Enric Gibert: you, Imants for the kind introduction and thank you for inviting me. It's a pleasure to be here today to you and the So yeah, Intel. as ⁓ you mentioned, my is not chemistry, biology, or any background. I'm computer engineer. My areas of expertise are high-performance computing, software development, artificial intelligence. I did.
Imants Zudans: Enric, thank you for joining us here today. Maybe let's start your story with Intel
Enric Gibert: the PhD in these fields at the University of Polytechnic of Catalonia here in Barcelona, I live. ⁓ But during the PhD, I did a research stay at the University of Illinois in Urbana-Champaign in the US. And doing the I joined Intel. Intel at that point in time ⁓ had opened a research lab ⁓ in Barcelona that was very near to the UPC. And it was, in fact, Directed my PhD advisor, ⁓ ⁓ Antonio González. So it was a natural step for me. ⁓ I stayed there for 10 years and what we were doing there was design the processors, the chips, the hardware accelerators ⁓ at a 5-10 year time frame. ⁓ So ⁓ when need to design what is going to be the microprocessor or the GPU or the hardware accelerator ⁓ the Polytechnic University of Catalonia, there was a big research group on computer architecture. And the group had a pretty strong reputation. So Intel decided to open an R &D facility that was 2002, 2003. for this time window, five, ten years now, you don't have the hardware, so you to start doing some sort of simulations and modeling benchmarking against current products and is we were doing. Nothing do with life we were designing chips and the hardware accelerators of the future and idea was a research lab but the idea was to impact the product lines of
Imants Zudans: So you spend a decade at Intel and then it was in 2014 that Intel decided to close Intel Barcelona site, right? What happened? Welcome to Imants podcast, where founders powering drug discovery share how they built their companies.
Enric Gibert: The lab began by just eight people and over 10 years. it reached a peak of 80 people, 60 of them a PhD. it was like a small research site compared to other sites at Intel. mean, 60, 80 people, we're collaborating a lot.
Imants Zudans: My guest today is Enric Gibert, CEO of Pharmacelera Enric spent a decade designing processors at Intel. Then he started building software for drug hunters. How does a chip engineer end up in medicinal chemistry?
Enric Gibert: with the people in Santa Clara, the headquarters of Intel are. We were also collaborating a lot with the people in Oregon, and and ⁓ of these unfortunate in which a big multinational ⁓ had promised investors that they would ⁓ have sales of 50 billion. ⁓ they didn't reach that. They got like 49.5 or 49.7. I don't remember. So they felt short with respect to expectations. then from a management perspective, they decided to start closing small and mid-size sites. They closed one in Mexico. they closed one in Germany that was more or less the similar size of us, 100 persons, and then they closed ⁓ ours and concentrated, and they decided to concentrate all the R &D efforts in bigger sites.
Imants Zudans: That's what we explore today. that sounds very exciting, very interesting.
Enric Gibert: as you say, it sounds a lot of fun. We had a blast. I had the privilege to work with people that were super smart, super inspiring. OK, and that I saw that was extremely important to keep me motivated. My mom always said that while I was at Intel, There was not a single day in which I complained when going to work. ⁓ Because we were having fun. mean ⁓ story is a lesson learned ⁓ ⁓ seen ⁓ companies ⁓ big companies in which ⁓ they need to reduce headcount or they need to do some ⁓ You start hearing rumors, okay? ⁓ spend like six or nine months ⁓ asking yourself, is it going to be me? I mean, and then ⁓ just... lower your performance, your engagement because you don't know how you, your group, your site is going to be affected. I've seen that ⁓ happening in other situations. This was the opposite. ⁓ And it has pros and cons, but I remember there was a Monday in which the head of the lab in Barcelona was discussing with his boss in Santa Clara the budget for that particular year. ⁓ That was on a Monday. And they were discussing what projects we would be running for that particular year, what was going to be the budget for the entire lab in Barcelona that year. That was on a Monday. This same boss landed on Wednesday, so 48 hours afterwards, to say that the site was closing, ⁓ that we were laid off. ⁓ So it was like ⁓ immediate action, no rumors. ⁓ And in two months we were gone, all of us. ⁓ So this started to think, if this happens again to me, what would I prefer? Would I prefer to know it with six or nine months in advance and start thinking, ⁓ my god, ⁓ my god, or I would like to have a very short notice? Of course, mean Intel. was in good shape financially, so we got a good compensation package. when explained the story, a lot of people tell me, they behaved bad. I said, well, I mean, was straight, period. no pain. that point in time, since we were notified, I think it was like two months or like that. everybody started looking for opportunities. and this is another interesting ⁓ story. that that were said that we would be laid off. three or four days afterwards, ARM the chip designer of chips for mobile phones, that they are based in Cambridge in the UK, landed in Barcelona with all our profiles from LinkedIn, and they started interviewing. So that was one of the best talent acquisition exercises I have ever seen. In 48 hours, 72 hours. the managers of ARM were in Barcelona all the team. So lot of people went to ARM, to Cambridge. I had the opportunity to go to Cambridge. Some people moved to the to the Silicon I also had an offer to stay at Intel. And some other people just wanted to look for other ⁓ alternatives. In particular case, yeah, I had been for years in a multinational. So I knew what was for big company ⁓ But my was for something ⁓ I had a blast at of course. but I still had a lot of and I said if I don't try now I will not I will never try it yeah together Enrik Herrero we decided explore opportunities in the life sciences and when we decided to Pharmacelera
Imants Zudans: discovery, was that a deliberate or were you considering other options as well?
Enric Gibert: was serendipity as things that happen in life. ⁓ Together with Enrik Herrero, we were colleagues Intel, although it is true we had never worked together. We knew of each other. We were in different projects. We had never coincided in the same project. So it took us a while to get to know ⁓ each other. And we common friends. We looked at ⁓ genomics, ⁓ fascinating from a research perspective. You must also understand our background. Both Enrik Herrero and myself, we are PhDs on computer engineers. So we are scientists, we are researchers. working on genomics, we saw that it was super exciting, super challenging. It is true that our first interactions ⁓ we saw that it was still a very immature field. So we would have had a lot of fun because it's an exciting topic. But most of the money was ⁓ grants, it was public funding. So it's hard, it's tough from a business perspective. So we ended up discarding genomics. And at that point in time, for these situations in life in which I was taking my elder daughter to school and ⁓ a mother of a boy that was attending the same class, she was working, or she is still working at Almirall, the largest pharmaceutical company in Spain, and she told me, Enric. There is an interesting profile that was allowed to let go at Almirall because at that point in time Almirall was ⁓ reshifting their focus. There's a person that has worked at Almirall on for 20 years. ⁓ brilliant. ⁓ should talk to him. ⁓ I'll do the introduction. ⁓ we met, ⁓ Herrero and myself, with Manel Lopez, this former molecular modeler from Almirall. ⁓ in a square, in a tapas place in Barcelona, and we had a beer together. And yeah, I mean, we started interacting with him, and brought us, or he teaching us what drug discovery is, because Enrik Herrero ⁓ and we didn't have any idea on what drug discovery meant. ⁓ not he taught us, or he explained to us what drug discovery. ⁓ was, but what drug discovery in the industry is and who is who. When we started iterating with Manel, Manel also said we would need to have someone with a solid and strong scientific ⁓ background from academia. And he presented us to Professor Javier Luque. He's a principal investigator at the University of Barcelona. ⁓ his areas of expertise are quantum mechanics. So all these ⁓ accurate algorithms that describe molecular interactions at the atomic and subatomic level. So we went and visited him at the university. At the very beginning, he said, ⁓ yeah, yeah, these two guys, these two engineers, these two freaks, that they think that they can revolutionize drug discovery. ⁓ I'm going to test them. He didn't explicitly mention that and I need to have a conversation with Javier if he listens to the podcast because I felt that he put us... ⁓ he said let's collaborate for six months. We are working on this publication, jump in as co-authors but ⁓ you need to some work. It was not as explicit as this but I could sense it. And yeah, after six months, he saw that we were serious, that we were professionals and that we were good people or good guys or people that was worth to have ⁓ interactions. So then we ⁓ incorporated Javier Lucas as CSO of the company. So that was the foundational time of PharmaCelera. Two computer engineers with a background in artificial intelligence and high performance computing. a drug hunter from industry that had applied modeling for 20 years ⁓ in the industry, let's call it real projects, ⁓ and then ⁓ an academic that brought the scientific excellence into the project. So serendipity. you dig little bit about ⁓ not only stories of companies, but in your personal life, mean, how many situations are just the outcome of serendipity and casualty and just ⁓ random interactions, right? ⁓ And you realize yeah, randomness and serendipity, they play an important role ⁓ in life. ⁓
Imants Zudans: What was the discussion, who would be the founders how would contribute potentially? It's your very diverse I assume that professor probably want to leave his very hard earned professor position. How did the team actually form?
