're just, like, set up differently.

Javier Tordable - 00:26:51: I mean, unfortunately, I don't think it's going to happen. But, again, like, I'm naturally a pessimistic person. Right? Like, I tend to think in terms of, you know, all the things that could go wrong and so on and so forth. And, of course, as an entrepreneur, you cannot do that. Right? Like, you always have to be optimistic because I judge you would not survive. When I think the 36,000 feet view, right, we are in this strange fall of the Roman Empire kind of phase, right, where the decline has already started. Right? It may go on for decades. Right? But I think, you know, unfortunately, things are going to get worse before they get better.

For many of us, right, that we're immersed in the system, we have salaries or advantages or we live inside of a bubble with a lot of these things don't affect us, you know, it may be fine. You know, many of us are mobile. We can go to different parts of the world where, like, we speak languages, we have capital, and so on and so forth. Right? So I think many people are just not going to be affected. The really sad part is that for a lot of people is it's a problem. Right? And without mincing words, right, like the fact that things are very expensive and it takes so long to get treatments and healthcare is so expensive means that people will unnecessarily die. In The United States and a lot of western countries, there's no question around it. Right? So I feel for that. Right? But at the same time, it's it is not a science problem. It is purely a political and economical problem.

Jon Chee - 00:28:07: Yep. Absolutely. So, like, to bring it back to, like, Pauling, we so, like, you laid out how you're, like, building this, like, framework for solo founding this thing. Talk a little bit about the broader development journey, And then and you mentioned that you're, you know, you're not paying customers. Like, talk about a little bit about as you got into go to market. Can you talk a little bit about that? Like, what type of organizations find the most resonance with your product and use cases and stuff like that?

Javier Tordable - 00:28:33: Yeah. So as we were talking a little bit before, typically, when somebody is beginning a new discovery program or they're starting a program that is parallel to something that may already have. Right? The first phase is to go through literature. You're trying to understand a target or trying to figure out how to pick a specific target for a disease. And then the next step is to hit the target. Right? Typically, you would have some modality. You would evaluate what modality makes sense for your target and for your indication, and then you want to, you know, inhibit or modulate, you know, somehow that target. And one of the ways to do that is using rational drug design and using computer simulations. So that process, it costs a certain amount of money, and it takes a certain amount of time. So what we're trying to do is make it faster and cheaper. So it's a pure economics play, so to speak. Right? We're not trying to come up with a truly novel method, right, that kind of replaces previous approach. We're just trying to make it much more efficient.

So there's a little bit of a difference with a lot of the companies that are popping up in this space. So if you are DeepMind, you're building AlphaFold, you're not trying to build a better molecular dynamic simulation platform to come up with a protein structure. Right? Like, you're trying to sidestep all the stuff entirely. And, of course, you know, there was a massive achievement, and there are probably many more of those things that are still out there. But there's also the more mundane problem of even when you know what to do, you still spend a lot of time and money setting that up. So a lot of the folks that that we're working with are small, medium-sized companies that have some of these programs. Sometimes it's something purely de novo where they have a specific indication area where they're interested in kind of entering a market. Sometimes they want to replicate what a competitor is doing. So maybe some company publishes a patent and they say, okay. This is interesting. This is an aerospace space. How can we come up with novel chemical matter, a novel molecule that is different enough for whatever is in the patent that it can be patented separately, but at the same time has similar mechanism of action. So a lot of those things can be done through computer simulation.

So from our perspective, there are certain things that we already know how to do. We've built agents that do them, and then we're very efficient. We're very quick at doing that. So we can do regular screens. We can do counter screening, just basically when you try to find molecules that hit a specific target but do not hit another target. So sometimes you do it for different pockets within one target. So let's say you have an enzyme that may have an orthostatic pocket where, you know, some cofactor binds, but then maybe another pocket and you want to try to find something that binds to the other pocket but not to the main pocket in order to reduce toxicity or whatnot. Sometimes it is a different target. In the human body, you have families of proteins as opposed to individual proteins typically. Right? So you would have many different proteins that are similar in different ways, but have slightly different function. And for a lot of diseases, you would be interested in preventing or kind of disabling one of the proteins in the same family, but without disabling all the other ones. So one of the common things that you would do with screen is say, can I find a small molecule that binds to one protein in the same family but does not bind to any of the proteins in that same family? Now that is, of course, more complicated, right, because these things tend to be very similar. So you kind of try to play with what are the small differences where you kind of, you know, leverage that. So we've been doing a few of those. You know, sometimes you're interested in what's called polypharmacology, which is essentially a molecule that would hit multiple targets at the same time because that has some benefits in your specific disease. Right?

