In this conversation, Enric traces the full arc: the Monday-to-Wednesday shutdown that ended his Intel career in 48 hours, the hardware “pharma box” that almost sent the company down the wrong path, twenty conference meetings that produced zero customers, and the moment he realized that scientists don’t buy better benchmarks. They buy trust.

Listen to the episode on Apple Podcasts or Spotify or here:

If you prefer reading, the full transcript is at the end of this page.

Guest

Enric Gibert

CEO and Co-founder of Pharmacelera. PhD in Computer Engineering from Universitat Politecnica de Catalunya (UPC). Ten years at Intel Corporation, where he accumulated 25 scientific publications and 20 patents in high-performance computing and hardware/software co-design. Executive management program at IESE Business School.

Enric on LinkedIn
Pharmacelera

Timestamps

• 00:00 — Introduction

• 00:36 — A Decade at Intel

• 03:38 — The Monday-Wednesday Shutdown

• 08:10 — Finding Drug Discovery

• 17:54 — The Pharma Box

• 25:33 — Crossing the Desert

• 29:23 — First Customers and Benchmarking

• 34:40 — Selling to Scientists

• 41:28 — Product-Market Fit

• 48:21 — Ten Years and Beyond

• 55:25 — Closing

One Thing to Know

Quantum mechanics is not quantum computing. QM is a physics theory that describes how molecules behave at the atomic level. QC is a new type of computer hardware. Pharmacelera uses QM algorithms running on conventional high-performance computers to model molecular interactions more accurately than the industry-standard tools.

Key People Mentioned

Enric Gibert (CEO, Co-founder) — Former Intel engineer. PhD in Computer Engineering, UPC. Brings high-performance computing expertise and Intel’s benchmarking culture to drug discovery. Enric’s LinkedIn profile

Enric Herrero (CTO, Co-founder) — Former Intel engineer. PhD in Computer Architecture, UPC. Research stays at UC San Diego and KTH Sweden. Specializes in memory hierarchies and hardware acceleration. Enric’s LinkedIn profile

Javier Luque (Scientific Director, Co-founder) — Full Professor at University of Barcelona. H-index of 84, 400+ publications. Creator of the MST quantum mechanical solvation model that forms Pharmacelera’s scientific core.

Manel Lopez (Co-founder, Senior Principal Scientist) — 20+ years at Almirall's molecular modeling group. Introduced Javier Luque to the team. After years in advisory roles, now back at Pharmacelera full-time. Manel's LinkedIn profile

Antonio González — Enric’s PhD advisor and founding director of the Intel Barcelona lab. Professor at Universitat Politècnica de Catalunya. Antonio’s UPC page

Venkat Reddy — CSO at General Inception, former CEO at Macrophage Pharma. Early champion of Pharmacelera’s technology. When Macrophage discontinued drug discovery, Reddy moved to General Inception and brought Pharmacelera in as their AI drug discovery strategic partner. Now sits on Pharmacelera’s board. Venkat’s LinkedIn profile

What Pharmacelera Makes

PharmScreen (Pharmacelera’s first product) — Ligand-based virtual screening software that uses 3D molecular descriptors derived from quantum mechanics calculations. Identifies drug candidates that traditional tools miss by exploring untapped chemical space. In published head-to-head benchmarks on GPCR targets, PharmScreen found 62 confirmed hits compared to 22 by competitors, with 67-75% of those hits unique to Pharmacelera’s approach. Enables scaffold hopping, helping chemists move beyond known chemical series to find structurally novel leads.

PharmQSAR — Predictive modeling for lead optimization. Helps chemists understand what parts of a molecule contribute positively and negatively to the specific molecular property of interest.

exaScreen (~2023) — Screens trillions of compounds in 3D using QM-derived molecular descriptors and a synthon-based approach. Scales linearly with building blocks. Integrated with EnamineREAL and eMolecules for direct ordering of physical compounds. Used for virtual screening and scaffold hopping.

MolXplore (2024) — Web-based graphical interface designed for medicinal chemists who need to visualize and evaluate virtual screening results without touching a command line. Pharmacelera’s tools are also available via CLI for computational chemists, API for HPC clusters and cloud deployment, and KNIME nodes for workflow integration.