Enric Gibert: with respect to positions, are right. Professor Javier Luque didn't to take the risk losing his academic he continued. ⁓ And he continues to be ⁓ a investigator at the University of Barcelona. ⁓ So we knew that his implication would be part time. At the same time, López was also looking a job. He needed a full-time job, so he not either to jump full-time to Pharmacelera because we didn't have a budget, didn't have resources. So he started in project as an position and being with us it was clear that he would not have by day duties in project because he needed a job at that point in time. So he would us. And then it was Enrik Herrero and myself as full-time employees with no salary. And at the very beginning, it was curious because, I I mentioned that we both knew each other from Intel, but we had never worked together. So there was a period in which we had to know each other. And at the very beginning, when we were... pitching the idea or when we were pitching the project, would step on each other. You see when we were going that was ⁓ saying ⁓ and would say A prime. ⁓ I would say A prime at that point in time, we participated in an acceleration program organized by the Mobile World Capital Barcelona, which is the an entity that tries to promote that the Mobile World Congress stays in Barcelona, technologies, including health. ⁓ we selected for program there, we were coached. ⁓ part ⁓ of the was ⁓ test. Why did you decide create a farmacellera or a project this one, what you there and what would you like do there. those coaching were extremely because we saw first that both Enrik Herrero and myself, we were aligned. We were not there for the money. Of course, mean, at the end of the day, money is important, but we were not there for the sake becoming rich, if I become So be it, I will not complain. But were not there either egos or for publicity. We were there for the challenge because we both thought without discussing it ⁓ us, that our motivation to a company like Pharmacelera was because we saw that there was a challenge and we love challenges. And in addition, during those sessions, became... more apparent that his role because of his personality, skills and his capacities put me on ⁓ a CTO and my role would be more Those coaching sessions were enlightening. And then, it was who was going to what, then I decided to do some... in management in order to complement ⁓
Imants Zudans: what surprised you ⁓ most about drug discovery?
Enric Gibert: What surprised me most about chemistry? maybe it would be similar if I explained to you how a processor is designed. So maybe, Imans, think, OK, a processor is just a bunch of 20, 30 engineers. working in a garage and no, no, no, it's humongous effort, ⁓ different departments, different duties. So is that also struck me drug discovery. Drug discovery, I thought that there were just chemists doing some experiments and you do tests, but I didn't realize of the of effort that is ⁓ behind ⁓ a drug discovery project. So 10,000 feet or from the distance you see, ⁓ I mean, ⁓ is important, this is challenging, ⁓ ⁓ a few chemists doing ⁓ X, Z. ⁓ But once you dig into it, you see or you start understanding the complexity. Similar with designing processors. I mean, say, OK, the contact point is this one? No, no, no. I mean, there's a plethora of groups coordinating and ⁓ or moving. the project from one stage to the other. So that was something struck drug discovery.
Imants Zudans: How did you decide what is the first application that you are going to make? Was there some scientist involved that said, you know, we need program that would do this? was decision made?
Enric Gibert: my area of expertise and Enrik Herrero's area expertise. So wanted to build something that was self-contained, ⁓ hardware machine, software installed on it, the software. Who was that decision made? this regard, took advantage Professor Javier Luque, PI from the University of Barcelona who became our CSO, was an expert quantum mechanics and quantum mechanics-based algorithms. So we decided to use this expertise to develop an engine order to very integrated with the hardware, with GPUs or even FPGA, Field Gate Programmable Arrays, which is that can be reconfigured, because then you build that is ad-hoc to that particular algorithm. ⁓ And we set time. ⁓ Yes. simulate molecular interactions with these QM-based ⁓ algorithms, with these accurate algorithms. And we chose virtual screening as our first use case. there, into I see something that if I had again, would try to avoid
Imants Zudans: And what is the advantage of that? Is it much faster? How much faster is it?
Enric Gibert: can be a hundred times faster, a times faster. Yeah, yeah.
Imants Zudans: Wow! So kind of build only ⁓ algorithm for virtual but actually have a server is tuned do that.
Enric Gibert: as an error, which is, have these algorithms, these QM-based algorithms for ⁓ screening. What I, Enric Givert and colleague, Enric Carrero, what do we like? We like hardware accelerators. So we said, going to build pharma box a piece of based on Yes, yes, absolutely. And idea was, we will sell this combined product, hardware plus software, to our customers, which will big pharmas and biotechs.
Imants Zudans: out of curiosity, how big the box is actually?
Enric Gibert: hardware accelerators or GPUs or even FPGAs. the idea was that it would be like a fancy workstation. It didn't have to be like big ⁓ HPC No, no, similar to workstation that today has GPUs. The idea was, instead of GPUs that are very well for specific ⁓ types of calculations, let's put a very strong. FPGA, which in size is similar to a GPU, and let's reprogram that FPGA to run our algorithms 100 or so three or four orders of magnitude faster than a traditional CPU. It sounded a scientific perspective super interesting, very aligned with our expertise, but there we of two potential issues or of roadblocks. ⁓ is in order to have hardware that is reprogrammed to run your software, your software needs be very stable. You must have a that is very stable. And at that point in time, the software was evolving very rapidly. We were having prototypes and and it here, tweaking it there, doing changes here, doing changes into the of molecules. So the software was evolving very So it was very hard to have a version very stable to map to hardware. And the second was that if we wanted to sell this, we would need not only to convince the chemists, the scientists, but we would also need convince the IT department. of the pharma. probably they would already have their own hardware approved list of ⁓ hardware providers. So ⁓ here comes company that says, OK, no, no, you need to buy these machines because this has all software ⁓ integrated. So we saw that that an additional complexity into the discussions. ⁓ And that point in cloud computing was becoming ubiquitous and although pharma embraced cloud at a pace than other segments, it was becoming that we didn't want to fight with ⁓ our own So dismissed the pharma But it was neat exercise. It was very interesting from a scientific perspective, but not very interesting from a business perspective, I think.
Imants Zudans: So do you think it's difficult sell or algorithm improvements would outpace the advantages that hardware solution
Enric Gibert: I think it's both. the latter, obviously, I mean, the software was evolving and is still evolving. And we're now expanding our applicability of QM to ⁓ other beyond and base virtual So the algorithm was evolving and is evolving faster the hardware can be. ⁓ And in addition, we realized of something. Me and Herrero as computer engineers at Intel, trying to optimize. we love accelerating and running software faster. But at end of the day, the value is in the outcome of the algorithms. value is on the molecules that are proposed by the algorithms. if you ask ⁓ a drug what you prefer? Let's give a drug discovery project. Do you to have? mediocre or molecules that are not very novel or that are not very diverse for your drug discovery project in a few minutes? Or is it okay for you to spend some hours, even doing more accurate simulations and get more relevant starting points? Molecules that are more novel, that are more diverse, that ⁓ allows to explore different scaffolds in parallel so that if one at the end of the day doesn't have the desired physicochemical properties, you can go into ⁓ other direction. I think that the value is on the latter. course, if ⁓ tell ⁓ drug hunter that he or she needs to six of simulations, he or she will say, OK, I will with something that faster and I can iterate several times. ⁓ But think that you are able to design technology that gives responses to molecular modelers in a reasonable amount of time, the value is on the results. The value is on the assets. The value is on the molecules that the algorithms propose. that also made us put ⁓ all focus on, OK, let's forget about the speed. Speed, of course, is important. We don't want a simulation to take ⁓ days or months. ⁓ but let's concentrate on the value that the modeler or the medicinal chemist will get.
Imants Zudans: Okay, so you essentially... Pivoted Right? You kind of had this idea about pharma box. I still find it very fascinating as a chemist. ⁓ two three orders of magnitude improvement sounds very relevant overall. you realize that's not the end solution Do you have clients ⁓ at this time ⁓ or you are working with academic partners?
Enric Gibert: At that point in time, in fact, we pivoted, as you said, and the decision was fast. mean, maybe we explored the option of the Pharma box, I say, for a year or something like that. Not more that. We just realized that it didn't make quite fast. We didn't have customers.
Imants Zudans: And you described to me that this is like deciding to cross the desert, right? Before crossing, that would be And then there is zone in front of you. You have to survive yourself, ⁓ family. That's a bold journey. You had to convince co-founders to on it. But also family. How did you them?
Enric Gibert: Mm-hmm. Yeah, yeah, I mean the early days are tough, okay, because you come from a comfort zone working for a multinational with a good salary to some extent, okay, so and then suddenly you enter into this roller coaster, okay, that ⁓ I always say when someone asks me, okay, but haven't you or don't you... sometimes or have doubts about whether should have moved to a big company or pursue your dreams and I joke, I said yeah, yeah, have, I make this question to myself 10 times per day ⁓ at least you can imagine that it's roller okay? On the other hand, mean having your, let's call your own destiny on your hands or or deciding yourself where you ⁓ to go, that's very rewarding. So here at home, support, it's tough, ⁓ we give you period in which you can okay, perfect. But then... ⁓
Imants Zudans: Did you discuss how long that period would be?
