Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Ep. 25 Javier Tordable Saw Biotech’s Blind Spot. Then Built a Company to Fix It.
34 Plays8 days ago
What does a former Google leader see when he looks at the drug discovery process? A broken system and a once-in-a-generation opportunity to rebuild it with AI.
Javier Tordable spent 16 years at Google leading technical teams, and helping global pharma companies modernize through AI in his role as Technical Director in the CTO office. Now he’s the founder and CEO of Pauling AI, an early-stage startup using molecular simulation to radically accelerate how new medicines are found. In this episode, Javier breaks down what biotech gets fundamentally wrong about AI and the cost of those misconceptions.
We explore the bold bets he’s making to reshape the industry, what biotech teams get wrong when explaining cutting-edge science, and how hosting his own podcast, The Pauling Perspective, has sharpened his lens on what truly resonates.
Also on the table: Linus Pauling, investor mind games, and the one prediction about drug development that might make traditionalists sweat.
This one’s for the engineers, the skeptics, and anyone daring to rethink what’s possible.
Recommended
Transcript
Introduction to Sparktime
00:00:01
Speaker
Welcome to Sparktime, where biotech's thought leaders, investors, CEOs, and industry experts break down the evolving story of life sciences. Hosted by Danny Stoltzfus and Will Riedel, two scientists and strategic communicators, we dive deep into how biotech leaders can shape the narrative, win investor confidence, and communicate breakthrough science in ways that truly resonate.
00:00:21
Speaker
From emerging trends and cutting-edge technologies to what investors and partners really want to hear, we go beyond the usual echo chamber, bringing you fresh insights, unexpected perspectives, and the strategies that set biotech's top players apart.
00:00:34
Speaker
If you want to sharpen your corporate messaging, decode industry shifts, hear from voices shaping the future of biotech, and get inspired, then you're in the right place. Let's get into it.
Interview with Javier Tordable
00:00:46
Speaker
Welcome again, everyone, to Sparktime. Today we had the fantastic opportunity to talk to Javier Tordable. Javier is a computer scientist by training and formerly worked for a long time at Google.
00:01:01
Speaker
We got to hear the personal story that drew him into biotech. And he also told us about how he's really interested in the biotech version of Google Translate, you know, that tool that's free and extremely useful and and maybe you use every day of your life.
00:01:19
Speaker
He also told us about how we're still sitting in the assembly era of molecular simulation and we're not yet in the computation era of molecular simulation. And if you're like me and you have no idea what that means,
00:01:32
Speaker
Stick around to hear about some computer science history as well. Enjoy. Today, we're joined by Javier Todable. He's a technology executive investor and entrepreneur.
00:01:45
Speaker
He's currently leading Pauling AI, a next generation molecular simulation platform. Javier is incredibly excited about the potential of AI in biotech to find new medicines.
00:01:56
Speaker
Previously, during 16 years that he spent at Google, Javier led technical work and managed teams in search, ads, supply chain, and cloud. As technical director in the CTO office, he focused on healthcare, pharma, and biotech, helping organizations use cloud, machine learning, and generative AI to improve just drug discovery and patient care.
00:02:17
Speaker
Prior to Google, Javier was a software engineer at Microsoft, and he previously studied in a PhD program in mathematics in Spain. So Javier, it's really fantastic to have you here today. How are you? Thank you so much, Will. It's my pleasure. Very, very excited to be here with you and and Danny and you know you know answer any questions that that you may have. Thank you for inviting me.
00:02:38
Speaker
Fantastic. Well, first of all, I'd love to dive into your history a little bit more. Would you mind sharing a little bit about your journey and how you ended up running Pauling AI, especially after being at Microsoft and Google?
00:02:50
Speaker
Definitely. So ah my training was as a computer scientist. I studied computer science and and mathematics. I started on a PhD program and I realized very early that academic life was not going to be for me.
00:03:06
Speaker
So I dropped out of that PhD. I was very lucky as well ah in the sense that I got an internship at Microsoft. um That was a very interesting experience. Microsoft doesn't normally hire interns from Europe.
00:03:20
Speaker
So I came to Seattle, spent three months working. i got an offer to come back full time. And I think that changed so a little bit in my my career path. So I spent a couple of years at Microsoft and then I had a chance to leave for Google.
00:03:33
Speaker
And I would say most of my career so far was was at Google. um So I was also fairly lucky in the sense that I got the chance to work in a whole bunch of different things, um ah you know mostly building infrastructure, building tools for other people you know across many different areas.
00:03:49
Speaker
um So I spent a lot of time in in in different parts of Google. But about three, four years ago, i I realized that I wanted to focus little bit more on maybe more
Javier's Career Transition to Biotech
00:04:02
Speaker
um impactful things than what I was doing. and And I think, you know, like so many other people, I guess, in in ah in the biotech world or in or in the healthcare world, you know, there was ah was a family story, you know, my my father passed away. That was very impactful for me. And and I decided that um a I wasn't as happy as as I could be doing what i was doing.
00:04:25
Speaker
And I made a switch. um And I kind of you know jumped into healthcare and life sciences. I think it was ah you know it wass very interesting. I sometimes tell people i had you know one of the best jobs you could imagine. So i was I was working at the studio office and I was basically...
00:04:43
Speaker
um ah working with almost every video game company in the world. Many of them are are Google Cloud customers. I was basically helping them build their games using Google Cloud and using AI and so on.
00:04:55
Speaker
And you know for somebody that ah that likes likes video games, that's very artistic, very creative, very visual, I think that was just fantastic, you know like working with all these you know amazing people. So i left that, I just decided to make a switch and then spent the next three, four years basically helping biotechs and pharma companies and so on and so forth, um you know, hospital systems um use technology to be more efficient.