Quotes from the Episode

On the Intel years:
“There was not a single day in which I complained when going to work. Because we were having fun.” (03:22)

On building what you know instead of what the market needs:
“What do Enric Gibert and my colleague Enric Herrero like? We like hardware accelerators. So we said, we’re going to build a box.” (18:54)

On the founder’s roller coaster:
“Haven’t you sometimes regretted or had doubts? For several years, I made this question to myself 10 times per day at least.” (26:05)

On selling to scientists:
“We were not selling pencils. We were selling sophisticated R&D software to people that are skeptical, critical, and very opinionated.” (37:00)

On the Bio International reality check:
“I had 20 meetings. When I came back, I said I have 15 opportunities, 10 customers. And we got zero.” (38:37)

On benchmarking philosophy:
“We are not designing hits, we are designing drugs.” (32:17)

Bonus: What We Didn’t Have Time For

Here’s what didn’t make the final cut.

The mistakes worth naming. Enric listed three: being too optimistic with early revenue projections, building a solution looking for a problem (”hammer seeking nails”), and underestimating what it takes to sell to skeptical scientists. Every founder in this space will recognize at least one.

What services teach you that software cannot. Enric’s early instinct was to prioritize software revenue because it recurs. But services turned out to be the faster path to traction. When you sit next to a scientist running your tool, you see which parameters they change, where they hesitate, what results confuse them. Software gives you binary feedback: renewed or not. Services give you a continuous signal. That signal drove critical product decisions, including the 3D visualization interface that made chemists trust the algorithm’s output.

Why not become a biotech? Every computational chemistry company faces this question. Pharmacelera chose to stay a tools company. Running your own drug pipeline sounds attractive: one successful asset could be worth billions. But it requires a completely different set of skills, a different funding structure (more speculative, longer timelines), and a different team composition. Pharmacelera’s compromise: a joint venture with General Inception to co-found a biotech that has privileged access to the platform, without converting the entire company into a drug developer.

Where the field is heading. The first generation of AI drug discovery companies fell short of expectations. Not failed, but short. The industry is now paying close attention to hybrid approaches: physics-based simulations combined with machine learning. Pharmacelera has been doing exactly this for a decade. Beyond hit identification, Enric sees the engine expanding into hit-to-lead and lead optimization stages, and into new modalities like PROTACs and molecular glues.

Other Interviews with Enric and the Team

• BiopharmaTrend: The Rise of Quantum Physics in Drug Discovery

• BioEmprendedores: El caso Pharmacelera (Spanish)

• On Quantum Theory, Drug Discovery, and Life Sciences Ecosystem in Spain

• Capital Cell: Entrevista a Enric Gibert, CEO de Pharmacelera (Spanish)

• FBG/UB: F. Javier Luque — “We aim to design new tools that increase the success rate of new drug candidates”

• BioXconomy: Pharmacelera bags $7.1M in funding

Links and Resources

• Pharmacelera website

• Pharmacelera science overview

• Case studies

Subscribe

If you work in drug discovery tools or are thinking about building a company in this space, subscribe to get notified when new episodes come out.

Subscribe now

Episode transcript

This transcript has been edited for clarity and readability. The meaning and sequence of the conversation are preserved.

Chapter 1: Introduction (00:00)

Imants:

Welcome to Imants Podcast, where founders powering drug discovery share how they built their companies.

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? That’s what we explore today.

Enric, thank you for joining us here today. Maybe let’s start your story with Intel.

Chapter 2: A Decade at Intel (00:36)

Enric: Thank you, Imants, for the kind introduction and thank you for inviting me. It’s a pleasure to be here talking today to you and the audience.

So yeah, Intel. As you mentioned, my background is not chemistry, biology, or any scientific background. I’m a computer engineer. My areas of expertise are high-performance computing, software development, artificial intelligence. I did the PhD in these fields at the Polytechnic University of Catalonia here in Barcelona, where I live. During the PhD, I did a research stay at the University of Illinois in Urbana-Champaign in the US. After doing the PhD, I joined Intel. Intel at that point in time had opened a research lab in Barcelona that was very near to the UPC. It was, in fact, directed by my PhD advisor, Professor Antonio Gonzalez. So it was a natural step for me.

I stayed there for 10 years and what we were doing there was designing the processors, the chips, the hardware accelerators at a 5-to-10-year time frame. When you need to design what is going to be the microprocessor or the GPU or the hardware accelerator for this time window, five, ten years from now, you don’t have the hardware. So you need to start doing simulations and modeling and benchmarking against your current products. That is what we were doing. Nothing to do with life sciences. We were designing the chips and the hardware accelerators of the future. It was a research lab, but the idea was to impact the product lines of Intel.

The Polytechnic University of Catalonia had a big research group on computer architecture. The group had a pretty strong reputation. So Intel decided to open an R&D facility. That was 2002, 2003. The lab began with just eight people and over 10 years it reached a peak of 80 people, 60 of them with a PhD. It was a small research site compared to other sites at Intel. 60, 80 people. We were collaborating a lot with the people in Santa Clara, where the headquarters of Intel are. We were also collaborating a lot with the people in Oregon, Israel, and Germany.