So we do a variety of different types of screen, and then we typically use physics-based tools as opposed to machine learning methods. You know, we combine a lot of different methods, but the basis of it is typically physics-based. So we do traditional docking and molecular dynamics and then a variety of analysis stemming from those molecular dynamic simulations. Now we started with small molecules because a lot of the tooling is more mature, which is better understood. We can justify that our protocols are good enough, but we're going to be doing all the modalities. You know? We'll do peptides. We'll do antibody design, etcetera, etcetera, etcetera.

One of the nice benefits of an unbiased AI system is that it can look objectively at a variety of different ways to perturb a cell to have a therapeutic effect. Now in real life, humans, they have experience that is tailored to specific area. Right? So within pharma, you would have a small molecule person versus a biologics person versus a cell therapy person and so on and so forth. And it's very hard to have an overarching point of view. Right? So if you have a novel disease or you're trying to come up with something that is truly different, at a big company, you would have a team of, you know, 20 people. You put them all together in a room. Right? And then something would come out of it. That's an expensive meeting. Right? Everybody can afford that. But an AI system can go and say, okay. Let me look at the cross product of different targets that may have in a specific pathway that is involved in a disease times modalities that I have, you know, for that. Right? So you could look and say, okay. There's this GPCR. It's a protein in the surface of the cell that acts as a receptor, and then something binds to it. And then it's kind of dissociates from another part of that protein, and that binds to another enzyme, and that complex binds to something else and then catalyzes some reaction happening and then something else causes the disease.

So an AI system can go and look at it and say, okay. I have the GPCR. I have the first part of it. I have the complex, and then I have maybe some transcription factor somewhere. Right? And then I have small molecules, peptides, antibodies, siRNAs, and so on and so forth. And they can go and say, let me just go through the entire cross product. I go through all 50 options and then just look at each one one by one systematically. Right? So an AI system can do things like that that are essentially impossible for humans today to do definitely to do on a cost-effective basis. So the idea is to build a platform that can do all these different things so that when somebody comes in and says, you know what? Here's my antigen that I just found out. Here's some virus. I only have annotations for some of these proteins. I don't really know how it works. I just have the sequence. Well, like, how would we come up with a novel antiviral for it? And then the system would go through every single protein, every single function, every interaction, try to do some selection of the right target, and then select the right modality. Right? You know, it could be an antibody. Like, maybe there's some surface protein in the of that virus that is very distinctive. So you would say, yeah, an antibody would perfectly work here. Right? So you would call and design an antibody to bind to it, you know, give you some sequence or, like, directly send it to one of these cloud labs and say, synthesize this antibody for me, express it in this different way. I need this amount of quantity with this level of purity or whatever. Ship it somewhere else, have it tested against that virus, and then go and iterate. Right? And do this whole process in a fully autonomous way.

Now, again, people do this today. Right? And to do the thing that I described, you could spend a year easily, right, doing that. So what we want is to be able to do that in a matter of days. Right? We want to go 10 times or a 100 times faster. Right? So, yeah, I mean, we've been basically building a lot of infrastructure for doing that. So far, we are, of course, a little bit more modest. Right? Like, we've been doing small molecules for a variety of different targets. So we've done, you know, a few ion channels, you know, GPCRs, traditional enzymes, so on and so forth. We can do a variety of different simulations, you know, with and without membrane, but we keep adding more and more.

Jon Chee - 00:35:42: Sick.

Javier Tordable - 00:35:43: We're still going in that journey. I have to say, like, we're still a fairly young company. Right? So there are a few things that we can do and others that we're still going through. But if things work out well, hopefully, we can raise money, get more customers, and so on and so forth. We're like, we'll get to a point where we can truly provide a platform that will come up with new cures just purely autonomously.