00:05:22
Speaker
And at some point, um about a year ago, I realized that that I wanted to... ah ah you know throw my hat in in the arena, so to speak, right and and and take a step at the idea that I was very, very excited about, I still am, um which is this notion of helping scientists find medicines better ah using language models and using automation.
00:05:45
Speaker
So I left ah Google, founded Pauling, found a few other people that are just as excited as I am about this. And we've been working on this for almost a year. And we're getting ready finally to to launch our product.
00:06:00
Speaker
Yeah, really fantastic. and And thanks for sharing that personal story too. Yeah, like I was just going to say, man, I wish someone had told me in my PhD program that there was a path outside of a research academic lab because i i tortured myself in there for about 15 years before I finally had that realization. So good on you for getting out much earlier than I did. But in all seriousness, I want to... um Just understand a little more more about your time at Google because I can't imagine having worked somewhere for that long. Like 16 years is an incredible feat to stay in any company. And so what I'd love to hear talk about was what did you have to unlearn when you stepped into biotech and away from tech?
00:06:42
Speaker
You know, I think... Let me answer the question in a slightly different way. So I remember I was reading while back some advice from ah Jeff Dean. For those of you that don't know him, you know he's a little bit of a legend inside of Google.
00:07:00
Speaker
He's one of the early engineers, he was like employee number 14 or something like that. and He's a fantastic programmer and an engineering leader. there you know a little bit just but People make jokes of of Jeff Dean, it's like the Chuck Norris of programming.
00:07:19
Speaker
And so I was reading a little bit of advice from him and ah this thing really resonated. When you're an engineer and you make career changes, it's useful to think in terms of a grid, where you have on the horizontal side, um common skills that may be useful across many different functions, and on the vertical side, um you have different areas where you could work.
00:07:44
Speaker
Or the other way around, it doesn't matter, right? These just two dimensions. So one of them would be, you know what, if if you're really, really good building databases, um you can build databases, set up databases, manage databases in many different industries, right?
00:07:58
Speaker
You can manage databases for you know for marketing, you can manage databases for biotech, right? um And the other way around is, you know if you are you know a biotech expert, you can work in different factors as well, right? you know you can be ah you know you can work in commercialization or you can run a company or you can you know you can manage a scientific team or whatnot.
00:08:17
Speaker
So he he said, you know it's very useful when you're making career changes to change only one of those i dimensions at a time. um So if you're an engineer, it's a good idea to continue doing the same type of engineering work that you would be doing, but maybe in a different space.
00:08:34
Speaker
Or if you are sure that you want to remain in a certain space, um you can change the the function or or a little bit of the skills instead of trying to change those two things at the same time. And when i when I started thinking about this, I ah you know kind of tried to do that, right? So I was still at Google.
00:08:53
Speaker
um I was an engineer. I was working on engineering problems. um I switched a little bit to work in little bit of ah ah a different function, and that helped me learn, helped me make connections, you kind of you start learning a little bit about the space, you know what matters, what do people care about.
00:09:09
Speaker
And then at some point I thought I was ready and then I made and made a change and and left the company. So I think that advice is very solid, right? You know, would say, you know, i would say you know other, know, maybe for other people, other engineers that are that are interesting that are interested in the biotech space, would say, you know, think about what kind of skills you can bring to the table and how those skills can be useful to companies within within biotech, which I would say is probably you know much much more and much wider than what people imagine.
00:09:42
Speaker
i I love that analysis because for me, like thinking really logically like that works and as scientist and I'm sure engineers are the same, like just breaking it down into like a a grid like that.
00:09:56
Speaker
It just seems so intuitive to me. So thank you for sharing that. I'm going to take that advice with me as I go forward and to whatever else I might do after to after what I'm doing right now. So that's really cool.
00:10:09
Speaker
Yeah. Great advice from a legend.
Tech and Biotech Synergies
00:10:11
Speaker
So Javier, you've you've been in biotech at Pauling for some time now, but even before that ah working in the CTO office.
00:10:22
Speaker
um I'm curious, since you've been working with biotech companies, um what what is something that biotech still fundamentally misunderstands about tech, especially ai And what do you think the cost of that misunderstanding is?
00:10:41
Speaker
ah that is That is such a big and and and complex question. Yeah, I'm sure there's something, there's at least one thing and I'm sure it's pretty big. We asked a questions here.
00:10:53
Speaker
Yeah. yeah so let me ah So let me let me try to try to answer, try to tackle it a little bit. So I think um in my mind,
00:11:05
Speaker
A lot of biotech is driven by but scientific or or business needs. but and And the people in that industry you know think as scientists or as business people.
00:11:18
Speaker
right So you would have, um you know maybe in the initial stages, you know people are driven by the science, like you what experiments have you done, what data did you get, you know what does the data show, like you know very, very much data driven.
00:11:31
Speaker
um And, yeah you know, maybe later stages, you know, it's a little bit about what is the the opportunity, right, to to bring a ah molecule to market and commercialize it and so on and so forth, right?
00:11:43
Speaker
um Whereas most of tech is driven, um or at least, you know, was driven, know, maybe not so much right now, but was driven by technology and ah engineering and how we can apply, you know, engineering is basically how do we apply technology to solve problems, right?
00:12:03
Speaker
So um sometimes in tech, people would build things because we thought it would be useful at some point in the future um because we enjoy the technology, regardless of how it's going to be applied later.
00:12:19
Speaker
um you know We build technology because it's useful for maybe one problem that we have at hand, and then that technology ends up being much, much more useful for a lot of other things like that. And sometimes it's very hard to predict how these things are going to turn out.