Enric Gibert: no, ⁓ no, no, no, no, no, And I to discuss ⁓ the list possible things from the company or that to do with Farmacelera at home. You don't want to mix both okay? You don't want to have your family your other board of directors member, okay? No,
Imants Zudans: So how did you get the first
Enric Gibert: How did we get the first client? the Barcelona Supercomputing Center. They had some algorithms drug They were complementary what we were doing. But they wanted to accelerate those algorithms. they subcontracted us in order to accelerate ⁓ their algorithms. So that like want to decouple ⁓ ⁓
Imants Zudans: but they are with you on a journey, you know, essentially. you know, it might not end up that, you know, it ends up with nothing. So I guess that, you know, I don't want to drill you too much on this topic, but think that younger founders, they of course see the success stories how...
Enric Gibert: Hehe. You service project that had nothing to do with our core engine, but that's our first source of income. Our second source of income was an academic group we were collaborating with. It was not the UBI, it was not the University of Barcelona, early we did benchmarking. Since we came, both Enrik and myself, we came from
Imants Zudans: you started something in couple years, revenue, investment, etc. But most of the times, as I see, it actually takes at least as long as they thought. And that's when the family starts to ask questions, you you said, you know, this going to be couple years and it's been five years and ⁓ affects them too.
Enric Gibert: from benchmarking processors or benchmarking potential processors that we were designing ⁓ to apples comparisons and doing this type ⁓ analysis. So one the first ⁓ tests that we did of our algorithms for, in the case of screening, was this ⁓ academic institution. some data two or three data sets. They small. And they said, ⁓ Yeah, yeah, yeah, yeah, absolutely. a roller coaster from personal perspective. I was the early 40s. I two daughters at that point in time. OK, so it's the same of doing or starting a company when you are 25 or 27 and you have less duties or less links ⁓ than when you are 40 something. Why don't we run these data sets in which we know molecules are active and which molecules are ⁓ active, OK? And just compare them to ⁓ our methods that they were using software from a software vendor, OK? And why don't we do a direct apples to apples comparison, OK? So we did that comparison, and the results very, good, OK? And so they were. At the same time, you have a little bit more experience and you know to balance a little bit the pros and cons. At end of the day, when I that I didn't want my family to be like another board member of is ⁓ of course, ⁓ discussed progresses, okay? the family saw that ⁓ the company was evolving, that were getting... very interested in a license of our software, okay, and that was one our first customers. Of course, that was an academic license. So you know that ⁓ in an academic institution, budgets are more restricted. So it was not a sale, it was our first software sale, and it was based on benchmarking. After that, we ⁓ we naive again, okay, and we like grants that we were getting investment from investors. So two years, would say that the family okay, this is to some extent safe. It's not Enrik now has gone with the crisis, with the year old crisis, no, no, okay. So ⁓ they the progress. ⁓ If you have benchmarking and you show that your solution is better, will come to us. Show me the data. Here's the data. And ⁓ realized afterwards that ⁓ solution is only, is This is % better, this is three times better, or even this is five times better, or 10. It needs to be more than that. Usability, easy ⁓ access, trust, not only benchmarking. ⁓ But yeah, ⁓ our second was based on a benchmarking study. we said, are the kings, we just need to show. ⁓ that our algorithms are more accurate, our algorithms are better, and people will come to us. No, no, no. does not work like that.
Imants Zudans: you took an idea is very common in world. You benchmarking and brought that to chemistry
Enric Gibert: I that benchmarking and not only in drug discovery, ⁓ we it also when we working the high-performance computing ⁓ community. Benchmarking per is ⁓ gigantic task, because, and full ⁓ issues, okay, and of debates. What do you choose as the baseline? What data sets you are using? What configurations you run? What of statistics you report? it's something can lot of controversy ⁓ and it can even become religious discussion. ⁓ So we were ⁓ reading the other ⁓ ⁓ opposed by Pat Walters complaining about the DUTE for learning. So we to go through all this There's people that report hitrate ⁓ and their main ⁓ metric ⁓ hitrate. This not good. in the that I can have 100 percent hitrate if I want. I prefer having ⁓ a much lower ⁓ hitrate but having two different scaffolds or two different families of molecules, rather than having all ⁓ the molecules that I test experimentally being active, but all of them being just small modifications of the same family. So you always have ⁓ type of questions. Have you taken novelty? Have you taken diversity into the equation? ⁓ are not designing hits we are designing drugs. ⁓ So it's a multi-parameter optimization. ⁓ So yeah, mean doing benchmarking is more than a science and you need to be quite open about it in the sense that, okay, here are the scenarios in which the technology works well because XYZ, And these are the scenarios in which the technology does not work well because ABC, There's no... one size all ⁓ in drug discovery. Every drug discovery project is different. So sometimes ⁓ you a different configuration of ⁓ a workflow ⁓ order succeed. And having always in mind that you are designing drugs, you are not designing ⁓ hits. So benchmarking tough. ⁓
Imants Zudans: customers, did look for customers or was this benchmarking study helping you? They were kind reading the paper and out to you.
Enric Gibert: No, we were looking for customers. We started attending some conferences. attended BioEurope, Bio Convention. ⁓ These are conferences, but I don't think that these are the best conferences to find the scientists. We still go BioEurope and Bio. Don't get me wrong. But if you want find you need to go to the medicinal chemist and the, where medicinal chemists and computational chemists okay? So you need to go to scientific At the end of the day, in our industry, you need to convince ⁓ by the and by the technology ⁓ and the that you obtain with that science and that technology. ⁓ The conversations always very scientific. So even the decider, ⁓ is a chemist or a biologist that he or she has decided to take like a managerial track a company, ⁓ but normally the deciders have a PhD in medicinal chemistry, in organic chemistry, ⁓ in biology in computational chemistry, okay, so they are PhDs. So you need to go to forums, okay, at the very beginning we went more to these business development forums that I don't think that are the correct audience. when you are a startup. Afterwards, yes, when you are fundraising, when you want to establish partnerships, you are ⁓ for more collaborative ⁓ alliances, ⁓ BioEurope and Bio are very good. But if you want to ⁓ ⁓ customers of slash AI tools, you need to go to the scientific events where the ComCams and the... ⁓ the medicinal chemists are going. it took us a while to realize that. It took us a while.
Imants Zudans: somewhere here have a book that says selling to scientists ⁓ is biggest challenge in sales. Scientists by nature are skeptical, critical, and they're also very smart. So I'm wondering how did go about that?
Enric Gibert: Yes True true true true. Yeah. No, we realized very early in the project that a Conversation ⁓ ⁓ sales conversation Is always driven by the science. Okay, and a scientific decision even ⁓ if if talking to a biotech the drug hunter or the medicinal chemist or the computational chemist the scientist and the decider at the same time. If you're talking to a big pharma, you to convince the science, the hunter the medicinal chemist, but you need also to convince ⁓ the manager. normally, the deciders also people that are scientists, so that they have a PhD in chemistry, organic chemistry, or biology. And the conversation still is scientific. ⁓ So we saw from the very beginning that we were not selling pencils. ⁓ We were selling sophisticated R &D software people that are skeptical, critical, and ⁓ opinionated. ⁓ So the that we incorporated to our business development efforts were mainly scientists had ⁓ or that skills and can build relationships. You cannot bring in persons that do not understand the science or that ⁓ have not worked ⁓ drug discovery ⁓ this is by the ⁓ purchaser or by buyer which is a scientist.
Imants Zudans: you were enough to realize this before starting selling or was this discovered in the process.
Enric Gibert: It was discovered in the process. At the very beginning, I mentioned, one of the learnings was, OK, we have the benchmarking. Now I need to go to, or they will come to me. Or when I just plot ⁓ or I the plot graphs in ⁓ a conference, will just say, ⁓ yes, yes, let's move forward. I remember ⁓ my first. Bio International Conference that I attended in San Diego in 2017. don't remember, we had 20 meetings. When I came back, said, OK, of these 20 meetings, I think that have opportunities, so 10 customers. And we got zero. I was or I am an engineer. So I was talking to scientists. that the time that I was realizing the drug discovery was very big. ⁓ And also realized that it's not only about benchmarking. Okay, it's also about having cases. Who have you worked with? Okay, where have you this? And all these how the software is. Okay, that was a process that took some time. Okay, to realize that we had to have scientists ⁓ that had communication and social skills so that they can engage on collaborations, on proof of concepts, on establishing conversation that understand the science and that can raise the interest. Yeah, of course. If a person that is interested, then we will bring the persons that are developing the software. ⁓ And that can explain you the very last argument ⁓ or ⁓ that want to. that you want to test, but at least people in business development, they need to understand the algorithms do and the outcome of the algorithms.
Imants Zudans: So now in your team you have ⁓ a lot of scientists, right?
Enric Gibert: Yeah, mean 75 % of our team has a PhD in chemistry or organic chemistry or computer science as we do. have several industrial PhDs, has also been a source of talent attraction and incorporation into the company. This also puts us at the forefront of science. So mean, very, very science first and a very, very science-oriented company.