Imants: That sounds very exciting, very interesting.

Enric: As you say, it sounds like a lot of fun. We had a blast. I had the privilege to work with people that were super smart, super inspiring. 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.

Chapter 3: The Monday-Wednesday Shutdown (03:38)

Imants: So you spend a decade at Intel and then it was in 2014 that Intel decided to close the Intel Barcelona site. What happened?

Enric: One of these unfortunate situations 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 fell short with respect to expectations. 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. They decided to concentrate all the R&D efforts in bigger sites.

That story is a lesson learned because I’ve seen other companies, big companies, in which they need to reduce headcount or they need to do some rework. You start hearing rumors. You spend six or nine months asking yourself, is it going to be me? And then you 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.

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. They were discussing what projects we would be running, 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, 48 hours afterwards, to say that the site was closing. That we were laid off.

Immediate action, no rumors. In two months we were gone, all of us. So this started me thinking, if this happens again to me, what would I prefer? Would I prefer to know it six or nine months in advance and start thinking, my god, my god? Or would I like to have a very short notice? Intel was in good shape financially, so we got a good compensation package. When I explain the story, a lot of people tell me, they behaved very badly. I say, well, no. It was straight. Period.

At that point in time, since we were notified, I think it was like two months or something like that. Everybody started looking for opportunities. And this is another interesting story. That week that we were told we would be laid off, three or four days afterwards, ARM, the chip designer of chips for mobile phones, based in Cambridge in the UK, landed in Barcelona with all our profiles from LinkedIn, and they started interviewing. That was one of the best talent acquisition exercises I have ever seen. In 48, 72 hours, the managers of ARM were in Barcelona interviewing all the team.

A lot of people went to ARM, to Cambridge. I had the opportunity to go to Cambridge. Some people moved to the US, to Silicon Valley. I also had an offer to stay at Intel. And some other people just wanted to look for other alternatives. In my particular case, I had been for 10 years in a multinational. I knew what working for a big company was. But my heart was asking for something different. I had a blast at Intel, of course. But I still had a lot of energy and I said, if I don’t try it now, I will never try it. So together with Enric Herrero, we decided to explore opportunities in the life sciences and that’s when we decided to start Pharmacelera.

Chapter 4: Finding Drug Discovery (08:10)

Imants: Drug discovery, was that a deliberate decision, or were you considering other options as well?

Enric: It was serendipity, as many things that happen in life. Together with Enric Herrero, we were colleagues at Intel, although it is true that 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. We had common friends.

We looked at genomics. Fascinating from a research perspective. You must also understand our background. Both Enric Herrero and myself, we are PhDs in computer engineering. We are scientists, we are researchers. Working on genomics, we saw that it was super exciting, super challenging. But our first interactions, we saw that it was still a very immature field. We would have had a lot of fun because it’s an exciting topic. But most of the money was grants, public funding. It’s tough from a business perspective. So we ended up discarding genomics.

And at that point in time, one of these situations in life. I was taking my elder daughter to school and a mother of a boy attending the same class, she was working at Almirall, the largest pharmaceutical company in Spain, and she told me, Enric, there is an interesting profile that was let go at Almirall because they were reshifting their focus. There’s a person that has worked at Almirall on modeling for 20 years. He’s brilliant. You should talk to him. I’ll do the introduction.

So we met, Enric 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. We started interacting with him, and he started teaching us what drug discovery is, because Enric Herrero and I, we didn’t have any idea what drug discovery meant. Not only did he explain to us what drug discovery was, but what drug discovery in the industry is and who is who.

When we started working with Manel, he also said we would need someone with a solid and strong scientific background from academia. He presented us to Professor Javier Luque, a principal investigator at the University of Barcelona. His areas of expertise are quantum mechanics. All these accurate algorithms that describe molecular interactions at the atomic and subatomic level.

We went and visited him at the university. At the very beginning, he said, these two guys, these two engineers, these two freaks, they think 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 through a trial. He said, let’s collaborate for six months. We are working on this publication, jump in as co-authors, but you need to do some work. It was not as explicit as this, but I could sense it.

After six months, he saw that we were serious, that we were professionals, and that we were good people worth having interactions with. So then we incorporated Javier Luque as CSO of the company. That was the foundational team 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 real projects. And an academic that brought the scientific excellence into the project. Serendipity.

If you dig a little bit about not only stories of companies, but even in your personal life, how many situations are just the outcome of serendipity and random interactions? Randomness and serendipity play an important role in life.