Jon Chee - 00:36:02: Very cool. And something that stood out to me is talking about how the AI system can, like, objectively just, like, assess. That stood out to me and also just when you're talking about, like, being at a large organization like Google where you have to, like, schmooze a little bit. It's like, what ideas comes to the top? And, you know, and then biases kind of start to play in when there's people involved. It's always like for me, I was always like I always would battle in my lab with the comp bio people. I'm like, no. We need to do all this in the wet lab. And, like, all the comp bio people are like, no. We'll just do it on the computer. And there's, like, kind of these biases. Yeah.

Javier Tordable - 00:36:37: Well, I mean, some things that you have to do on the computer and other things. And some things that you have to do in real life, by the way. Right? So it's still that is it to do in real life.

Jon Chee - 00:36:44: For sure. And with something even on my side and Excedr, what we're trying to do is, like, we just have, like, so much data. We have, like, conversation data. We have, like, all the HubSpot, like, email data. Like, we have data from, like, events, and we, like, feed it in. And then we just have basically, what we're doing now is, like, we're having just, like, AI, like, kicks like, we used to have, like, these, like, calls where we're all kind of kicking off the week. Each department just, like, says their thing. But now we, like, flipped it where we're having now, like, AI kind of, like, dig into the data and then do the presentation to start the week, and then we chime in. And, like, it's an interesting thing because, like, when a sales team is, like, forecasting things, is it as good as you say it is, or are you just, like, think it is and AI is very much a great just, like, cutting through that? You're just, like, no. This is at risk, actually. Like and it kinda gives this level of, like, clarity, like, where like, it's a good it's objective. It has no feelings. It's just like, here it is. This is what it is.

Javier Tordable - 00:37:46: Doesn't have a quota. Doesn't care about getting fired at the end of the quarter. Right?

Jon Chee - 00:37:49: Exactly. Has zero does not care. And then so and it's kind of really changed the way we communicate internally because now it's just like, the robot literally just like we know where it's sourcing it from. Like, we can see this, and this rubric is now, like, very clear. And that's just, like, sales as an example. And then when we do, like, marketing analytics and stuff as well, we're just like, okay. Like, it does way better attribution than we can do. Like like, way better. And it's just kind of interesting seeing the nature of work change with all of this. I guess maybe I'll let you do the setting of the table. Like, you know, you said it takes, like, a year. What is the status quo? If Pauling wasn't around, what would you typically be doing in that year time?

Javier Tordable - 00:38:28: Yeah. I mean so it normally comes down to a couple of different options. Right? So, normally, sometimes people just don't believe in computational chemistry. Right? So they want to do everything experimentally. That costs time and money. Right? But in some cases, you know, that's just the best way to do it. I would say in many cases, computational methods can help and do help.

Now for people that do computational methods, typically, it's one of two things. They either have an in-house team, so may have a person, two people, three people if it's a small company. If it's a large company, they may have hundreds of people to do this, but they have some in-house teams that are typically overworked and other staff as usual, right, where people have some particular level of expertise. And in some ways, they're trying to use some of this technology, usually not as effectively as they could. Right? The reason why software engineering is getting disrupted so much and the reason why these models are so good at writing software is because the models were built by people who write software. Right? So it's a problem that is very, very well understood, and it's a tool that is built by the same people that suffer from that specific problem. Now that's not the case in computational chemistry. You don't have computational chemists building LLMs specifically for computational chemistry. So people are trying to use some of these tools, but I think for the most part, they are not as efficient as it could be.

And the alternative is to outsource that. This is an area where it's very common to use a CRO or a chemistry research organization. You know, of course, there are many CROs that do experiments of various types in cell lines or organoids or animal studies, etcetera, etcetera, etcetera. As a matter of fact, I think most biotechs don't do their own animal studies. Right? Like, they just hire a CRO to do it, who is an expert in sourcing and taking care of those animals and running the experiments and so on. So the same thing happens for computational chemistry. So somebody would hire a CRO and says, you know, here's my target. Here's the kind of constraints, you know, in the chemical space that I want to get. Here's what I want from my molecule. The CRO will go and tell them, yeah, this is going to take six weeks, and we're going to charge you $20,000, and then we'll give you a list of your top 100 compounds.