00:12:32
Speaker
So, um you know, if you think of, you know, a lot of the Google, a lot of the foundational work that was done at Google, for example, in in the early 2000s, you know, in distributed system design and so on and so forth, um a lot of the people that worked on this probably could not have predicted how you would usher this whole area of cloud, right?
00:12:52
Speaker
um if you If you look at... you know when mobile devices, for example, started becoming popular, um you you know a lot of a lot of that was built, you know not necessarily because you people had in mind that these devices were going to sell ah by the billions and and generate tons of billions of dollars in profit, but because people thought, okay, this technology is coming up, it's ready, it's going to be useful, we can do all sorts of cool things with it.
00:13:16
Speaker
I suspect a lot of the work that was done in AI before it became immensely popular, right? The work that was done in 2012 to 2015, right? Like very, very early work on language models and so on was done because People, ah you know, researchers ah thought um it it had promise. ah It was interesting. It was intellectually stimulating.
00:13:46
Speaker
um But a lot of it didn't really make much money. It's one of those things that, you know, like, ah people like you know, folks would would come up with, you know, ah you know all sorts of language model related research for Translate, for example, in in Google Translate.
00:14:02
Speaker
um Google Translate, as far as I remember, has never been a paid product, right? like So it it never really generated revenue for the company, right? But people thought, okay, you know maybe we can actually solve this whole you know human language translation problem. right Maybe we can preserve you know other languages. Maybe we can um you know make the internet ah more usable to everybody in the world. right And there wasn't really a... You know couldn't just go and say, oh, I'm just going to you know pay $9 a month for Google Translate, right and and that makes money for Google. right like
00:14:33
Speaker
That never existed there, but there was a little bit of ah of a different impetus to it. So kind of a long story, right but I think... um In my mind, a lot of biotech is unfortunately short-sighted and hasn't invested in this type of technologies that um that have spawned massive businesses for tech as a whole.
00:14:56
Speaker
but um you know you I think now, for example, it's it's very easy to see how AI is this you know massive you know you know business that is booming. OpenAI, Microsoft, and many others are making you know lots of money with it and and Google itself.
00:15:09
Speaker
um 10 years ago, it was very, very hard to predict that. So um I don't think biotech in general as an industry is making the kinds of investments that is going to lead to to this type of you know massive businesses you know down in the future.
00:15:23
Speaker
So, you know um Yeah, I'll live with that. yeah i can I can continue going on, but that's probably the main the main difference to me. right and And I think in in some ways, ah it's a little bit unfortunate to see that. And it's maybe the reason why a crossover between ah you know folks coming from from biotech and coming from from big tech may you know bring interesting innovations, right interesting technology that may be useful for for everybody.
00:15:53
Speaker
I resonate with that sentiment that you're sharing. It's like, it it is short-sighted, right? Because, it's hard in biotech to see five, 10, 15 years and into the future and say, well, if I invest this now, this is what I'm creating. I mean, so that's one aspect. And then like one example that we've talked about recently on the podcast was, you know, like to get CRISPR to where it needs to be is probably going to be like a 30 year journey. And that's,
00:16:24
Speaker
to sustain momentum through, you know, everything going on in the industry for that length of time and keep getting the funding and things is really challenging, especially, you know, in climates that aren't risk on, if you like, is probably the right way to say it. So,
00:16:41
Speaker
i don i don't I don't know if I have a lot of coherent thoughts, but I understand the depth of the problem and I do agree that short-sightedness is is part of the cause of that.
Challenges in Biotech Tools
00:16:50
Speaker
um So really interesting to hear your your perspective, Javier.
00:16:55
Speaker
Yeah. um I want to like completely change direction for a second here and rewind to the torture of my PhD once again. didn't think I'd ever do that twice in the same podcast. But one of the things I battled with immensely was as a chemist, I was doing a lot of molecular modeling and simulations. And, you know, I could easily remember like one simulation taking weeks and weeks and weeks to complete because there was like 200 atoms in the mo in the molecule, right? And
00:17:27
Speaker
I remember at the time thinking, wow, if only we had computers that were so much better, like how much more could we do? and ah I'm obviously dating myself a little bit by revealing that. But I'm curious because you've obviously had the realization that molecular simulation needed to be upgraded. And i mean, what what what did you see that other people had missed and made people like me suffer through their PhDs with, you know, ah crappy molecular simulation?
00:17:57
Speaker
um You know, I think that is that is such a ah funny anecdote, right? And I think most people ah went through the same, right? Like most of chemists that I know, ah computational chemists, you know, went through similar things.
00:18:10
Speaker
um You know, if I could tie to ah what I said before about... ah like long-term investments. I think part of the reason ah that is the case is because ah people have not made the the right investments in building tools in making these things easier to use.
00:18:32
Speaker
There was... um But one of the main differences between ah technology or traditional, you know, or big tech and other industries is that people have made a lot of investments over time that ended up resulting in increases in efficiency.
00:18:47
Speaker
ah So, for example, You know, when when computers were were first invented, people programmed them using, ah you know ah you know, little punch cards, right?
00:18:59
Speaker
And then assembly was invented. And then assembly became very popular, but it's a very, very hard language, right? So so folks, you know, universities and other places decided, okay, why don't we create higher level languages and ah compilers and and other things? And then, you know, open source came in and so on and so forth.