Imants: What was the discussion around who would be the founders and how each would contribute? It’s a very diverse team. I assume the professor probably didn’t want to leave his hard-earned position. How did the team actually form?

Enric: With respect to positions, you are right. Professor Javier Luque didn’t want to take the risk of losing his academic position. He continued, and he continues to be a principal investigator at the University of Barcelona. So we knew that his involvement would be part-time.

At the same time, Manel Lopez was also looking for a job. He needed a full-time job, so he was not ready either to jump full-time to Pharmacelera because we didn’t have a budget, didn’t have resources. He started in the project in an advisor position, part-time. It was clear that he would not have day-to-day duties in the project because he needed income at that point. So he advised us. And then it was Enric Herrero and myself as full-time employees with no salary.

At the very beginning, it was curious because we both knew each other from Intel, but we had never worked together. There was a period in which we had to get to know each other. When we were pitching the idea, we would step on each other. I was saying A, and he would say A’, I would say A’‘.

At that point in time, we participated in an acceleration program organized by the Mobile World Capital Barcelona, an entity that tries to promote that the Mobile World Congress stays in Barcelona, digital technologies, including health. We were coached. Part of the coaching was: why did you decide to create Pharmacelera or a project like this one, what drives you there, and what would you do there.

Those coaching sessions were extremely valuable because we saw first that Enric Herrero and myself, we were very aligned. We were not there for the money. Of course, at the end of the day, money is important. But we were not there for that. If I become rich, so be it, I will not complain. But we were not there for egos either or publicity. We were there for the challenge. We thought, individually without discussing it between us, that our motivation to build a company like Pharmacelera was because we saw a challenge and we love challenges.

During those coaching sessions, it became apparent that because of his personality, his skills and his capacities, Enric Herrero fit more into a CTO role, and my role would be more CEO. Those sessions were very enlightening. Once it was clear who was going to do what, I decided to do some degrees in management to complement my previous background.

Imants: What surprised you about drug discovery?

Enric: What surprised me most about chemistry? Maybe it would be similar if I explained to you how a processor is designed. Maybe, Imants, you think a processor is just a bunch of 20, 30 engineers working in a garage. No, no, no. It’s a humongous effort. Different departments, different duties. That is something that also struck me from drug discovery. I thought that there were just chemists doing some experiments. But I didn’t realize the amount of effort that is behind a drug discovery project.

From 10,000 feet, from a distance, you see this is important, this is challenging, but it’s a few chemists doing X, Y, Z. But once you dig into it, you start understanding the complexity. Similar with designing processors. There’s a plethora of groups passing or moving the project from one stage to the other. That was something that struck me about drug discovery.

Chapter 5: The Pharma Box Mistake (17:54)

Imants: How did you decide what is the first application that you are going to make? Was there some scientist involved that said, we need a program that would do this? How was that decision made?

Enric: In that regard, the advantage was that Professor Javier Luque, PI from the University of Barcelona who became our CSO, was an expert in quantum mechanics and quantum mechanics-based algorithms. So we decided to use this expertise to develop an engine in order to simulate molecular interactions with QM-based algorithms, with these accurate algorithms. We chose virtual screening as our use case.

Looking in retrospect, I see something that if I start again, I would try to avoid. It was an error. We had these algorithms, these QM-based algorithms for virtual screening. What do I, Enric Gibert and my colleague Enric Herrero, like? We like hardware accelerators. So we said, we’re going to build a fharmas box, a piece of hardware based on hardware accelerators or GPUs or even FPGAs.

That was my area of expertise and Enric Herrero’s area of expertise. We wanted to build something that was self-contained. A hardware machine, the software installed on it, the software very integrated with the hardware, with GPUs or even FPGAs, Field Programmable Gate Arrays, which is hardware that can be reconfigured. You can build hardware that is ad hoc to that particular algorithm.

Imants: And what is the advantage? Is it much faster? How much faster is it?

Enric: It can be a hundred times faster, a thousand times faster.

Imants: So you kind of built not only an algorithm for virtual screening but actually a server that is tuned to do that.

Enric: Yes, absolutely. The idea was, we will sell this combined product, hardware plus software, to our customers, which will be big pharmas and biotechs.

Imants: Out of curiosity, how big is the box?

Enric: The idea was that it would be like a fancy workstation. It didn’t have to be a big HPC system. Similar to a workstation that today has GPUs. The idea was, instead of GPUs that are very good 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 1,000 times faster, so three or four orders of magnitude faster than a traditional CPU. It sounded from a scientific perspective super interesting, very aligned with our expertise.