Imants Zudans: Was there ever a moment you thought that this is difficult, moment where you came close to ⁓
Enric Gibert: Not really, mean the first years as we commented was traversing the desert but once started having traction with recurring including some big pharmas we had the sensation that we're not going to make I would say after first two or three years, we had a clear that, OK, is doable, OK, and this is going to work out.
Imants Zudans: When did you start to feel that selling becomes easier? That it's not so much struggle anymore when it is easier to get the next ⁓ client.
Enric Gibert: a ⁓ I have ⁓ really good answer for. I guess when you start seeing that... at the end of the day, confidence. Creating confidence and seeing that you have created confidence in the users or in the customers. ⁓ a project with biotech in the UK that was working on ⁓ what was... I don't now, on fibrosis, yes, ⁓ on fibrosis, ⁓ And ⁓ did, think from perspective, we did a pretty... the first software users renew the license, then you start seeing, this is recurrent revenue. So there is a market. There is a need. We're providing a solution that ⁓ is valued by users, and not because they have tried once, but they are renewing. ⁓ That's an important ⁓ nice piece of work because results were very encouraging but at that point in time that biotech company or their investors, their shareholders decided to put all the efforts in clinical stage assets and discontinue drug discovery although the work was good they decided to discontinue drug discovery ⁓ the person was running ⁓ was the CSO Another important milestone is, we as a company, provide access to software, okay, so we sell software licenses. On the other hand, we also provide services, okay, and we thought, okay, the traditional theory, okay, software is like 90 % recurrent revenue. In services, you need to chase ⁓ every single person. No, when we started seeing recurrences in services as well, that meant of this company was Macrophage Pharma. name is ⁓ Reddy. He moved to Inception. This company from San Francisco and he became CSO of General Inception. And other means, we had our slide deck through other friends to General Inception. So when joined General Inception as a... that ⁓ in previous projects with that same customer, we delivered. So they were repeating with us. That creates comfort. If you have both users of the software renewing, this means that it's providing answers or ⁓ results that meaningful for them. And this is why they want to ⁓ continue engaging you. ⁓ And they you even ⁓ for some and functionalities. A CSO, one of his first duties was, okay Venkat, we have received this slide deck, okay. ⁓ This company from Barcelona, it seems that it's cool technology, but we don't have any way to assess whether are good or not. I mean, please assess they are. And Venkat ⁓ no, yeah, yeah. ⁓ I know Yeah, serendipity again. mean, I know them, I work with them. that so they are interested on expanding the usage of the technology and also recurrent in services. When we started seeing ⁓ customers repeating with us and ⁓ subcontracting project number two, project number three us, that also creates a lot of comfort. that's a point in which
Imants Zudans: Serendipity again.
Enric Gibert: It's never easy, you can never relax, but it's a point in which you say, okay, I mean, the market is responding, I mean, we're providing something that is meaningful to the drug discovery industry. technology is brilliant, the guys are ⁓ good, you can trust them. then we ⁓ a strategic alliance with General Inception, ⁓ became shareholders of our company and are their artificial intelligence drug discovery strategic partner. So we support for companies in their portfolio. So this serendipity, the is very small ⁓ and suddenly someone we had worked in the past and was happy with the technology and with our work.
Imants Zudans: When was that for you?
Enric Gibert: For me that was ⁓ four or five years ago. Yes. became the CSO of this company, Gniter, and then that brought a new partner ⁓ on the table. That a lot ⁓ of confidence that the project will move forward.
Imants Zudans: So approximately at the start of COVID, something like that, right?
Enric Gibert: Yeah, yeah.
Imants Zudans: Do you miss on chips?
Enric Gibert: ⁓ some extent, yes, ⁓ a researcher. It's my passion. I've always ⁓ to work chip design ⁓ since I that particular subject ⁓ during ⁓ engineering ⁓ I was very fortunate because at that point in time there were not many ⁓ chip design companies in Spain. There was like a big R &D group at the UPC, the Polytechnic University of Catalonia. And I was fortunate to ⁓ land there, OK, because that opened the opportunities everywhere. And then Intel came to Barcelona. I miss it because that's my background. ⁓ Similar to I miss programming, I miss doing research on that particular topic. But I just chose, also because of my personality and the paths of life, just chose to move ahead in this managerial role. But people at Pharmacellera, mean the people that are doing the algorithms, they know that when we have a one-to-one conversation or when we go for lunch, I love hearing them telling me that they have tried the hydrophobic profile like this way. And I get into the details of the algorithms, into the details of how the algorithms runs, what is the benchmarking, what is the execution time, what are the hardware requirements. So I will always have this. in my heart, okay? But it is true that I have just chosen to move ahead in the management career path.
Imants Zudans: Okay, so it is 22 it is ⁓ peak hype for Biotech. There is a lot of money ⁓ being thrown in Biotech. ⁓ You are revenue, you have clients, ⁓ you've that your algorithms, your tools are useful, they ⁓ drug discovery. Did consider exit at that time?
Enric Gibert: Well, mean, 10 years is a long time, so we... We had conversations, ⁓ both led by us and led by others that have approached us, companies that are working on the similar space but have complementary technology. ⁓ So we could have something bigger and with a grown customer base. companies not even having computational, ⁓ more ⁓ ⁓ lab capabilities and lab skills that could complement by some sort of ⁓ artificial intelligence and quantum mechanics machinery in order to complement those. Yeah, have had several conversations. mean, ⁓ are open ⁓ this regard to explore not only collaboration or strategic alliances, but even joint ventures. ⁓ are open to that. ⁓ is true that it's not easy ⁓ when you join forces not as a partner that we have several already. But if you join forces more on a joint venture with another organization, there must be a very clear alignment on the culture, on the goal where you want to reach, and on the chemistry of the teams. So it's not easy. So long story short, yeah, we've had in these last 10 years ⁓ conversations
Imants Zudans: ⁓ do you get feedback from, from users?
Enric Gibert: ⁓ in order to do something bigger and or to join forces with other organizations both in Spain and outside of Spain yeah
Imants Zudans: But you decided to go still on your own, right?
Enric Gibert: Yeah, no. An interesting conversation, for example, was with a big pharma company. And computational he was the simulations, and he was handing the results ⁓ to the chemist. And ⁓ the results, the prioritized by the platform, he was handing that ⁓ to the using smiles ⁓ or a ⁓ structure of the molecule. ⁓ And he well, I mean, The molecules that the technology is proposing are so novel and are so diverse that medicinal chemists are challenging me about the decisions that I do using your technology. ⁓ And then I need to show them for some specific scaffolds that they say, ⁓ no, no, let's not go this route. Why the system chose it? once you see the interactions in 3D and the hydrophobic profile of the molecule, then they say, ⁓ now I understand why this molecule was prioritized ⁓ over others that from a medicinal chemist or from a 2D perspective made more sense. also you need to make ⁓ that move medicinal chemists out of their comfort zone ⁓ and known chemistry. ⁓ It took us three years to get this type of feedback. then we to release a graphical user interface in which you could in 3D very well ⁓ why the was proposing these molecules. And this type of feedback, it took us three years to get it. But once we got it, that is very valuable. ⁓
Imants Zudans: So first 10 years are behind you. What is in front of you? What are the next five years?
Enric Gibert: So I think that after the first generation of artificial intelligence drug discovery companies fell short with respect to the expectations. I don't want to say that they failed. They didn't fail. I artificial intelligence now is super important and will continue being super important for drug discovery. But after they didn't meet the expectations, most of them, the industry is paying A lot of attention on the combination of ⁓ artificial intelligence slash machine learning with physics-based simulations. Every drug discovery project is different. So you some sort of physics-based ⁓ methods in order to characterize that particular system that you are trying to design drugs, not hits, for. ⁓ In this regard, mean, a pharmaceutical is well positioned, because we apply quantum mechanics, so physics-based simulations combined with artificial intelligence. And now, in addition, there is all these ⁓ exponentially growing chemical spaces, these ultra-large ⁓ chemical spaces that can be synthesized and tested rapidly. is ⁓ important ⁓ being able to test hypotheses a hit ID stage. very rapidly and very cheaply and based on this data start working on some candidates and then spend more money on ⁓ optimizing the properties of those particular ⁓ families that you have selected. So in our particular case we have the foundations, we have the engine, now it's a matter of ⁓ applying it beyond HitID. We have been... working a lot the hit ID stage, but we also have success cases in the hit to hit and hit optimization stages. ⁓ applying our engine in these ⁓ additional ⁓ discovery stages. And course, mean, working on new modalities molecular glues we just incorporated ⁓ an application scientist has experience ⁓ in applying computational for protax and molecular glues. ⁓ So all these type of new moralities and new type of targets, this is an area in which the company can also grow with additional success cases.
Imants Zudans: 10 years, ⁓ congratulations. I see there is very significant chapter in front you. So ⁓ I wish you success ⁓ and endurance. And maybe we'll have you back a year or two ⁓ see ⁓ the progress. So thank you, Enric, for joining me. was Enric. ⁓
Enric Gibert: Perfect. Thank you for inviting me. It was a privilege. And yeah, let's see if we can talk again in two years and see where we were at at that point in time. That would be interesting.