00:19:17
Speaker
And the end result of that, if you look at it at ah at a big scale, it is that in, you know, 40, 50 years, we went from um ah situation where you had, where you needed hundreds of people, really, really smart, um you know, lots and lots of years of work to build anything to a point where,
00:19:39
Speaker
Software engineers are very, very productive. You can have an average software engineer today, you know, using open source code, you know, create a website, you know, that, you know, you click buttons, it does things, right? 40 years ago, that would have required hundreds of people, literally, literally hundreds people and years of work. Now it's something that one person with the right, you know, tools and open source code and so on can use very, very easily.
00:20:01
Speaker
So um I think that ah development of technology has not happened successfully in biotech in general. ah Now, there are things that that we can probably do right now. So, you know for example, you know we can ah build better tools for scientists ah to do this type of simulations.
00:20:23
Speaker
We can take the technology that is out there for for this type of simulations and then make it more um you know make it easier to use, ah make it more resilient. ah We can build layers on top of it that make it easier for,
00:20:37
Speaker
you for language models or for other automation tools to use, and so on and so forth. um I think, in my mind, a lot of the potential of ah computational chemistry tools has not really been exploited.
00:20:55
Speaker
like in In some ways, we're still in the era of assembly programming, ah where somebody would... would go and spend you know two days of work you know running a molecular simulation in Gromax and then you know seeing what happens and then learning what every single little flag does and this and that.
00:21:12
Speaker
like There's no reason why it has to be like that, right? like you know There's no reason why you couldn't have an iPhone app where you you know tap a button and and then you know a speaker says, you know what do you want simulate today? And you say, well, I want to simulate a benzene molecule s floating around in you know water or whatever. you know Go ahead and do it. And then two seconds later, you have that thing you know recreated and and working.
00:21:34
Speaker
Like there's no inherent limitation why why we can't build that, right? It's just a matter of having the right incentives. um So in my mind, um you know like that would have been yeah A few years ago, that would have been very, very hard. It would have required a lot of people, a lot of time.
00:21:51
Speaker
Now, we are at a point where it's becoming easier to build those kinds of tools with a reasonable number of people, with a reasonable investment. you know like Language models are are kind of enabling us to do some some of those things and so on and so forth.
00:22:05
Speaker
So yeah I would say, you know back to your question, you know maybe the the thing that I saw a little bit different is um If we can leverage some of the tooling, some of the infrastructure that that is coming up, you know, in particular ah language models and and and other tools in order to make computational chemistry more usable and easier, then, you know, we'll probably get some benefit, right? In the sense of ah enabling people to make scientific discoveries or or other discoveries that useful for, you know, for dry discovery, um you know, better than they couldn't up to now.
00:22:42
Speaker
Wow. It like blows my mind well. I think this is like, I don't know, the second or third conversation we've had about the importance of building tools to accelerate drug development in 2025. I feel like this is going to be like the year of the tool.
00:22:58
Speaker
Well, I love hearing, Javier, how how these threads have have really come together in what you told us earlier, that tools like Google Translate, maybe they have some some use, but maybe the the payout isn't clear. and um And what you just said, that computation...
00:23:16
Speaker
you know the computation era hasn't been tapped in in biology, in in biotech, or in drug discovery, and or or when it comes to molecular
AI's Role in Biotech
00:23:25
Speaker
simulation. And we're still kind of in this assembly area era. So, i love hearing all of that come together.
00:23:33
Speaker
in your answer. But i'm I'm curious now, when you're talking to investors or you're talking to partners, what what is that piece that that makes everything click that when they know they need your help and they know they need ah Pauling AI to to solve the problem for them?
00:23:49
Speaker
What is that moment in your mind? Yeah, I would say that are, I think that are maybe two messages that resonate.
00:24:01
Speaker
The first one is a little bit more um down to earth utilitarian. I think most people can understand, right? So that is basically what technology does across many different industries. And it's not particular to biotech, which is automation, right? Automation and building better tools.
00:24:18
Speaker
So if you have a specific process that people normally do for whatever reason, and then you make that process faster and cheaper because we automate it, then there is value there.
00:24:31
Speaker
And that applies across a lot of different things. So like you can you can look at you know marketing, right? People spend time writing up an email, you know writing that copy, you know reviewing it, creating some some assets or looking at stock photographs and then pasting them in the email and then sending them email to people. Hopefully, you know they click and they buy a product, right?
00:24:49
Speaker
That is a process that humans do and it takes a certain amount of time. So then you can go and say, okay, I can use a language model. I can build a infrastructure. infrastructure that can create a tool that automates that entire thing so that you know you click a button and then you know instead of spending you know two hours doing this work, it takes 10 seconds and then you review it and send it and that's it.
00:25:08
Speaker
pleasure So automation has value by itself. and And I think the first thing that we're trying to sold that would ah we're working at pulling is, can we automate many of the things that computational chemists do? To the point where, for you know yeah for a lot of projects, a computational chemist would spend months months doing work that is not necessari necessarily um adding a lot of value to the business. right And if you can go and say, instead of spending months, ah spend you know a day or a couple of days because the tools are so much better than to do these things for you automatically.
00:25:40
Speaker
And that is valuable. That is interesting. I think investors in general understand that. Like that is not industry specific. Now, the second point that I think is very interesting and is where the real promise comes in is once you have an automated way of doing these processes, then the The improvement is not just quantitative, it's also qualitative because you can do things that you would never have thought about doing before.
00:26:11
Speaker
And I'll give you an example ah that is biotech specific that I tell you know a lot of people and yeah and I think medicinal chemists, for example, in really like it, which is reverse docking.
00:26:23
Speaker
um So once you have, you know sometimes people, you know they do you this process, they do a heat screening, they start testing you know back and forth and then they think, okay, I have a you know a potential lead candidate. So now let me try to do some kind of reverse docking to see if it could you know if it could possibly be toxic by trying to predict, know would it bind to some of these proteins that in a way typical toxicity, right? you know So liver enzymes so or calcium channels in the heart or whatever.