But we became aware of two potential issues, two roadblocks. First, in order to have hardware that is reprogrammed to run your software, your software needs to be very stable. At that point in time, the software was evolving very rapidly. We were having prototypes and tweaking it here, tweaking it there, doing changes here, doing changes into the alignment of molecules. So the software was evolving very rapidly. So it was very hard to have a version very stable to map to hardware.

The second learning was that if we wanted to sell this, we would need not only to convince the chemists, the scientists, but also need to convince the IT department of the pharma. They would already have their own approved list of hardware providers. So here comes a company that says, you need to buy these machines because this has all the software integrated. We saw that was an additional complexity into the discussions. And at that point in time, cloud computing was becoming ubiquitous. Although pharma has embraced cloud computing at a slower pace than other segments, it was becoming obvious that we didn’t want to fight with our own hardware.

So we dismissed the pharma box. It was a neat exercise. Very interesting from a scientific perspective, but not very interesting from a business perspective, I think.

Imants: Do you think it was difficult to sell, or that the algorithm improvements would outpace the advantages that the hardware solution offers?

Enric: I think it’s both. The latter? Obviously! The software was evolving and is still evolving. We’re now expanding our applicability of QM algorithms to other areas beyond ligand-based virtual screening. The algorithm was evolving and is evolving faster than the hardware can be.

In addition, we realized something. Enric Herrero and I as computer engineers working at Intel, trying to optimize. We love accelerating software and running software much faster. But at the end of the day, the value is in the outcome of the algorithms. The value is in the molecules that are proposed by the algorithms.

If you ask a drug hunter, what do you prefer? Given a drug discovery project, do you prefer to have mediocre molecules, molecules that are not very novel or not very diverse, in a few minutes? Or is it okay for you to spend some hours, even days, doing more accurate simulations and get more relevant starting points? Molecules that are more novel, more diverse, that allow you 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 in another direction.

I think the value is in the latter. Of course, if you tell the drug hunter that he or she needs six months of simulations, they will say, I’ll go with something faster that I can iterate several times. But if you are able to design technology that gives responses to molecular modelers in a reasonable amount of time, the value is in the results. The value is in the assets. The value is in the molecules that the algorithms propose. That also made us put all the focus on the quality of results. Speed 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: So you essentially pivoted. You had this idea about the pharma box. I still find it very fascinating as a chemist. Two or three orders of magnitude improvement sounds very relevant. But you realized that’s not the end solution. Did you have clients at this time already, or were you still working with academic partners?

Enric: At that point in time, we pivoted, as you said, and the decision was quite fast. Maybe we explored the option of the pharma box, I would say, for a year or something like that. Not more than that. We just realized that it didn’t make sense quite fast. We didn’t have customers.

Chapter 6: Crossing the Desert (25:33)

Imants: You described to me that this is like deciding to cross the desert. Before crossing, that would be Intel. And then there is a dry zone in front of you. You have to survive yourself, your family. That’s a bold journey. You had to convince your co-founders to embark on it. But also the family. How did you convince them?

Enric: The early days are tough because you come from a comfort zone working for a multinational with a good salary. And then suddenly you enter into this roller coaster. I always say, when someone asks me, don’t you sometimes regret or have doubts about whether you should have moved to a big company or pursued your dreams? I always joke: I ask myself that question 10 times per day at least. So you can imagine that it’s a roller coaster.

On the other hand, having your own destiny in your hands, deciding yourself where you want to go, that’s very rewarding. So here at home, full support. It’s tough, we will give you a period in which you can try. Perfect.

Imants: Did you discuss how long that period would be?

Enric: No. And I tried to discuss as little as possible things from the company or things that had to do with Pharmacelera at home. You don’t want to mix both things. You don’t want to have your family being your other board of directors member. I want to decouple that.

Imants: But they are with you on a journey. It might end up with nothing. I don’t want to drill you too much on this topic, but I think younger founders see the success stories. You started something, and in a couple years, revenue, investment, etc. But most of the time, as I see, it actually takes at least twice as long as they thought. That’s when the family starts to ask questions. You said this is going to be a couple years and it’s been five years. It affects them too.

Enric: Absolutely. It’s a roller coaster from a personal perspective. I was in my early 40s. I had two daughters at that point in time. It’s not the same as starting a company when you are 25 or 27 and you have fewer duties, fewer ties, than when you are 40-something.

At the same time, you have a little bit more experience and you know how to balance the pros and cons. When I said that I didn’t want my family to be like another board member of Pharmacelera, of course I discussed the progress. The family saw that the company was evolving, that we were getting grants, that we were getting investment from investors. After two years, I would say the family felt, this is to some extent safe. It’s not that Enric has gone crazy with a 40-year-old crisis. They saw the progress.