Imants Zudans: That was Enric Giber, CEO of ParmaCellera. I'm Imants Zudans. Thank you for listening.
Imants Zudans: Enric, thank you for joining us here today. Maybe let's start your story with Intel
Enric Gibert: the PhD in these fields at the University of Polytechnic of Catalonia here in Barcelona, I live. ⁓ But during the PhD, I did a research stay at the University of Illinois in Urbana-Champaign in the US. And doing the I joined Intel. Intel at that point in time ⁓ had opened a research lab ⁓ in Barcelona that was very near to the UPC. And it was, in fact, Directed my PhD advisor, ⁓ ⁓ Antonio González. So it was a natural step for me. ⁓ I stayed there for 10 years and what we were doing there was design the processors, the chips, the hardware accelerators ⁓ at a 5-10 year time frame. ⁓ So ⁓ when need to design what is going to be the microprocessor or the GPU or the hardware accelerator ⁓ the Polytechnic University of Catalonia, there was a big research group on computer architecture. And the group had a pretty strong reputation. So Intel decided to open an R &D facility that was 2002, 2003. for this time window, five, ten years now, you don't have the hardware, so you to start doing some sort of simulations and modeling benchmarking against current products and is we were doing. Nothing do with life we were designing chips and the hardware accelerators of the future and idea was a research lab but the idea was to impact the product lines of
Imants Zudans: So you spend a decade at Intel and then it was in 2014 that Intel decided to close Intel Barcelona site, right? What happened? Welcome to Imants podcast, where founders powering drug discovery share how they built their companies.
Enric Gibert: The lab began by just eight people and over 10 years. it reached a peak of 80 people, 60 of them a PhD. it was like a small research site compared to other sites at Intel. mean, 60, 80 people, we're collaborating a lot.
Imants Zudans: My guest today is Enric Gibert, CEO of Pharmacelera Enric spent a decade designing processors at Intel. Then he started building software for drug hunters. How does a chip engineer end up in medicinal chemistry?
Enric Gibert: with the people in Santa Clara, the headquarters of Intel are. We were also collaborating a lot with the people in Oregon, and and ⁓ of these unfortunate in which a big multinational ⁓ had promised investors that they would ⁓ have sales of 50 billion. ⁓ they didn't reach that. They got like 49.5 or 49.7. I don't remember. So they felt short with respect to expectations. then from a management perspective, they decided to start closing small and mid-size sites. They closed one in Mexico. they closed one in Germany that was more or less the similar size of us, 100 persons, and then they closed ⁓ ours and concentrated, and they decided to concentrate all the R &D efforts in bigger sites.
Imants Zudans: That's what we explore today. that sounds very exciting, very interesting.
Enric Gibert: as you say, it sounds a lot of fun. We had a blast. I had the privilege to work with people that were super smart, super inspiring. OK, and that I saw that was extremely important to keep me motivated. My mom always said that while I was at Intel, There was not a single day in which I complained when going to work. ⁓ Because we were having fun. mean ⁓ story is a lesson learned ⁓ ⁓ seen ⁓ companies ⁓ big companies in which ⁓ they need to reduce headcount or they need to do some ⁓ You start hearing rumors, okay? ⁓ spend like six or nine months ⁓ asking yourself, is it going to be me? I mean, and then ⁓ just... lower your performance, your engagement because you don't know how you, your group, your site is going to be affected. I've seen that ⁓ happening in other situations. This was the opposite. ⁓ And it has pros and cons, but I remember there was a Monday in which the head of the lab in Barcelona was discussing with his boss in Santa Clara the budget for that particular year. ⁓ That was on a Monday. And they were discussing what projects we would be running for that particular year, what was going to be the budget for the entire lab in Barcelona that year. That was on a Monday. This same boss landed on Wednesday, so 48 hours afterwards, to say that the site was closing, ⁓ that we were laid off. ⁓ So it was like ⁓ immediate action, no rumors. ⁓ And in two months we were gone, all of us. ⁓ So this started to think, if this happens again to me, what would I prefer? Would I prefer to know it with six or nine months in advance and start thinking, ⁓ my god, ⁓ my god, or I would like to have a very short notice? Of course, mean Intel. was in good shape financially, so we got a good compensation package. when explained the story, a lot of people tell me, they behaved bad. I said, well, I mean, was straight, period. no pain. that point in time, since we were notified, I think it was like two months or like that. everybody started looking for opportunities. and this is another interesting ⁓ story. that that were said that we would be laid off. three or four days afterwards, ARM the chip designer of chips for mobile phones, that they are based in Cambridge in the UK, landed in Barcelona with all our profiles from LinkedIn, and they started interviewing. So that was one of the best talent acquisition exercises I have ever seen. In 48 hours, 72 hours. the managers of ARM were in Barcelona all the team. So lot of people went to ARM, to Cambridge. I had the opportunity to go to Cambridge. Some people moved to the to the Silicon I also had an offer to stay at Intel. And some other people just wanted to look for other ⁓ alternatives. In particular case, yeah, I had been for years in a multinational. So I knew what was for big company ⁓ But my was for something ⁓ I had a blast at of course. but I still had a lot of and I said if I don't try now I will not I will never try it yeah together Enrik Herrero we decided explore opportunities in the life sciences and when we decided to Pharmacelera
Imants Zudans: discovery, was that a deliberate or were you considering other options as well?
Enric Gibert: was serendipity as things that happen in life. ⁓ Together with Enrik Herrero, we were colleagues Intel, although it is true we had never worked together. We knew of each other. We were in different projects. We had never coincided in the same project. So it took us a while to get to know ⁓ each other. And we common friends. We looked at ⁓ genomics, ⁓ fascinating from a research perspective. You must also understand our background. Both Enrik Herrero and myself, we are PhDs on computer engineers. So we are scientists, we are researchers. working on genomics, we saw that it was super exciting, super challenging. It is true that our first interactions ⁓ we saw that it was still a very immature field. So we would have had a lot of fun because it's an exciting topic. But most of the money was ⁓ grants, it was public funding. So it's hard, it's tough from a business perspective. So we ended up discarding genomics. And at that point in time, for these situations in life in which I was taking my elder daughter to school and ⁓ a mother of a boy that was attending the same class, she was working, or she is still working at Almirall, the largest pharmaceutical company in Spain, and she told me, Enric. There is an interesting profile that was allowed to let go at Almirall because at that point in time Almirall was ⁓ reshifting their focus. There's a person that has worked at Almirall on for 20 years. ⁓ brilliant. ⁓ should talk to him. ⁓ I'll do the introduction. ⁓ we met, ⁓ Herrero and myself, with Manel Lopez, this former molecular modeler from Almirall. ⁓ in a square, in a tapas place in Barcelona, and we had a beer together. And yeah, I mean, we started interacting with him, and brought us, or he teaching us what drug discovery is, because Enrik Herrero ⁓ and we didn't have any idea on what drug discovery meant. ⁓ not he taught us, or he explained to us what drug discovery. ⁓ was, but what drug discovery in the industry is and who is who. When we started iterating with Manel, Manel also said we would need to have someone with a solid and strong scientific ⁓ background from academia. And he presented us to Professor Javier Luque. He's a principal investigator at the University of Barcelona. ⁓ his areas of expertise are quantum mechanics. So all these ⁓ accurate algorithms that describe molecular interactions at the atomic and subatomic level. So we went and visited him at the university. At the very beginning, he said, ⁓ yeah, yeah, these two guys, these two engineers, these two freaks, that they think that they can revolutionize drug discovery. ⁓ I'm going to test them. He didn't explicitly mention that and I need to have a conversation with Javier if he listens to the podcast because I felt that he put us... ⁓ he said let's collaborate for six months. We are working on this publication, jump in as co-authors but ⁓ you need to some work. It was not as explicit as this but I could sense it. And yeah, after six months, he saw that we were serious, that we were professionals and that we were good people or good guys or people that was worth to have ⁓ interactions. So then we ⁓ incorporated Javier Lucas as CSO of the company. So that was the foundational time of PharmaCelera. Two computer engineers with a background in artificial intelligence and high performance computing. a drug hunter from industry that had applied modeling for 20 years ⁓ in the industry, let's call it real projects, ⁓ and then ⁓ an academic that brought the scientific excellence into the project. So serendipity. you dig little bit about ⁓ not only stories of companies, but in your personal life, mean, how many situations are just the outcome of serendipity and casualty and just ⁓ random interactions, right? ⁓ And you realize yeah, randomness and serendipity, they play an important role ⁓ in life. ⁓
Imants Zudans: What was the discussion, who would be the founders how would contribute potentially? It's your very diverse I assume that professor probably want to leave his very hard earned professor position. How did the team actually form?