Now and an in-house team, you know, may say, well, it's gonna take me two months to do the same thing. Right? But, of course, you only pay a salary. Right? You don't pay extra. Right? You may have to set up your own, like, GPU cluster and buy them workstations and so on and so forth. And depending where you are in the world, the salary of that person may be very high too. Right? So if you're in San Francisco, it may pay a quarter million dollars, you know, fully loaded cost to that person to do simulations for you.

So our value proposition where we come in is somebody would tell us about what they want to do, and then we try to be drastically faster than any of those two options. So we typically would do projects within, you know, three, four, five days all the way from getting a target to getting a list of hits. Of course, that is just the first phase. Right? So after somebody, you know, test those molecules, they may come back and say, hey. This works. This doesn't work. Let's run through an optimization process. Let's focus more on that. Continue searching in one way or another. Right? It's almost never one and done. Right? But we will do that radically faster. And in general, we can do it also cheaper. Some people do have infrastructure in house that they have already paid for. Right? So in that case, it's kind of harder to compete with the cost of electricity, but we can definitely do it cheaper than almost any other CRO out there.

And I would say, again, as I mentioned, right now, there are certain things that we know how to do, we've done for some customers we're very good at. You know, we have fully automated. But benefit is as we keep doing more and more projects, building more and more infrastructure, the repertoire of the kinds of simulations and the kind of analysis that we can do keeps expanding. So if you put yourself in a position, you have an in-house team, you know, you hired one person, that person is a small molecule person, and then you wanna say, well, we want to expand into biologics because for this specific or the other way around, actually, that's even more common. Right? So you have an antibody for a specific cytokine that you use for an autoimmune disease, and you must say, well, this is great, but you have to inject it every day or every week, and it becomes a pain. So if you had a small molecule that has the same effect, that would be great. Well, if you had a quarter million dollar antibody person and you tell them, next week, we're going to do small molecules, they'll be like, what am I going to do with that information? Right? You can go and hire somebody else or you can send this person back for training, spend another year. Right? And then they will come back and maybe they can help you. So we try to essentially overcome that limitation. Right? Somebody can send and says, yeah. I have the specific target. Find me a large 700 molecular weight molecule that binds to it, unlikely to create immunogenicity or whatever. Right? And we'll go and say, okay. Well, let's just put our agents to work on that. And then, hopefully, we can do that faster and cheaper than they can.

Jon Chee - 00:42:53: Sick. In my head, I can just, like, envision the path dependency that kind of, like right? You're just, like, you're from biologics. Like, oh, shit. Like, we cannot go from here. Like, we can't make that turn. It was a one-way door.

Javier Tordable - 00:43:05: Well, but now it's all the rage. Right? Like, even with GLP-1s, you know, it's all, like, oral GLP-1s and, like, for glypromone. Right? Like, you know, the new Lilly small molecule, GLP-1 inhibitor. Right? I mean, for I think for a lot of people. Right? Again, back to another economics question. Right? Like, you know, biologics are expensive. Right? If you can go from a biologic to a small molecule that has the same purpose, not only is it more convenient for people, it makes it more accessible for people. It's also cheaper. Right?

Jon Chee - 00:43:29: Yep. Yep. Spot on. And, like, that optionality is, like, for real. It's got, like, that value proposition you're which is like, yeah. Like, we'll give this a spin. Let's see what we can do on the small molecule side. That's really, really cool. And, like, I've was talking to some other founders that it's just like, what a time to be building. Like, it's freaking cool. Like, I have been getting less sleep because I'm just, like, tinkering with, like, and building things way more. And I'm so tired, but I'm having so much fun. Like, it's just freaking awesome? And hearing what you're building is sweet. Like and as you're, like, looking forward, let's say, one year, two years, what's in store for you guys? Yeah.

Javier Tordable - 00:44:12: I don't think it's shocking. Right? I mean, it is exciting, but it's not as always shocking. Right? So expanding across all dimensions. Right? So now we can do one type of modality. We want to do more type of modalities. Right? We want to do products. We want to do molecular tools. We want to do more exotic kind of mechanisms for small molecules, peptides, antibodies, etcetera. We want to expand across the type of simulations that we do. Right? So we do enzymes. We do transparent proteins. Right? It would be great to be able to simulate RNA DNA transcription factor complexes. Right? And small molecules that modulate transcription factor. That's a class of therapeutic molecule that has barely been explored. Right? Like, you know, there's a little bit of literature, but it's very rare. There are no good computational methods for that. So we'd like to do that. You know, of course, becoming more mature as a business, right, we'll have, you know, more established processes and so on and so forth. And I think, you know, as I mentioned, the ultimate goal is, right now, we're working with other companies offering services acting a little bit like as a CRO in the building technology that we charge for. But, of course, the ultimate goal is to work on assets, to work around therapeutics. Right?