Chapter 7: First Customers and Benchmarking (29:23)

Imants: So how did you get the first client?

Enric: How did we get the first client? It was the Barcelona Supercomputing Center. They had some algorithms for drug discovery that were complementary to what we were doing, but they wanted to accelerate them. They subcontracted us to accelerate their algorithms. That was a service project that had nothing to do with our core engine, but it was our first source of income.

Our second source of income was an academic group we were collaborating with. It was not UB, not the University of Barcelona. We did benchmarking. Since both Enric Herrero and I came from benchmarking processors or benchmarking potential processors, doing apples-to-apples comparisons and that type of analysis. One of the first tests we did of our algorithms for the case of virtual screening was with this academic institution. They had some data sets, two or three data sets. They were small. And they said, why don’t we run these data sets in which we know which molecules are active and which molecules are not active, and just compare them to our current methods? They were using software from a big software vendor. Why don’t we do a direct apples-to-apples comparison?

We did that comparison and the results were very very good. They were very interested in having a license of our software, and that was one of our first customers. Of course, that was an academic license. In an academic institution, budgets are more restricted. So it was not a big sale, but it was our first software sale, and it was based on benchmarking.

After that, we thought, we were naive again. We said, if you have benchmarking and you show that your solution is better, people will come to us. Show me the data, here’s the data. We realized afterwards that a solution is not only “this is better.” This is 10% better, this is three times better, or even five times better, or ten. It needs to be more than that. Usability, easy access, trust. It’s not only benchmarking. But our second sale was based on a benchmarking study. So we said, we are the kings, we just need to show that our algorithms are more accurate, and people will come to us. No, no, no. This does not work like that.

Imants: You took an idea which is very common in the chip world, benchmarking, and you brought that to chemistry.

Enric: Benchmarking, not only in drug discovery, we saw it also when we were working for the high-performance computing community. Benchmarking per se is a gigantic task, full of issues and full of debates. What do you choose as the baseline? What data sets are you using? What configurations do you run? What type of statistics do you report? It can create a lot of controversy and it can even become a religious discussion.

We were reading a post by Pat Walters complaining about the DUD-E dataset for machine learning. We had to go through all this process. There are people that report hit rate and their main metric is hit rate. This is not good in the sense that I can have 100 percent hit rate if I want. I prefer having a much lower hit rate 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 chemical family.

You always have these questions. Have you taken novelty into the equation? Have you taken diversity? We are not designing hits. We are designing drugs. It’s a multi-parameter optimization.

Doing benchmarking is more than a science and you need to be quite open about it. Here are the scenarios in which the technology works well because X, Y Z. And here are the scenarios in which the technology does not work well because A, B, C. There’s no one size fits all in drug discovery. Every drug discovery project is different. Sometimes you need a different configuration of a workflow to succeed. And always keeping in mind that you are designing drugs, you are not designing hits. Benchmarking is tough.

Chapter 8: Selling to Scientists (34:40)

Imants: The customers, did you look for customers, or was the benchmarking study helping you? Were they reading the paper and reaching out to you?

Enric: No, we were looking for customers. We started attending some conferences. We attended BioEurope, the Bio International Convention. These are good conferences, but I don’t think they are the best conferences to find the scientists. We still go to BioEurope, don’t get me wrong. But if you want to find customers, you need to go where medicinal chemists and computational chemists are. You need to go to scientific events.

In our industry, you need to convince by the science and by the technology and the results that you obtain with that science and that technology. The conversations are always very scientific. Even the decider is a chemist or a biologist who has decided to take a managerial track within a company. Normally even the deciders have a PhD in medicinal chemistry, in organic chemistry, in biology, or in computational chemistry. So you need to go to these forums. At the very beginning we went more to business development forums. I don’t think those are the correct audience when you are a startup.

Afterwards, yes. When you are fundraising, when you want to establish partnerships, when you are looking for more collaborative alliances, BioEurope and Bio are very good. But if you want to find customers of computational/AI tools, you need to go to the scientific events where the computational chemists and the medicinal chemists are going. It took us a while to realize that.

Imants: Somewhere here I have a book that says selling to scientists is the biggest challenge in sales. Scientists by nature are skeptical, critical and they’re also very smart. How did you go about that?

Enric: We realized very early in the project that a sales conversation is always driven by the science. It’s a scientific decision. Even if you’re talking to a biotech, the drug hunter or the medicinal chemist or the computational chemist is the scientist and the decider at the same time. If you’re talking to a big pharma, you need to convince the scientist, the drug hunter or the medicinal chemist, but you also need to convince the manager. Normally, the deciders are also scientists. They have a PhD in chemistry, organic chemistry, or biology. And the conversation is still scientific.