00:26:49
Speaker
So sometimes you would spend little bit of time running those simulations, trying to understand them. um but but that process ah takes a certain amount of time and takes some expertise, right?
00:27:02
Speaker
So it may take, i don't know, a few hours, half a day, right? You know, for every, you know, protein that you want, you know, go ahead and simulate. So there's only so many of them that you can do because you need a person, you need an expert that needs to do that.
00:27:15
Speaker
um Now, in if I go and tell you, you know what, um we can do this thing in a fully automated way, you know, for a dollar per protein. It would be like, well, let's just run it against, you know, all, you know, 10,000, 20,000, you know, human proteins, which you have a structure, all of it, right? I give you a check for $10,000 and I save myself, you know, one month and hopefully increase my probability of of success with this specific drug, you know, from A to B, right? Or reduce my likelihood of having a toxic compound from, you know, XX percent to Y way percent, right?
00:27:44
Speaker
yeah ah So all that in exchange for $10,000 or $20,000, in many cases, that's a no-brainer. But that can only happen if you have this fully automated way. right The moment that you need ah some you know human in the loop, you know you cannot enable this this sort of fix.
00:28:01
Speaker
So think it's just one of the examples in which once you have a fully automated platform that does this work, then you could do all sorts of different things that you can't you know even conceive right now because you know we are depending on having humans do the work.
00:28:17
Speaker
It's really interesting, a does this ho human in the loop concept, because we've had other people were on the other side of this argument saying, but you need someone in there guiding yourself.
00:28:29
Speaker
to make sure that it's not nonsense or that it's scientifically or chemically accurate for want of a better way of describing it. Do you ever get that sort of pushback when you talk about removing humans completely from there the, from the loop?
00:28:44
Speaker
Yeah. I mean, I would say, um Within drag discovery, nobody is removing ah you know experiments and data, right? There's there's no way around it. you know the The whole point is you know how long and how much does it take to go from an idea to ah to a molecule that you can test?
00:29:03
Speaker
um there are that are ah you know I think two schools you know in in in the world of of AI, for example, you know that ah you know a lot of people subscribe to. On the one hand, you have folks that want to understand what the models are doing. They want to understand you know how they make decisions, how they work, you know the way they work.
00:29:22
Speaker
They want a kind of a white box you know type of models because they they think you know humans need to need to be able to understand. And in some fields, I would say, You know, maybe, right?
00:29:34
Speaker
um If you have a model that is making a diagnosis ah for a patient, maybe you could say, OK, I really need to understand you know every single detail on how and why you made every single diagnosis.
00:29:48
Speaker
um On the other hand, there's there's a camp of folks that believe that um the only thing that matters is our benchmarks and if our models satisfy those benchmarks. And as long as they work, it doesn't really matter um how they do, what they do, or or why.
00:30:02
Speaker
and I subscribe a little more to that second camp. It was a little bit of a change for me, right, coming from a traditional engineering background. But at some point, I realized, you know the only thing that matters is, does this system, does this, you know, black box solve my problem or not?
00:30:19
Speaker
So if I gave you a magical AlphaFold7 model where you feed it you know some disease and it gives you a magical molecule that that has all the you know ah properties that you want, it has all the AdMID properties, and low toxicity and and all the things that you could possibly want, and um you know and it works well, it binds to your target, you know super high affinity and so on and so forth, but then you have no idea you know what ah you know the the mechanism of action you know would be or or you know or or why the model came up to that conclusion.
00:30:55
Speaker
Does it really matter if you understand how the Blacks box works or not? I would say it probably doesn't. You know, there are medicines today in the market, right, for which we do not really understand how they work, but they work and they're approved and they cure people's diseases, right? So, you know, isn dr do we as humans really need to understand every single decision? I would say probably not. And and it's a little bit of a handicapped for the people that really need to understand that in order to adopt technology in general.
00:31:28
Speaker
Yeah, that white box, black box description is one that we've encountered multiple times. And you're right, there are people on both sides of that fence. And I think...
00:31:40
Speaker
you know, one thing that we think about in terms of communicating AI and the technologies and the tools that are advancing drug discovery is, well, what happens when it gets to the regulators and what are they going to think? And, you know, I think there's, but again, there's probably two camps there as well, but I think um you know you would know better than me about the potential pitfalls of not understanding the full process. But I really like there the way that you describe that.
00:32:09
Speaker
You know, there's lots of things that we don't know and yet drugs are still approved. So is that really a reason to not ah not give it approval for a new drug that's used a ah black box tool, as you described? So really i really like thinking about that from a different perspective. So thanks for sharing that, Javier. Yeah.
00:32:28
Speaker
Thank you. I mean, it's it's kind of it's hard to hard to say what perspective will eventually work. But if I had to bet, I would say you know black boxes would typically... when you know Most of us, for example, like you know we don't understand how our cars work. I mean, like, you zealand rep but most people, I bet, do not understand how an internal combustion engine works, ah you know much less you know how a lithium-ion battery works. And we jump in our cars you know and go to work and you know it doesn't matter because the car takes you there and it's safe and you know hopefully it will take you back home. And you know we don't really care so much about you know the internals of it. right
00:33:05
Speaker
So I think at some point, you know industries will in general will go through this same process with with AI, in which you know what matters is is the benchmark, what matters is you know how you test, how you evaluate the tools or the infrastructure that you have. And it satisfies those benchmarks and it works well. then people will stop caring so much about, you know, how or why it doesn't make decisions.