Enric Gibert: with respect to positions, are right. Professor Javier Luque didn't to take the risk losing his academic he continued. ⁓ And he continues to be ⁓ a investigator at the University of Barcelona. ⁓ So we knew that his implication would be part time. At the same time, López was also looking a job. He needed a full-time job, so he not either to jump full-time to Pharmacelera because we didn't have a budget, didn't have resources. So he started in project as an position and being with us it was clear that he would not have by day duties in project because he needed a job at that point in time. So he would us. And then it was Enrik Herrero and myself as full-time employees with no salary. And at the very beginning, it was curious because, I I mentioned that we both knew each other from Intel, but we had never worked together. So there was a period in which we had to know each other. And at the very beginning, when we were... pitching the idea or when we were pitching the project, would step on each other. You see when we were going that was ⁓ saying ⁓ and would say A prime. ⁓ I would say A prime at that point in time, we participated in an acceleration program organized by the Mobile World Capital Barcelona, which is the an entity that tries to promote that the Mobile World Congress stays in Barcelona, technologies, including health. ⁓ we selected for program there, we were coached. ⁓ part ⁓ of the was ⁓ test. Why did you decide create a farmacellera or a project this one, what you there and what would you like do there. those coaching were extremely because we saw first that both Enrik Herrero and myself, we were aligned. We were not there for the money. Of course, mean, at the end of the day, money is important, but we were not there for the sake becoming rich, if I become So be it, I will not complain. But were not there either egos or for publicity. We were there for the challenge because we both thought without discussing it ⁓ us, that our motivation to a company like Pharmacelera was because we saw that there was a challenge and we love challenges. And in addition, during those sessions, became... more apparent that his role because of his personality, skills and his capacities put me on ⁓ a CTO and my role would be more Those coaching sessions were enlightening. And then, it was who was going to what, then I decided to do some... in management in order to complement ⁓
Imants Zudans: what surprised you ⁓ most about drug discovery?
Enric Gibert: What surprised me most about chemistry? maybe it would be similar if I explained to you how a processor is designed. So maybe, Imans, think, OK, a processor is just a bunch of 20, 30 engineers. working in a garage and no, no, no, it's humongous effort, ⁓ different departments, different duties. So is that also struck me drug discovery. Drug discovery, I thought that there were just chemists doing some experiments and you do tests, but I didn't realize of the of effort that is ⁓ behind ⁓ a drug discovery project. So 10,000 feet or from the distance you see, ⁓ I mean, ⁓ is important, this is challenging, ⁓ ⁓ a few chemists doing ⁓ X, Z. ⁓ But once you dig into it, you see or you start understanding the complexity. Similar with designing processors. I mean, say, OK, the contact point is this one? No, no, no. I mean, there's a plethora of groups coordinating and ⁓ or moving. the project from one stage to the other. So that was something struck drug discovery.
Imants Zudans: How did you decide what is the first application that you are going to make? Was there some scientist involved that said, you know, we need program that would do this? was decision made?
Enric Gibert: my area of expertise and Enrik Herrero's area expertise. So wanted to build something that was self-contained, ⁓ hardware machine, software installed on it, the software. Who was that decision made? this regard, took advantage Professor Javier Luque, PI from the University of Barcelona who became our CSO, was an expert quantum mechanics and quantum mechanics-based algorithms. So we decided to use this expertise to develop an engine order to very integrated with the hardware, with GPUs or even FPGA, Field Gate Programmable Arrays, which is that can be reconfigured, because then you build that is ad-hoc to that particular algorithm. ⁓ And we set time. ⁓ Yes. simulate molecular interactions with these QM-based ⁓ algorithms, with these accurate algorithms. And we chose virtual screening as our first use case. there, into I see something that if I had again, would try to avoid
Imants Zudans: And what is the advantage of that? Is it much faster? How much faster is it?
Enric Gibert: can be a hundred times faster, a times faster. Yeah, yeah.
Imants Zudans: Wow! So kind of build only ⁓ algorithm for virtual but actually have a server is tuned do that.
Enric Gibert: as an error, which is, have these algorithms, these QM-based algorithms for ⁓ screening. What I, Enric Givert and colleague, Enric Carrero, what do we like? We like hardware accelerators. So we said, going to build pharma box a piece of based on Yes, yes, absolutely. And idea was, we will sell this combined product, hardware plus software, to our customers, which will big pharmas and biotechs.
Imants Zudans: out of curiosity, how big the box is actually?
Enric Gibert: hardware accelerators or GPUs or even FPGAs. the idea was that it would be like a fancy workstation. It didn't have to be like big ⁓ HPC No, no, similar to workstation that today has GPUs. The idea was, instead of GPUs that are very well for specific ⁓ types of calculations, let's put a very strong. FPGA, which in size is similar to a GPU, and let's reprogram that FPGA to run our algorithms 100 or so three or four orders of magnitude faster than a traditional CPU. It sounded a scientific perspective super interesting, very aligned with our expertise, but there we of two potential issues or of roadblocks. ⁓ is in order to have hardware that is reprogrammed to run your software, your software needs be very stable. You must have a that is very stable. And at that point in time, the software was evolving very rapidly. We were having prototypes and and it here, tweaking it there, doing changes here, doing changes into the of molecules. So the software was evolving very So it was very hard to have a version very stable to map to hardware. And the second was that if we wanted to sell this, we would need not only to convince the chemists, the scientists, but we would also need convince the IT department. of the pharma. probably they would already have their own hardware approved list of ⁓ hardware providers. So ⁓ here comes company that says, OK, no, no, you need to buy these machines because this has all software ⁓ integrated. So we saw that that an additional complexity into the discussions. ⁓ And that point in cloud computing was becoming ubiquitous and although pharma embraced cloud at a pace than other segments, it was becoming that we didn't want to fight with ⁓ our own So dismissed the pharma But it was neat exercise. It was very interesting from a scientific perspective, but not very interesting from a business perspective, I think.
Imants Zudans: So do you think it's difficult sell or algorithm improvements would outpace the advantages that hardware solution
Enric Gibert: I think it's both. the latter, obviously, I mean, the software was evolving and is still evolving. And we're now expanding our applicability of QM to ⁓ other beyond and base virtual So the algorithm was evolving and is evolving faster the hardware can be. ⁓ And in addition, we realized of something. Me and Herrero as computer engineers at Intel, trying to optimize. we love accelerating and running software faster. But at end of the day, the value is in the outcome of the algorithms. value is on the molecules that are proposed by the algorithms. if you ask ⁓ a drug what you prefer? Let's give a drug discovery project. Do you to have? mediocre or molecules that are not very novel or that are not very diverse for your drug discovery project in a few minutes? Or is it okay for you to spend some hours, even doing more accurate simulations and get more relevant starting points? Molecules that are more novel, that are more diverse, that ⁓ allows to explore different scaffolds in parallel so that if one at the end of the day doesn't have the desired physicochemical properties, you can go into ⁓ other direction. I think that the value is on the latter. course, if ⁓ tell ⁓ drug hunter that he or she needs to six of simulations, he or she will say, OK, I will with something that faster and I can iterate several times. ⁓ But think that you are able to design technology that gives responses to molecular modelers in a reasonable amount of time, the value is on the results. The value is on the assets. The value is on the molecules that the algorithms propose. that also made us put ⁓ all focus on, OK, let's forget about the speed. Speed, of course, is important. We don't want a simulation to take ⁓ days or months. ⁓ but let's concentrate on the value that the modeler or the medicinal chemist will get.
Imants Zudans: Okay, so you essentially... Pivoted Right? You kind of had this idea about pharma box. I still find it very fascinating as a chemist. ⁓ two three orders of magnitude improvement sounds very relevant overall. you realize that's not the end solution Do you have clients ⁓ at this time ⁓ or you are working with academic partners?
Enric Gibert: At that point in time, in fact, we pivoted, as you said, and the decision was fast. mean, maybe we explored the option of the Pharma box, I say, for a year or something like that. Not more that. We just realized that it didn't make quite fast. We didn't have customers.
Imants Zudans: And you described to me that this is like deciding to cross the desert, right? Before crossing, that would be And then there is zone in front of you. You have to survive yourself, ⁓ family. That's a bold journey. You had to convince co-founders to on it. But also family. How did you them?
Enric Gibert: Mm-hmm. Yeah, yeah, I mean the early days are tough, okay, because you come from a comfort zone working for a multinational with a good salary to some extent, okay, so and then suddenly you enter into this roller coaster, okay, that ⁓ I always say when someone asks me, okay, but haven't you or don't you... sometimes or have doubts about whether should have moved to a big company or pursue your dreams and I joke, I said yeah, yeah, have, I make this question to myself 10 times per day ⁓ at least you can imagine that it's roller okay? On the other hand, mean having your, let's call your own destiny on your hands or or deciding yourself where you ⁓ to go, that's very rewarding. So here at home, support, it's tough, ⁓ we give you period in which you can okay, perfect. But then... ⁓
Imants Zudans: Did you discuss how long that period would be?