So there are many different things that I'm interested in that I'd love to be able to work. One of the things maybe this kind of, like, going a little bit down the weeds. Right? But, I mean, one of the things that I think is super, super interesting and one of our academic collaborators is in this space is the impact of chronic infections on neurodegenerative diseases. Some of these are better known and understood than others, but I think this is one of the areas that is going to be a massive potential longevity, expanding intervention in the future. So some of these things, like the association between Epstein-Barr virus and multiple sclerosis, Right? It's very well known. Epstein-Barr virus that seems to be present in ninety, ninety-five percent of people that actually suffer from multiple sclerosis that contributes to the demyelination of neurons. So the rationale again, this is not novel. Many people have thought about this for the years. But the point would be, well, if you could come up with an antiviral for this virus or somehow prevent the virus from working, you could prevent the neurodegenerative disease. Right? And I think that intuitively makes a lot of sense.

But the same thing, there are many other examples of similar infectious diseases. There was a shocking study in my mind that came from Taiwan originally around people that were vaccinated for herpes virus. And in that study, because of the way that it was done and because of the kind of constraints around which they started vaccinating people, they were able to prove causal relationship between protection against herpes virus and reduction in the probability that there would be Alzheimer's to a point where if you had been exposed to herpes virus at some point in your life, you were essentially thirty percent more likely to develop Alzheimer's, all those things being equal. So what suggests is that if you could come up with an antiviral and this is a little bit more complicated because herpes is very complex virus. You know, it doesn't just stay in blood, penetrates neurons. It would form certain conformations that become latent for many years, and it would reactivate over time. And it is very, very hard to actually reach with most modalities. Creates a little bit of a cluster, right, that protects it and so on and so forth. So it's much harder than it seems from a technology perspective. But if you could somehow prevent that virus from working, you could have a meaningful effect preventing neurodegenerative disease. And in my mind, once this is a little bit understood and the technology matures, I think some of these things would be ideal use cases for technology like ours that can exhaustively find and explore novel targets. Right?

I really like the example of a virus because in a virus, you have a limited number of proteins. Right? You may have 100 to 100 proteins, and you can systematically go through every single one of those and try to identify which one would be easier or harder to drug. You can go through essentially every single protein-protein interaction. You can disrupt protein interactions. There are much smaller organisms where you can reason through them exhaustively in a way that you probably would not be able to do with more complex organisms. And if you could completely save and without off-target toxicity effects, if you could somehow inhibit this, then you would have a hopefully positive consequence, not just for that possibly acute infection, but for the long term, you know, effects of that infection, you know, the inflammation and many other, you know, age-related effects. As a matter of fact but, I mean, you probably know this better than I do. Right? Like, one of the reasons why older people die, right, is because the immune system eventually becomes busy with all sorts of things instead of protecting against infections. So older people may die from the same cold that you and I go through, and we have no issue whatsoever. Right? But if you're 95 years old and you get that cold, maybe fatal for you. Right? So being able to fight infectious disease, I think, is still very, very, very important.

You know, in my mind I mean, this is kind of high speculation. Right? But I think one of the interesting things that came up from the GLP-1 drugs is that when you reduce caloric intake, you get a whole bunch of side effect. Right? So it's not just for diabetes and obesity. Right? But it kind of reduces the cravings. It improves psychological well-being. Right? Like, you have all sorts of really amazing side effects. I think if we were able to reduce essentially infectious disease and reduce inflammation, that would probably have a very substantial effect in many other things across our bodies, including longevity.

Jon Chee - 00:49:23: Absolutely. And it's really cool to hear to how, like, one, the ultimate goal to actually start developing assets, that'll be super rad. That's a future that I'll be rooting for you from the cheap seats. Like, just like, I'll be up in this case rooting for you. That's really cool. And I guess a question for you too. Like, you know, you talk about, like, I believe you said a research partnership with academia. Are you guys looking for, like, academic research, like, partnerships? Are you primarily focused on industry? You mentioned some of your early customers are kind of, like, the small to medium size. Do you also look at the large folks, All of the above?