00:33:29
Speaker
Yep. Yep. I think those those are great analogies. And I want to add one more is that most people don't know how a toilet works, which is something that's in your home. but But, it but ah you know, exactly until it breaks.
00:33:41
Speaker
But I think, you know, hearkening back to our conversation with ah Jeff Baker, former deputy director of the FDA or of an FDA office, we know what the agency
Communication in Science
00:33:50
Speaker
wants to see. And that's, efficacious or or that the product is efficacious, that it's safe and that the process is reproducible. And so, you know, plugging those metrics in, um, with, uh, with an AI assisted, uh, uh, drug development process, I, you know, it, it still makes a lot of sense. Like you said, Javier. Yeah.
00:34:11
Speaker
Yeah. And I also think you revealed that one of your secret recipes for communication, because you're obviously an expert at translating complex things right to people of all sorts. And I love that you jumped straight to analogies, because this is one thing we try and tell people all the time, especially scientific founders, is like, look,
00:34:35
Speaker
Not everyone's going to grasp the scientific beauty and complexity of what you're doing. So let's find an analogy to communicate it. And clearly you're a master at that, Javier. So I just wanted to highlight that whilst I was still thinking about it. But Oh, thank you. don't know if I would consider myself master, but I've definitely sat through a lot of really, really boring scientific presentations even in my few years. I thought, you know, this would be so interesting and so exciting, but the person just cannot make it exciting. And so I'm going to try to apply that little bit to myself. Yes. Well, you're doing a great job, so I appreciate it. And
00:35:14
Speaker
that having following on from what you just said, you obviously see have seen people make a lot of mistakes around communication, um especially at the frontier of tech and biotech.
00:35:26
Speaker
So what do you wish they would do instead? is Is it use more analogies? Is it just make it less boring? Like what what else would you say would be really effective? Yeah. Yeah, I mean, ah in my mind, analogies are our great tools because when you can put something, no matter how complex, in the same words and same situations that somebody finds in their day-to-day, that makes it much, much easier to understand, right?
00:35:53
Speaker
ah But I would say... You know, most most people, ah especially most scientists, they don't really know how to weave a story into what um into what they describe.
00:36:08
Speaker
You know, in my mind, um so much of scientific discovery is, you know, such, you know, so so many challenges, so much, you know, intellectual, you know, um you know, problems that are overcome, right? Like through, you know, you know deep thought and running experiments and people that thought one way and then they changed their mind and so on and so forth.
00:36:29
Speaker
um And when you read ah really good ah scientific communications, you know, like, you know, there's so many books out there, you know, Siddhartha and Makarjee, for example, you know, like wrote a whole bunch of books about cancer and so on and so forth. And and there are many other you know really, really good scientific communicators. When you read that, you think, wow, this is this is so challenge so interesting, right? you know how How people overcame all these challenges.
00:36:51
Speaker
um And then you go to you know your average scientific conference, right? And it's like, well, we run this experiment, we run this other experiment, we sold this data, here's what we found.
00:37:03
Speaker
We did this way, we did it that way, we found something else and we changed gears and then blah, blah, blah, blah. blah blah blah And like, literally like, it's so easy to just fall asleep, right? um I would say, you know, my main advice, sort of like, if could if I could say anything, right, is, you know, for folks to maybe develop a little bit the the skills of,
00:37:24
Speaker
um kind of crafting a story, right? Of like, you know, well, you know, here's what we thought and then here's why it didn't work and then here's what we thought again and like how this is surprising and and then, you know, well, and this, you know, is really exciting because now, you know, it could open the door for something else, right? And and of course, you know, the experiments are going to be, you know, ah you know, sometimes they're very boring and, you know, they're experiments in animals and there's all the data you have to explain and so on and so forth and and scientists have to be careful, right? You know, to put a lot of,
00:37:53
Speaker
you know um you know claims that are too bold that other people are going to you know grab and and take out of context and so on and so forth. But I think there's still a bit of room for maybe being a little bit more, i don't know, exciting, right? and when went and Yes, you're definitely speaking my love language now because I 100% agree that it could be a little more dramatic and twisting and turning and you know it's it's much more engaging so i don't know i could go on and on about that but i won't bore if wrong with it so everyone knows what i think already
00:38:30
Speaker
Okay. So, Javier, I want to try and put a bow on it for our audience. um and And by it, I mean your career. You've gone from Google Ads to drug discovery. So, what is the through line in your career that connects all of those elements that actually makes perfect sense once you say it out loud?
00:38:52
Speaker
Yeah, um oh ah I'll give you two answers. I would say the real answer right is throughout my career, what I've been doing is building tools. That's what I'm good at right? i I know how to look at a problem, look at a process and build a tool that makes that process better.
00:39:12
Speaker
um That is a very, very general set of skills. um you know I used it, yeah i built you know I built tools for search, I built tools for ads, I built tools for you know supply chain back when I was at Google in the supply chain team and so on and so forth. So I would say that that is really the the thread that ties things together.
00:39:29
Speaker
um But the other thing that that comes to mind is you know There was this quote, and I think it's from Peter Thiel, which I heard ah many years ago and it really hit me, which is, ah we have the brightest minds of our generation trying to improve click-through rates.
00:39:52
Speaker
And I thought it was so, so deep and so sad. It is really, really sad. Like you find incredibly smart people that are just trying to figure out how to get Fox to to click on ads online and buy shit.
00:40:06
Speaker
and And I was like, as a society, right? Like, why are we spending so much money, so much effort on this, right? ah Instead of, you know, curing diseases, instead of, like, actually, you know, discovering novel science, right?