We saw from the very beginning that we were not selling pencils. We were selling sophisticated R&D software to people that are skeptical, critical, and very opinionated. So the people that we incorporated into our business development efforts were mainly scientists that have communication skills and that can build relationships. You cannot bring in persons that do not understand the science or that have not worked in drug discovery, because this is perceived by the buyer, who is a scientist.

Imants: Were you smart enough to realize this before starting to sell, or was this discovered in the process?

Enric: It was discovered in the process. At the very beginning, one of the learnings was, we have the benchmarking, now they will come to us. Or when I show the plots and graphs at a conference, people will just say, yes, let’s move forward. I remember my first Bio International Conference that I attended in San Diego in 2017. We had 20 meetings. When I came back, I said, out of these 20 meetings, I think I have 15 opportunities, so 10 customers. And we got zero.

I was an engineer talking to scientists. That was when I was realizing drug discovery was very big. And I also realized it’s not only about benchmarking. It’s about having success cases. Who have you worked with? Where have you applied this? How usable is the software? That was a process that took some time. To realize we had to have scientists with communication and social skills so that they could engage on collaborations, on proof of concepts, on establishing a conversation. They need to understand the science and raise the interest. If there’s a person that is interested, then we bring the people that are developing the software, who can explain the very last parameter that you want to test. But at least the people in business development need to understand what the algorithms do and the outcome of the algorithms.

Imants: So now in your team you have a lot of scientists, right?

Enric: 75% of our team has a PhD in chemistry or organic chemistry or computer science, as we do. We have several industrial PhDs, which has also been a source of talent attraction and incorporation into the company. This puts us at the forefront of the science. Very science-first and a very science-oriented company.

Imants: Was there ever a moment you thought this is too difficult, a moment where you came close to quitting?

Chapter 9: Product-Market Fit (41:28)

Enric: Not really. The first years, as we discussed, was traversing the desert. But once we started having traction with recurring customers, including some big pharmas, we never had the sensation that we’re not going to make it. After the first two or three years, we had a clear view that this is doable and this is going to work out.

Imants: When did you start to feel that selling becomes easier? That it’s not so much struggle anymore, that it is easier to get the next client?

Enric: That’s a question I have a really good answer for. When you start seeing that the first software users renew the license. Because then you start seeing, this is recurrent revenue. There is a market. There is a need. We’re providing a solution that is valued by users, and not only because they have tried it once, but because they are renewing. That’s an important milestone.

Another important milestone: we as a company sell software licenses. On the other hand, we also provide services. The traditional theory says software is 90% recurrent revenue, and in services, you need to chase every single customer. But when we started seeing recurrences in services as well, that meant that in previous projects with that same customer, we delivered. So they were repeating with us. That creates comfort.

If you have software users renewing, it means the software is providing results that are meaningful for them. That’s why they want to continue engaging with you. They even ask for new features and functionalities, so they are interested in expanding the usage of the technology. And also recurrence in services. When we started seeing customers repeating with us, subcontracting project number two, project number three, that also creates a lot of comfort.

It’s never easy, you can never relax. But it’s a point at which you say, the market is responding. We’re providing something meaningful to the drug discovery industry.

Imants: When was that for you?

Enric: For me, that was four or five years ago.

Imants: So approximately at the start of COVID, something like that?

Enric: Yes.

At the end of the day, it is confidence. Creating confidence and seeing that you have created confidence in the users and the customers.

We did a project with a biotech in the UK that was working on.... I don’t remember now. On fibrosis. We did, I think from my perspective, a pretty nice piece of work. The results were very encouraging. But at that point in time, that biotech company, or their investors, their shareholders, decided to put all efforts in clinical-stage assets and discontinue drug discovery. So although the work was good, they decided to discontinue drug discovery.

The person who was running that, the CSO of this company, Macrophage Pharma, his name is Venkat Reddy. He moved to General Inception, a company Igniter from San Francisco, and he became the CSO of General Inception. By other means, we had sent our slide deck through other friends to General Inception. So when Venkat joined General Inception as CSO, one of his first duties was, Venkat, we have received this slide deck. This company from Barcelona, it seems like cool technology, but we don’t have any way to assess whether they are good or not. Please assess who they are. And Venkat said, I know them. I’ve worked with them.

Imants: Serendipity again.

Enric: The technology is brilliant, the guys are good, you can trust them. So we signed a strategic alliance with General Inception. They became shareholders of our company and we are their artificial intelligence drug discovery strategic partner. We provide support for companies in their portfolio. So this is serendipity.