Javier Tordable - 00:49:53: I mean, of course, like, if anybody is listening to this and they want to discuss, you know, happy to do that. But I would say right now, there are a couple of things that are more interesting and a few things that are a bit less interesting. So academic partners, maybe a little bit less interesting. Right? Like, a lot of those are unpaid or, like, people don't necessarily have a lot of money. Right? So we have or at least I personally have a bunch of different indications that I would like to work on. And it's not complicated to find people that are working on that same space and then try to strike a partnership. Right? So we've done a few of those already. Larger pharma tends to be slower and harder to work with. Right? We haven't put as much effort into it. At some point, you know, a year from now, rather, we'll have the whole sales team. Hopefully, we'll do all the stuff. We'll go to every conference and so on and so forth.

So I think the sweet spot for us is small, medium-sized biotech funded, right, but not necessarily tons of resources. Right? And where the value proposition of saving money and especially being able to operate faster, right, to be able to compete with some of these other companies that are coming from China elsewhere becomes much, much more interesting. And, of course, investors, like any other company, right, you know, Now, you know, raising money. If you're a VC, you know, you're investing in the AI for sciences space, hit me up. I'd love to discuss. We're trying to build the fastest, lowest cost computational chemistry here in the world. Right? And if we manage to get the resources, we will do that. Right? I have no doubt about that. But the ultimate goal is not to create a CRO. The ultimate goal is to cure diseases. Right? So this is a tool that we use for that. Right? It's not the it's not the ultimate purpose. Very cool.

Jon Chee - 00:51:23: Very cool. Well, Javier, this has been super fun and wide ranging as well. I always think about these conversations as, like, a road trip almost, and then, like, taking these detours into the forest, and then we find our way back, and then we take another detour and then find our way back. So this has been really, really fun, and I've learned a lot. And in traditional closing fashion, got two questions for you. First question is, would you like to give any shout outs to anyone who supported you along the way?

Javier Tordable - 00:51:47: Yeah. I mean, there are many people. I think, you know, some of them I already mentioned. Right? Some of my managers at Google, you know, Asaf and Patrick and Will and others. The folks that invested in my, and Gandhi and a few other people. Right? You know? So lots of thanks. And, of course, you know, many other people that have you know? I think in our world, nobody does anything completely alone. Right? But there's a lot of people to mention.

Jon Chee - 00:52:10: Absolutely. I mean, it's like it's hard shit that we're working on. Like, it's like it's not easy. So no. Absolutely. And the last question is, what's the best piece of advice someone else has given you?

Javier Tordable - 00:52:21: It's hard to say. I mean, I'm a person that actually would listen to advice, and I listen to it objectively. I don't usually ask for advice, but when I get it, I would reason about it. And then I am not ashamed to steal advice from other people. For sure. For sure. I definitely have a lot of things that I would just sometimes tell to people. When I give advice, you know, I would just think that I was like, oh, yeah. I mean, I blatantly stole that from somebody else. I have no problem whatsoever with that. But one thing that I think is very generally applicable, not as many people as they should use, is to do things that make you happy because you are in charge of your own happiness. Many people think that the world needs to provide for them or other people need to provide for them or things will somehow turn out or or they have to, you know, work hard, and then eventually, they'll be able to be free and enjoy life however they want. I don't think that's a good idea. I think it's a better idea to align what you do, what you spend most of your time on with the things that make you happy. Right? So that takes different form depending on different people. Right? Spending time with friends, family, or working on interesting things, or traveling, or whatever, maybe for different people. Right? But we are responsible for our own happiness.

Jon Chee - 00:53:26: Absolutely. I couldn't agree more. I talked about this at Excedr, and the company is just like, at least this is for me. It's just like you spend a lot of your waking hours at work better like the work that you're doing. Again, it's not all puppies and rainbows, but, like, goddamn, you better like it. Like, you only got one of these. You only got one life. Just a just like snapshots in time, and you better

Javier Tordable - 00:53:47: You're gonna enjoy to like it too. Right? So you may initially like it, and then it's great, or you can train yourself to enjoy, to find pleasure on doing things well, right, in whatever job happens to be.