Enric Gibert: no, ⁓ no, no, no, no, no, And I to discuss ⁓ the list possible things from the company or that to do with Farmacelera at home. You don't want to mix both okay? You don't want to have your family your other board of directors member, okay? No,
Imants Zudans: So how did you get the first
Enric Gibert: How did we get the first client? the Barcelona Supercomputing Center. They had some algorithms drug They were complementary what we were doing. But they wanted to accelerate those algorithms. they subcontracted us in order to accelerate ⁓ their algorithms. So that like want to decouple ⁓ ⁓
Imants Zudans: but they are with you on a journey, you know, essentially. you know, it might not end up that, you know, it ends up with nothing. So I guess that, you know, I don't want to drill you too much on this topic, but think that younger founders, they of course see the success stories how...
Enric Gibert: Hehe. You service project that had nothing to do with our core engine, but that's our first source of income. Our second source of income was an academic group we were collaborating with. It was not the UBI, it was not the University of Barcelona, early we did benchmarking. Since we came, both Enrik and myself, we came from
Imants Zudans: you started something in couple years, revenue, investment, etc. But most of the times, as I see, it actually takes at least as long as they thought. And that's when the family starts to ask questions, you you said, you know, this going to be couple years and it's been five years and ⁓ affects them too.
Enric Gibert: from benchmarking processors or benchmarking potential processors that we were designing ⁓ to apples comparisons and doing this type ⁓ analysis. So one the first ⁓ tests that we did of our algorithms for, in the case of screening, was this ⁓ academic institution. some data two or three data sets. They small. And they said, ⁓ Yeah, yeah, yeah, yeah, absolutely. a roller coaster from personal perspective. I was the early 40s. I two daughters at that point in time. OK, so it's the same of doing or starting a company when you are 25 or 27 and you have less duties or less links ⁓ than when you are 40 something. Why don't we run these data sets in which we know molecules are active and which molecules are ⁓ active, OK? And just compare them to ⁓ our methods that they were using software from a software vendor, OK? And why don't we do a direct apples to apples comparison, OK? So we did that comparison, and the results very, good, OK? And so they were. At the same time, you have a little bit more experience and you know to balance a little bit the pros and cons. At end of the day, when I that I didn't want my family to be like another board member of is ⁓ of course, ⁓ discussed progresses, okay? the family saw that ⁓ the company was evolving, that were getting... very interested in a license of our software, okay, and that was one our first customers. Of course, that was an academic license. So you know that ⁓ in an academic institution, budgets are more restricted. So it was not a sale, it was our first software sale, and it was based on benchmarking. After that, we ⁓ we naive again, okay, and we like grants that we were getting investment from investors. So two years, would say that the family okay, this is to some extent safe. It's not Enrik now has gone with the crisis, with the year old crisis, no, no, okay. So ⁓ they the progress. ⁓ If you have benchmarking and you show that your solution is better, will come to us. Show me the data. Here's the data. And ⁓ realized afterwards that ⁓ solution is only, is This is % better, this is three times better, or even this is five times better, or 10. It needs to be more than that. Usability, easy ⁓ access, trust, not only benchmarking. ⁓ But yeah, ⁓ our second was based on a benchmarking study. we said, are the kings, we just need to show. ⁓ that our algorithms are more accurate, our algorithms are better, and people will come to us. No, no, no. does not work like that.
Imants Zudans: you took an idea is very common in world. You benchmarking and brought that to chemistry
Enric Gibert: I that benchmarking and not only in drug discovery, ⁓ we it also when we working the high-performance computing ⁓ community. Benchmarking per is ⁓ gigantic task, because, and full ⁓ issues, okay, and of debates. What do you choose as the baseline? What data sets you are using? What configurations you run? What of statistics you report? it's something can lot of controversy ⁓ and it can even become religious discussion. ⁓ So we were ⁓ reading the other ⁓ ⁓ opposed by Pat Walters complaining about the DUTE for learning. So we to go through all this There's people that report hitrate ⁓ and their main ⁓ metric ⁓ hitrate. This not good. in the that I can have 100 percent hitrate if I want. I prefer having ⁓ a much lower ⁓ hitrate but having two different scaffolds or two different families of molecules, rather than having all ⁓ the molecules that I test experimentally being active, but all of them being just small modifications of the same family. So you always have ⁓ type of questions. Have you taken novelty? Have you taken diversity into the equation? ⁓ are not designing hits we are designing drugs. ⁓ So it's a multi-parameter optimization. ⁓ So yeah, mean doing benchmarking is more than a science and you need to be quite open about it in the sense that, okay, here are the scenarios in which the technology works well because XYZ, And these are the scenarios in which the technology does not work well because ABC, There's no... one size all ⁓ in drug discovery. Every drug discovery project is different. So sometimes ⁓ you a different configuration of ⁓ a workflow ⁓ order succeed. And having always in mind that you are designing drugs, you are not designing ⁓ hits. So benchmarking tough. ⁓
Imants Zudans: customers, did look for customers or was this benchmarking study helping you? They were kind reading the paper and out to you.
Enric Gibert: No, we were looking for customers. We started attending some conferences. attended BioEurope, Bio Convention. ⁓ These are conferences, but I don't think that these are the best conferences to find the scientists. We still go BioEurope and Bio. Don't get me wrong. But if you want find you need to go to the medicinal chemist and the, where medicinal chemists and computational chemists okay? So you need to go to scientific At the end of the day, in our industry, you need to convince ⁓ by the and by the technology ⁓ and the that you obtain with that science and that technology. ⁓ The conversations always very scientific. So even the decider, ⁓ is a chemist or a biologist that he or she has decided to take like a managerial track a company, ⁓ but normally the deciders have a PhD in medicinal chemistry, in organic chemistry, ⁓ in biology in computational chemistry, okay, so they are PhDs. So you need to go to forums, okay, at the very beginning we went more to these business development forums that I don't think that are the correct audience. when you are a startup. Afterwards, yes, when you are fundraising, when you want to establish partnerships, you are ⁓ for more collaborative ⁓ alliances, ⁓ BioEurope and Bio are very good. But if you want to ⁓ ⁓ customers of slash AI tools, you need to go to the scientific events where the ComCams and the... ⁓ the medicinal chemists are going. it took us a while to realize that. It took us a while.
Imants Zudans: somewhere here have a book that says selling to scientists ⁓ is biggest challenge in sales. Scientists by nature are skeptical, critical, and they're also very smart. So I'm wondering how did go about that?
Enric Gibert: Yes True true true true. Yeah. No, we realized very early in the project that a Conversation ⁓ ⁓ sales conversation Is always driven by the science. Okay, and a scientific decision even ⁓ if if talking to a biotech the drug hunter or the medicinal chemist or the computational chemist the scientist and the decider at the same time. If you're talking to a big pharma, you to convince the science, the hunter the medicinal chemist, but you need also to convince ⁓ the manager. normally, the deciders also people that are scientists, so that they have a PhD in chemistry, organic chemistry, or biology. And the conversation still is scientific. ⁓ So we saw from the very beginning that we were not selling pencils. ⁓ We were selling sophisticated R &D software people that are skeptical, critical, and ⁓ opinionated. ⁓ So the that we incorporated to our business development efforts were mainly scientists had ⁓ or that skills and can build relationships. You cannot bring in persons that do not understand the science or that ⁓ have not worked ⁓ drug discovery ⁓ this is by the ⁓ purchaser or by buyer which is a scientist.
Imants Zudans: you were enough to realize this before starting selling or was this discovered in the process.
Enric Gibert: It was discovered in the process. At the very beginning, I mentioned, one of the learnings was, OK, we have the benchmarking. Now I need to go to, or they will come to me. Or when I just plot ⁓ or I the plot graphs in ⁓ a conference, will just say, ⁓ yes, yes, let's move forward. I remember ⁓ my first. Bio International Conference that I attended in San Diego in 2017. don't remember, we had 20 meetings. When I came back, said, OK, of these 20 meetings, I think that have opportunities, so 10 customers. And we got zero. I was or I am an engineer. So I was talking to scientists. that the time that I was realizing the drug discovery was very big. ⁓ And also realized that it's not only about benchmarking. Okay, it's also about having cases. Who have you worked with? Okay, where have you this? And all these how the software is. Okay, that was a process that took some time. Okay, to realize that we had to have scientists ⁓ that had communication and social skills so that they can engage on collaborations, on proof of concepts, on establishing conversation that understand the science and that can raise the interest. Yeah, of course. If a person that is interested, then we will bring the persons that are developing the software. ⁓ And that can explain you the very last argument ⁓ or ⁓ that want to. that you want to test, but at least people in business development, they need to understand the algorithms do and the outcome of the algorithms.
Imants Zudans: So now in your team you have ⁓ a lot of scientists, right?
Enric Gibert: Yeah, mean 75 % of our team has a PhD in chemistry or organic chemistry or computer science as we do. have several industrial PhDs, has also been a source of talent attraction and incorporation into the company. This also puts us at the forefront of science. So mean, very, very science first and a very, very science-oriented company.
Imants Zudans: Was there ever a moment you thought that this is difficult, moment where you came close to ⁓
Enric Gibert: Not really, mean the first years as we commented was traversing the desert but once started having traction with recurring including some big pharmas we had the sensation that we're not going to make I would say after first two or three years, we had a clear that, OK, is doable, OK, and this is going to work out.