00:40:23
Speaker
um You know, so when when you look at, you know, like, you know, largest companies in the world, right? Like, you know, when you look at, you know, Google and and and and Facebook and Apple and so on and so forth, I mean, you know, not to discount, you know, they build great products that that people like and and buy and that's why they have a lot of money, right? But at the same time, you know, maybe some of the some of the effort and some of the,
00:40:45
Speaker
resources could be dedicated to things that eventually make more of a difference. so And it always really drove me crazy how somebody would would go and and spend $1,000 on a phone you know without thinking twice, right but then they would not pay $9.99 a month right for some you know you know a service that helps them with their mental health, or a gym subscription, or an online nutrition you know website right that helps them eat healthier or whatnot.
00:41:11
Speaker
i think you know As a society, we have a pretty hard time investing in the things that are really important, that really matter.
Pauling AI's Mission
00:41:20
Speaker
So, I mean, I don't have a solution for it, but but in my mind, you know I i just decided, you know what, I'm just not going to be part of the problem anymore. I'm just going to focus on things that are more important.
00:41:31
Speaker
Yeah. Well, I very much feel that way about ah has clearing out things in my life in general and just focusing on what's important. So I think we're we're aligned on that one as well. And i of course, as soon as I...
00:41:48
Speaker
I saw the name Polling ai I immediately went back in time to when I was in Oslo and at the Nobel Prize Museum. And of course, he won two Nobel Prizes. It was one of the few people highlighted there as having done that as a, I think he was independent as well and winning them. So is he related to the origin of of the company name? And if so, what do you think he would think of what you're doing today?
00:42:15
Speaker
Yeah, yeah, of course. um I mean, the the company is named after after Linus Pauling. um i Yeah, I don't think I can dare to put words in his mouth. Really? Didn't take the date. Yeah, I mean, I think that's maybe a little bit too much. But I have to say, I have a lot of foundations for scientists. In some ways, kind of also took inspiration from...
00:42:43
Speaker
ah from Tesla, right? You know, which is another company, but it's named after a famous ah famous scientist and and engineer. And yeah, I think there's something interesting about ah being bold, trying to solve major problems, you know, mainly in the world.
00:43:02
Speaker
And you know I thought, ah yeah, I was actually surprised that ah that nobody had named the company after Linux polling. The Nobel Prize Club is a very small club, right? Yeah.
00:43:16
Speaker
So ah you know you know the domain was available. I thought, okay, this this is it. you know this is This is my company name. Very cool.
00:43:25
Speaker
So I want to hear a little bit about, you know, the bold bet that you're making at Pauling. What is the the big bold bet that you're making and and um what are you excited about in the future and in changing the industry if it comes true?
00:43:41
Speaker
Which we hope it does. Yeah, I hope so too. I mean, I think at a high level, right? Like, you know, this there's this notion that or this idea that we can make computational chemistry much better, much easier, much faster.
00:44:02
Speaker
And we can um take ah process that takes, that requires you know six months, a year, a couple of years and make it radically better and faster and cheaper.
00:44:15
Speaker
And by doing that, we will make it easier ah to find new medicines. That's the ultimate goal, right? So it's it's not ah for the sake of, you know, just building better tools and then sell subscription software, right?
00:44:30
Speaker
um Drug discovery is a really, really bad way to make money. It is... Yes. It is very tough and very hard and it takes years and years of investment, right? So so the goal is it's not so much, you know, going to you know to build, you know, some tool and and start monetizing it right away.
00:44:46
Speaker
the The ultimate goal is um We have, like we as a society have this massive problem that it is very hard and very expensive to find medicines. It costs hundreds of millions or billions of dollars. It takes many, many years.
00:45:02
Speaker
It's becoming worse. um And and the the challenges that we have, you know, keep, getting worse, right? You know, antibiotic resistance, all sorts of, you know, different, you know, ah you know diseases and, you know, you know many, many of us are, you know, getting cancer, degenerative diseases and so on and so forth.
00:45:19
Speaker
So the problem is getting worse. um the the the The idea is, you know can we build better tools that will help us find new medicines better and faster than we can today? but and And hopefully we can do that using technology, using AI you know you know in a smart way.
00:45:39
Speaker
um you know And if we're successful, you know hopefully everybody would benefit, right? you know either Either us or other people that that use the product will be able to find medicines and and cure diseases and make people's lives better, right? like that's ah That's the hope.
00:45:53
Speaker
That's what we do, what we do. I just feel like there's such a surge of people in your boat with you who really want to make medicines more accessible, utilize tools that we don't yet have, but desperately need. I feel like it has to happen, right? It just, it's a matter of when. And I,
00:46:16
Speaker
i was I want to encourage everyone to be rethinking how they do drug development because I think the current model isn't sustainable and that's probably a whole different topic that we could have another hour conversation about. But Javier, if people are interested in and learning more about polling ai um first of all, tell us what's coming up for you in the next year or so at at polling and you know how how's the best way for people to contact you that want to learn more?
00:46:45
Speaker
Yeah, a year a long time, so I don't think I can answer that question. But but we we've been working at this for ah close to a year. We're getting ready to launch.
00:46:57
Speaker
um And then ah we will be looking for, ah you know, we have a few um early stage partners with whom we're collaborating. We'll be I'm looking forward to working with with a few more folks. of you know I'm happy to talk with anybody that is interested in the space. um we will be ah We will be hiring, you know we will be you know looking for more more customers and and so on and so forth.
Closing Remarks and Future Invitations
00:47:20
Speaker
um I would say um reach out ah ah either on you know Twitter or LinkedIn, any social media platform. um I'm not going to share my email, but it's not really hard to find. Absolutely. Just for the sake of not leaving it in the in the recording. ah But, you know, so I'm ah happy to reach out to anybody. I'm happy to talk with anybody that is that is interested in this. And if I can help them, you know, maybe in in their own journey in drug discovery, I'd love to do that.