The world is very small. Suddenly someone that we had worked with in the past and who was happy with the technology and our work became the CSO of this company, Igniter, and that brought us a new partner to the table. That creates a lot of confidence that the project will move forward.

Imants: How do you get feedback from users?

Enric: An interesting conversation, for example, was with a big pharma company. A computational chemist was running the simulations and handing the results to the medicinal chemist. The prioritized molecules by the platform, he was handing those to the chemist using SMILES or a 2D structure of the molecule. And he said, the molecules that the technology is proposing are so novel and so diverse that medicinal chemists are challenging me about the decisions I make using your technology. I need to show them, for some specific scaffolds that they say, no, let’s not go this route, why the system chose them.

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 chemistry or 2D perspective made more sense. You also need to make sure that you move medicinal chemists out of their comfort zone and their known chemistry.

It took us three years to get this type of feedback. Then we had to release a graphical user interface in which you could see in 3D very well why the technology was proposing these molecules. That type of feedback took us three years to get. But once we got it, it is very valuable.

Chapter 10: Ten Years and Beyond (48:21)

Imants: So it is 2022, peak hype for biotech. There is a lot of money being thrown into biotech. You are generating revenue, you have clients, you’ve demonstrated that your algorithms and tools are useful, that they improve drug discovery. Did you consider an exit at that time?

Enric: Ten years is a long time. We had conversations, both led by us and led by others that approached us. Companies working in a similar space but with complementary technology. We could have had something bigger, with a grown customer base. Some companies not even having computational capabilities, but more lab capabilities and lab skills that could be complemented by artificial intelligence and quantum mechanics.

We have had several conversations. We are open to exploring not only collaboration or strategic alliances, but even joint ventures. It is true that it’s not easy when you join forces, not as a partner, which we have several already, but if you join forces on a joint venture with another organization. There must be a very clear alignment on the culture, on the goal that you want to reach, and on the chemistry of the teams. So it’s not easy.

Long story short, in these last 10 years we’ve had conversations about doing something bigger and joining forces with other organizations, both in Spain and outside of Spain.

Imants: But you decided to still go on your own, right?

Enric: Yeah.

Imants: Do you miss working on chips?

Enric: To some extent, yes. I’m a researcher. It’s my passion. I’ve always wanted to work on chip design since I took that particular subject during my engineering degree.

I was very fortunate because at that point in time there were not many chip design companies in Spain. There was a big R&D group at the UPC, the Polytechnic University of Catalonia. I was fortunate to land there because that opened opportunities everywhere. And then Intel came to Barcelona. I miss it because that’s my background.

I miss programming. I miss doing research on that particular topic. But I chose, because of my personality and the paths of life, to move ahead in this managerial role. But people at Pharmacelera, 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 tell me that they have tried the hydrophobic profile this way. I get into the details of the algorithms, into the details of how they run, what is the benchmarking, what is the execution time, what are the hardware requirements. I will always have this in my heart. But I have chosen to move ahead in the management career path.

Imants: The first 10 years are behind you. What is in front of you? What are the next five years?

Enric: After the first generation of artificial intelligence drug discovery companies fell short with respect to expectations. I don’t want to say that they failed. They didn’t fail. Artificial intelligence is now 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 to the combination of artificial intelligence/machine learning with physics-based simulations.

Every drug discovery project is different. You need some sort of physics-based methods to characterize that particular system that you are trying to design drugs, not hits, for. In this regard, Pharmacelera is well positioned, because we apply quantum mechanics, physics-based simulations combined with artificial intelligence.

Now, in addition, there are these exponentially growing chemical spaces, these ultra-large chemical spaces that can be synthesized and tested rapidly. This is super important for being able to test hypotheses in 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.

In our particular case, we have the foundations, we have the engine. Now it’s a matter of applying it beyond hit ID. We have been working a lot on the hit ID stage, but we also have success cases in the hit-to-lead and lead optimization stages. So applying our engine in these additional drug discovery stages. And working on new modalities like molecular glues. We have just incorporated an application scientist that has experience in applying computational methods for PROTACs and molecular glues. All these new modalities and new types of targets, this is an area in which the company can grow with additional success cases.

Chapter 11: Closing (55:25)

Imants: Ten years, congratulations. I see that there is a very significant chapter in front of you. I wish you success and endurance. Maybe we’ll have you back in a year or two to see what’s the progress. Thank you, Enric, for joining me.

Enric: Thank you for inviting me. It was a privilege. Let’s see if we can talk again in two years and see where we are at that point in time. That would be interesting.

Imants: That was Enric Gibert, CEO of Pharmacelera. I’m Imants Zudans. Thank you for listening.