Build a Great Company and a Great Business Follows ⦨ Genialis CEO & Co-Founder Rafael Rosengarten ⦨ Midstage Institute

“Build a Great Company and a Great Business Follows”

Genialis CEO & Co-Founder Rafael Rosengarten: Build a Great Company and a Great Business Follows

Show Notes

The ongoing fight against cancer requires all hands on deck. Fortunately, tech startups are starting to harness the power of AI and machine learning to help in the fight against cancer. Genialis is one startup that uses AI in the precision medicine industry to ensure cancer patients are getting the right treatment.

Genialis CEO and co-founder Rafael Rosengarten recently sat down with startup coach Roland Siebelink to discuss the role his startup is playing in treating cancer patients more effectively. Rafael also spoke about the journey Genialis has been on and his personal journey as a first-time founder.

  • What it was like being a founder without a business background.
  • The challenge of succeeding with a startup that hasn’t raised a lot of money.
  • How startup founders can recognize when it’s time to pivot their company.
  • Why Genialis doesn’t consider employee headcount an indicator of success.
  • What it means to build a great company before building a great business.
  • How Genialis links its values to its strategic initiatives.


Roland Siebelink: Hello and welcome to the Silicon Valley Momentum Podcast. My name is Roland Siebelink and today I have with me yet another CEO of a fast-growing and momentous startup. It’s Genialis, and with us today is Rafael Rosengarten, the CEO and co-founder. Hello Rafael. Thank you for joining us.

Rafael Rosengarten: Hey, Roland. Thanks for having me.

Roland Siebelink: Of course. Always the first question - your company Genialis, what does it do? Who does it serve? And what difference is it making in the world?

Rafael Rosengarten: Genialis is what I call a computational precision medicine company. Let me unpack that for people who are not in the biotech space. We are focused on making sure patients - and in our cases, typically cancer patients - have the best possible therapy. The way we do that is we build machine learning models that try to understand each individual patient’s specific biology and then match their biology - or their disease biology - with the best possible available drug. And in the case where there isn’t a best possible available drug, then we work with companies to try to develop new drugs. The idea is to make sure that every single cancer patient’s disease is treated individually and is treated in the most efficacious possible way.

Roland Siebelink: Okay. Many of us have already had experience either ourselves or some family member or a loved one going through cancer trajectories, and then I think for many people this makes a lot of sense. But for those of us that haven’t, I think you’re talking about precision medicine and that seems to be the big change and evolution in treating cancer in the last decade or so. Can you explain a little bit more the context of why all this computational power is so important?

Rafael Rosengarten: Sure. I think it’s important to remember that cancer is not one disease, it’s many. And even when you think about something like breast cancer or lung cancer, each one of those is not one disease; it’s many different kinds of disease. What we in the biomedical community are starting to really appreciate is that these diseases are defined not so much by where they pop up in the body but by certain characteristics of the genes and the genetics and the RNA and the proteins and all the molecules in the cells so that a disease in one part of the body can actually be more similar to a disease in a different part of the body than an adjacent tumor might be. And so, we are interested in understanding the molecular nature of each disease.

Why is it important? It’s important for a number of reasons. The first, of course, is the human reason. Cancer is still a major killer. There just are not good interventions for a lot of different kinds. And as you mentioned, almost all of us have ourselves or had loved ones deeply impacted by one cancer or another. With that said, precision medicine stands to revolutionize medicine for most diseases. We just happen to start with cancer because we think the problem is really acute there.

But there’s some other reasons - and these are economic reasons why it’s a worthwhile place to work. Of all of the therapeutic areas, cancer is arguably the most expensive in which to develop new drugs. The estimates vary, but the estimates are somewhere between one and two and a half billion dollars per new medicine that gets to market. But here’s the thing, 96 to 97% of investigational cancer drugs that enter clinical trials fail to exit clinical trials. The failure rate is exorbitantly high.

Roland Siebelink: You’re gonna explain to us why that is, right?

Rafael Rosengarten: I’ll give you some hypotheses. By the time you get to the clinical trials, you’ve already spent at least half of that one to 2 billion, so it’s a really expensive failure. It’s a late failure. Of course, to think back to the real reason we do this, which is about the patients, you have patients who go into clinical trials - very often, they’re out of options. You’re in a clinical trial because the off-the-shelf medicines aren’t gonna work.

Roland Siebelink: It’s the last chance in a way that maybe something might work.

Rafael Rosengarten: That’s exactly right. It creates hope. It’s an opportunity cost, and so we really wanna make sure we’re optimizing this. There’s a big financial incentive. There’s a big human and ethical incentive. Your question though is why are these clinical trials so failure prone? The silly answer, the reductionist answer, is because biology is really complicated. A lot of these potential medicines, these investigational drugs do work, but for a fraction of the patients. Not for everybody. But the challenge is to get regulatory approval, you have to clear some threshold of efficacy. You have to work for a certain number of people. You have to be better than the previous mouse trap. And sometimes you’re just not going to be for that number of people.

This is where we come in. If the drug’s only gonna work for 20% of patients with a particular disease, what if we could identify those 20% beforehand, confidently? Then you can get the effective efficacy way higher. You could work for 80%, for example, the biomarker positive patients as opposed to 20% of all comers. And then you’ve got a smaller niche, but it’s gonna really work and that drug will help those people. That’s what we’re in this business to do.