Jon Chee - 00:53:57: Yeah. And it's hard. Like, I think too, it's not a masochistic pleasure. It's like, you know, it's just like you know what it is like to start a company and you're running through walls all the time, getting no's all the time. Eventually, you start to relish it, and you're just like, I'm here for it, and I couldn't agree more. It is your responsibility. So well, Javier, thank you again for your time. This has been super fun. Next time I'm up in Seattle I'm in Seattle a lot actually. This I have a lot of friends, childhood friends who are actually now living in Seattle, went to UW and SU. So the next time I'm up there, I'll give you a shout.

Javier Tordable - 00:54:30: Love that.

Jon Chee - 00:54:31: Yeah. You guys have fantastic food and coffee, so there's plenty to do. And it's also, like you said, it's beautiful out there.

Javier Tordable - 00:54:36: So Yeah. Especially in summer. So, you know, if you get a chance to visit, let me know.

Jon Chee - 00:54:40: Is it super, super nice right now?

Javier Tordable - 00:54:41: Yes. Surprisingly, actually. It is one of the funny things in Seattle right off scratch. Like, people will tend to talk about the weather, surprisingly big amount of time. But the weather was supposed to be pretty bad this week. When I checked yesterday, it was going to be about ten days straight of rain and 60-degree weather. But today is actually beautiful. Clear skies. I think it's, like, 70. Yeah. It's almost 70 degrees.

Jon Chee - 00:55:02: That's the dream. I was gonna say, like, my wife and I, every time we go up there, like, when it's, like, summertime and clear, it's just like there really is no better place. We're talking about, like, our trip to Hawaii and just, like, similarly, just like, how does this exist? Like, when you're up in Seattle, like, you're just, like, surrounded by pure lush, and you're just right by the water. There's nothing better. So next time, I'll catch you when the sun's out in Seattle. This has been super fun. Thanks, Javier.

Javier Tordable - 00:55:25: It was my pleasure. Thank you for having me.

Outro - 00:55:28: Thanks for listening to our four-part series featuring Javier Tordable. From self-taught kid in Spain through dual degrees in math and computer science, sixteen years at Google building infrastructure at scale, and a deeply personal pivot into life sciences after his father's death from cancer to solo founding Pauling AI on the conviction that agentic AI can collapse the cost and timeline of drug discovery. Javier's story shows what it looks like when someone spends decades accumulating exactly the skills the world's hardest problems will require, then builds the company to solve them.

If you enjoy the show, please subscribe, leave a review, or share it with a friend. Join us for our next series featuring Philip Bawden, CEO of LabShares, a Greater Boston Lab Services and shared laboratory company, helping emerging biotech and life science teams get to work quickly with flexible, capital-efficient access to space, equipment, and support. Before joining LabShares, Philip spent more than two decades in healthcare and life sciences investing and company building, serving as a managing partner at Gallum Partners, a managing director at Longfellow Healthcare Partners, and a general partner at Riverside Partners after starting his career at Fraser Healthcare Partners. He also earned his MBA from Harvard Business School and his bachelor's degree in cell and molecular biology from Duke, where he was a captain and most valuable athlete on the varsity swimming team.

At LabShares, Philip is focused on making it easier for biotech teams to scale by giving them state-of-the-art shared labs, advanced equipment, and the operational infrastructure to move fast without the burden of building everything themselves. Philip's journey from healthcare investor to operator to CEO shows what it looks like when someone combines capital, strategy, and a deep understanding of the life sciences ecosystem to remove friction for science-minded founders, making this a conversation you won't want to miss.

The Biotech Startups Podcast is produced by subscribe. Excedr provides research labs with equipment leases on founder-friendly terms to support paths to exceptional outcomes. To learn more, visit our website, www.excedr.com. On behalf of the team here at Excedr, thanks for listening.

The Biotech Startups Podcast provides general insights into the life science sector through the experiences of its guests. The use of information on this podcast or materials linked from the podcast is at the user's own risk. The views expressed by the participants are their own and are not the views of Excedr or sponsors. No reference to any product, service, or company in the podcast is an endorsement by Excedr or its guests.