Imants Zudans: When did you start to feel that selling becomes easier? That it's not so much struggle anymore when it is easier to get the next ⁓ client.
Enric Gibert: a ⁓ I have ⁓ really good answer for. I guess when you start seeing that... at the end of the day, confidence. Creating confidence and seeing that you have created confidence in the users or in the customers. ⁓ a project with biotech in the UK that was working on ⁓ what was... I don't now, on fibrosis, yes, ⁓ on fibrosis, ⁓ And ⁓ did, think from perspective, we did a pretty... the first software users renew the license, then you start seeing, this is recurrent revenue. So there is a market. There is a need. We're providing a solution that ⁓ is valued by users, and not because they have tried once, but they are renewing. ⁓ That's an important ⁓ nice piece of work because results were very encouraging but at that point in time that biotech company or their investors, their shareholders decided to put all the efforts in clinical stage assets and discontinue drug discovery although the work was good they decided to discontinue drug discovery ⁓ the person was running ⁓ was the CSO Another important milestone is, we as a company, provide access to software, okay, so we sell software licenses. On the other hand, we also provide services, okay, and we thought, okay, the traditional theory, okay, software is like 90 % recurrent revenue. In services, you need to chase ⁓ every single person. No, when we started seeing recurrences in services as well, that meant of this company was Macrophage Pharma. name is ⁓ Reddy. He moved to Inception. This company from San Francisco and he became CSO of General Inception. And other means, we had our slide deck through other friends to General Inception. So when joined General Inception as a... that ⁓ in previous projects with that same customer, we delivered. So they were repeating with us. That creates comfort. If you have both users of the software renewing, this means that it's providing answers or ⁓ results that meaningful for them. And this is why they want to ⁓ continue engaging you. ⁓ And they you even ⁓ for some and functionalities. A CSO, one of his first duties was, okay Venkat, we have received this slide deck, okay. ⁓ This company from Barcelona, it seems that it's cool technology, but we don't have any way to assess whether are good or not. I mean, please assess they are. And Venkat ⁓ no, yeah, yeah. ⁓ I know Yeah, serendipity again. mean, I know them, I work with them. that so they are interested on expanding the usage of the technology and also recurrent in services. When we started seeing ⁓ customers repeating with us and ⁓ subcontracting project number two, project number three us, that also creates a lot of comfort. that's a point in which
Imants Zudans: Serendipity again.
Enric Gibert: It's never easy, you can never relax, but it's a point in which you say, okay, I mean, the market is responding, I mean, we're providing something that is meaningful to the drug discovery industry. technology is brilliant, the guys are ⁓ good, you can trust them. then we ⁓ a strategic alliance with General Inception, ⁓ became shareholders of our company and are their artificial intelligence drug discovery strategic partner. So we support for companies in their portfolio. So this serendipity, the is very small ⁓ and suddenly someone we had worked in the past and was happy with the technology and with our work.
Imants Zudans: When was that for you?
Enric Gibert: For me that was ⁓ four or five years ago. Yes. became the CSO of this company, Gniter, and then that brought a new partner ⁓ on the table. That a lot ⁓ of confidence that the project will move forward.
Imants Zudans: So approximately at the start of COVID, something like that, right?
Enric Gibert: Yeah, yeah.
Imants Zudans: Do you miss on chips?
Enric Gibert: ⁓ some extent, yes, ⁓ a researcher. It's my passion. I've always ⁓ to work chip design ⁓ since I that particular subject ⁓ during ⁓ engineering ⁓ I was very fortunate because at that point in time there were not many ⁓ chip design companies in Spain. There was like a big R &D group at the UPC, the Polytechnic University of Catalonia. And I was fortunate to ⁓ land there, OK, because that opened the opportunities everywhere. And then Intel came to Barcelona. I miss it because that's my background. ⁓ Similar to I miss programming, I miss doing research on that particular topic. But I just chose, also because of my personality and the paths of life, just chose to move ahead in this managerial role. But people at Pharmacellera, mean the people that are doing the algorithms, they know that when we have a one-to-one conversation or when we go for lunch, I love hearing them telling me that they have tried the hydrophobic profile like this way. And I get into the details of the algorithms, into the details of how the algorithms runs, what is the benchmarking, what is the execution time, what are the hardware requirements. So I will always have this. in my heart, okay? But it is true that I have just chosen to move ahead in the management career path.
Imants Zudans: Okay, so it is 22 it is ⁓ peak hype for Biotech. There is a lot of money ⁓ being thrown in Biotech. ⁓ You are revenue, you have clients, ⁓ you've that your algorithms, your tools are useful, they ⁓ drug discovery. Did consider exit at that time?
Enric Gibert: Well, mean, 10 years is a long time, so we... We had conversations, ⁓ both led by us and led by others that have approached us, companies that are working on the similar space but have complementary technology. ⁓ So we could have something bigger and with a grown customer base. companies not even having computational, ⁓ more ⁓ ⁓ lab capabilities and lab skills that could complement by some sort of ⁓ artificial intelligence and quantum mechanics machinery in order to complement those. Yeah, have had several conversations. mean, ⁓ are open ⁓ this regard to explore not only collaboration or strategic alliances, but even joint ventures. ⁓ are open to that. ⁓ is true that it's not easy ⁓ when you join forces not as a partner that we have several already. But if you join forces more on a joint venture with another organization, there must be a very clear alignment on the culture, on the goal where you want to reach, and on the chemistry of the teams. So it's not easy. So long story short, yeah, we've had in these last 10 years ⁓ conversations
Imants Zudans: ⁓ do you get feedback from, from users?
Enric Gibert: ⁓ in order to do something bigger and or to join forces with other organizations both in Spain and outside of Spain yeah
Imants Zudans: But you decided to go still on your own, right?
Enric Gibert: Yeah, no. An interesting conversation, for example, was with a big pharma company. And computational he was the simulations, and he was handing the results ⁓ to the chemist. And ⁓ the results, the prioritized by the platform, he was handing that ⁓ to the using smiles ⁓ or a ⁓ structure of the molecule. ⁓ And he well, I mean, The molecules that the technology is proposing are so novel and are so diverse that medicinal chemists are challenging me about the decisions that I do using your technology. ⁓ And then I need to show them for some specific scaffolds that they say, ⁓ no, no, let's not go this route. Why the system chose it? once you see the interactions in 3D and the hydrophobic profile of the molecule, then they say, ⁓ now I understand why this molecule was prioritized ⁓ over others that from a medicinal chemist or from a 2D perspective made more sense. also you need to make ⁓ that move medicinal chemists out of their comfort zone ⁓ and known chemistry. ⁓ It took us three years to get this type of feedback. then we to release a graphical user interface in which you could in 3D very well ⁓ why the was proposing these molecules. And this type of feedback, it took us three years to get it. But once we got it, that is very valuable. ⁓
Imants Zudans: So first 10 years are behind you. What is in front of you? What are the next five years?
Enric Gibert: So I think that after the first generation of artificial intelligence drug discovery companies fell short with respect to the expectations. I don't want to say that they failed. They didn't fail. I artificial intelligence now is super important and will continue being super important for drug discovery. But after they didn't meet the expectations, most of them, the industry is paying A lot of attention on the combination of ⁓ artificial intelligence slash machine learning with physics-based simulations. Every drug discovery project is different. So you some sort of physics-based ⁓ methods in order to characterize that particular system that you are trying to design drugs, not hits, for. ⁓ In this regard, mean, a pharmaceutical is well positioned, because we apply quantum mechanics, so physics-based simulations combined with artificial intelligence. And now, in addition, there is all these ⁓ exponentially growing chemical spaces, these ultra-large ⁓ chemical spaces that can be synthesized and tested rapidly. is ⁓ important ⁓ being able to test hypotheses a hit ID stage. very rapidly and very cheaply and based on this data start working on some candidates and then spend more money on ⁓ optimizing the properties of those particular ⁓ families that you have selected. So in our particular case we have the foundations, we have the engine, now it's a matter of ⁓ applying it beyond HitID. We have been... working a lot the hit ID stage, but we also have success cases in the hit to hit and hit optimization stages. ⁓ applying our engine in these ⁓ additional ⁓ discovery stages. And course, mean, working on new modalities molecular glues we just incorporated ⁓ an application scientist has experience ⁓ in applying computational for protax and molecular glues. ⁓ So all these type of new moralities and new type of targets, this is an area in which the company can also grow with additional success cases.
Imants Zudans: 10 years, ⁓ congratulations. I see there is very significant chapter in front you. So ⁓ I wish you success ⁓ and endurance. And maybe we'll have you back a year or two ⁓ see ⁓ the progress. So thank you, Enric, for joining me. was Enric. ⁓
Enric Gibert: Perfect. Thank you for inviting me. It was a privilege. And yeah, let's see if we can talk again in two years and see where we were at at that point in time. That would be interesting.
Imants Zudans: That was Enric Giber, CEO of ParmaCellera. I'm Imants Zudans. Thank you for listening.