00:47:50
Speaker
And, you know, hopefully, you know, things will will work out ah for us. and And, know, maybe I'll be back here, you know, a year from now saying, well, you know, here's all the things that we did over the last year. That'll be a much, much more interesting podcast.
00:48:05
Speaker
Phenomenal. I mean, really, that that was an extremely interesting discussion because we've never had a computer scientist on the podcast. But, you know, it's been so fun. So I guess we'll have to have more computer scientists and we'll have to have you back on the podcast, Javier. And of course, we never talked about it. But Javier, um you have your own podcast and we encourage our listeners to go listen to the Pauling Principle podcast. So,
00:48:28
Speaker
um it's It's been such a pleasure to have you on, Javier. Thanks so much for joining us. Thank you. Thank you so much. The pleasure was all mine. I i really enjoyed the conversation. So thank you and and thank you for the opportunity to to join you.
00:48:44
Speaker
hearing Javier talk about, he kept using analogies and like, I'm not especially or at all tech savvy, but I feel like I could fully capture what he was talking about just for the way he was describing it. And I mean, of course he gave us a big shout out because he was talking about storytelling and drama and ups and downs. And we always, you know, talk about that as well, Will, but I don't know, did that just resonate with me because I'm so un-tech savvy or did you find that really appealing as well that he spoke in analogies?
00:49:18
Speaker
Well, a few thoughts and the first of them is that during the discussion, Dani, you did say that that that analogies are your love language. um did Did you mean to say that?
00:49:30
Speaker
Yes, absolutely. yeah When it comes to tech, yes. Because otherwise I have no idea what's going on. Okay, got it. Okay, that's funny. um I think the second is, yeah, analogies just work.
00:49:43
Speaker
ah We don't need to know how. It's such a good illustration that we don't need to know how everything works if we can show that it does work and that it's safe and it's it's something that's extraordinarily useful. So that is universal.
00:49:58
Speaker
um I also think it's so fun to talk to people in completely different fields because the analogies they use that you know they they they can throw around. don't know what saying that Javier was doing this, but... they tend to have analogies that you've never heard before and you think, oh my God, that's so interesting. And so maybe maybe that's what he was speaking about or maybe those those were um you know off the cuff, something that Javier says specifically. But I love analogies. I love how they bring you into the story. Of course, they show you a piece of past drama and now suddenly you as the audience are on the journey with that person. So completely agree.
00:50:33
Speaker
Yeah, well, speaking of analogies, here's what I never want to hear again. we are building the plane whilst flying it. I think that is well overused in the biotech space, and I'll be very happy to never hear that one again.
00:50:48
Speaker
Start saying that we're building the boat as as we're... we're sailing it? well that's going to lead to us drowning unless you've got a good life jacket, so... I don't think I'll get in that boat with you, thanks. So i i i i thought it was so interesting.
00:51:02
Speaker
that One of the earlier things that Javier said about Google Translate as as a big example of ah a tool that was built not necessarily to... um bring money in directly, but absolutely adds value. And it's so very interesting that that was built more so as can we do it rather than, well, how much money is this going to bring in? And there i think there are a lot of reasons in biotech why maybe we don't build tools like that. And and one of them perhaps being that biotech isn't
00:51:38
Speaker
necessarily driving revenue and and and rather the funding model is very different. at least that's my knee-jerk reaction to that. But I also wonder how these sorts of tools have been built by you know bigger companies like pharma companies and um why we don't see more of them and um and and what it would lead to if if we did have tools that perform very useful tasks that
00:52:07
Speaker
you know, what are the tools that I'm not thinking of? Surely they exist, right? huh Yeah, well, it's interesting that you say that, and it actually ties back to the Google Translate piece, because we just talked recently with Marie from Clinials, and her whole thing is translating using tool.
00:52:28
Speaker
Yeah. So you tied those two together nicely. And I mean, I would have never thought about that as a tool that we needed in biotech. But he you know she also came from a tech background. So again, this kind of idea of cross-pollination of industries is really bringing about things that you know we otherwise wouldn't have and probably desperately need and don't even realize it.
00:52:49
Speaker
A sneak preview for our our listeners is that we are secretly scheming to get Jeff Baker back onto the podcast. And, well, I think we should talk to him about this concept of black box versus white box said um you know, get his thoughts, not representing the FDA, of course, but just, you know, as someone who's spent so much time in a highly regulated space, like, does it matter if we have all, does it matter if it's a black box? Does it matter if we don't exactly understand where the information came from? Like,
00:53:23
Speaker
I mean, I really liked Javier's rationale of why we don't need to have that. But I mean, we we often face this question from clients in the space of AI drug development is like, how are you going to handle regulatory approvals? So I think, you know, when we get him back on, we should like just pester him mental until he tells us what everyone should be saying.
00:53:44
Speaker
Right. but Well, knowing Jeff, he won't tell everyone what to say. But, you know, I've always, always want to have Jeff back on, friend of the pod, Jeff.
00:53:55
Speaker
um But i i think I know what he's going to say. And I think it's going to, you know, make a lot of sense with what he said before. But it's it's a question that comes up again and again. There are people on both sides of it.
00:54:07
Speaker
So let's let's keep collecting those opinions. Yeah, for sure. Maybe we can take bets on what he's going to say. Well, thanks again to our listeners for joining Sparktime. We welcome you to join next time as we continue to explore the ideas, the thinkers and the innovations that drive biotech forward.
00:54:24
Speaker
We hope to see you there.