Roland Siebelink: You said you’ve been in business for a while, so what is the origin story -of Genialis? What made you start it with your co-founders? What has been the story of the development so far?

Rafael Rosengarten: There’s a long version that won’t fit on this podcast, and then maybe the sanitized version. The origin story is that my technical co-founder and I were both at Baylor College of Medicine in Houston at the time. I was in what I call an Nth year postdoc, so I was on the hamster wheel doing academic research. I enjoyed the research, but there wasn’t an academic future in the crystal ball. But I knew that I wanted to do something with commercial importance. My co-founder, meanwhile, had actually just started the company in his native Slovenia. He had done his PhD in a really rich tradition of artificial intelligence. And he had come to Baylor College of Medicine in search of a biomedical problem to apply these technologies to. And we actually worked in adjacent laboratories through mutual academic connections.

I would call myself one of the first customers. He had come with a little commercial pilot project to Baylor College of Medicine. I was using the tech in my own research and totally drank the Kool-Aid. It was just mind blowing how transformative it was in my hands. Literally shaving years off of the research problems that we had to do by being able to construct an AI model that could predict the biological relationships rather than having to do years and years of laborious experiments.

Roland Siebelink: When was this more or less, Rafael? How many years ago?

Rafael Rosengarten: This was the early 2000-teens. The company emerged from stealth in 2016 as a Delaware C Corporation that had subsumed the Slovenian subsidiary. And we raised our first venture capital in 2017 and were off to the races then. But the idea was that I knew I wanted to do something commercial. I was the biologist, he was the tech guy. We had a couple other co-founders more on the commercial side. I was all-in for the potential of the technology.

Then the question becomes what’s the problem you wanna solve? And from a technology standpoint, we reasoned that the first major hurdle to the adoption of these advanced data mining technologies like machine learning and others was getting the data itself in a usable format in a useful place. The truth is in biomedicine most data are small data, not big data. They’re big in the sense that it’s hard for a graduate student to maybe analyze it in a spreadsheet. But they’re not big in a deep learning social network context.

The technology suite that we set out to build is software and some of it is a SaaS platform and others are more bespoke where we can aggregate all these smaller but really precious clinical data sets and harmonize them in a way that we can create big data from small. We can create data sets that are machine learning ready from disparate data. That was the main technological focus. And then in terms of where to point it, around 2016-2018, there was this huge wave of AI for dot-dot-dot companies, AI that was gonna do something. And this is true in all verticals.

But in biomedicine, you saw a lot of these AI for drug discovery companies. These were startups that were claiming that their technology - whatever it was - would help discover new drugs, new targets, new molecules, et cetera. And some of those have emerged very successful, in my opinion. But we reason that if that’s what you’re doing, you’re still many years away from helping patients because there are still bottlenecks along the way. Things that are just gonna be slow because there are manual bottlenecks, animal experiments, et cetera. The risk associated with a new target or a new molecule is very high. You don’t solve that 96% failure rate in the clinic. That’s where we decided to go. We felt that we had the most potential for impact to actually change people’s lives quickly by getting new drugs to market faster with much less error.

Roland Siebelink: That’s the differentiating hypothesis. It sounds like where most of your colleagues or competitors in the field were focusing on early-stage drug discovery, as you said, our impact will be much higher if we do it at the clinical testing stage. Is that correct?

Rafael Rosengarten: In 2017, that felt fairly contrarian. Today, I would argue that the entire diagram of drug development from the earliest stages to the last stages and on to the market are all - I’m not gonna say saturated - but they’re all very active in these kinds of technological approaches. Frankly, all for the better, there are 9,000 diseases that we don’t know how to treat or cure. Let’s get after it.

Roland Siebelink: Absolutely. To continue a little bit on the business story - your technical co-founder and you a biologist, as I understand, you were not really experienced in business, trained in business. How did that journey treat you?

Rafael Rosengarten: It’s been pretty brutal to be honest. I liken being a startup CEO, especially as a first time founder, which I am, it’s a bit like being on a roller coaster, but with a very short periodicity I feel like every day I spend half of the day hiding under my desk and the other half standing on top of it celebrating our small victories. When you’re a small startup, especially when you haven’t raised mammoth amounts of funding - and we’ve done okay but not huge amounts - a lot of things feel existential. Losing one big account feels like an existential threat. Being late on a product release can feel existential. It’s been a lot of trial by fire.

I will say that when I joined as a co-founder, I wasn’t the CEO initially; one of our other business co-founders was. But he raised our initial capital and had a particular vision for the company. When we switched roles or when I took over, we did a pretty hard pivot. And for anyone out there who’s pivoted their company, that can be a pretty painful experience as well. You feel like you’re in a dark wood for a while until you regain that traction. But that was okay. We went from running out of cash and pivoting to break even and profitable in less than a year, about a year. Once you’re break-even and profitable, then you’re the master of your own destiny and you can start doing things again.

Roland Siebelink: Absolutely. Maybe that experience of the pivot is worth delving into a little bit more - to a degree you can share, of course. In retrospect, can you talk through what were the drivers that absolutely made it imperative to pivot? And how could you recognize them?

Rafael Rosengarten: Ultimately, the reason why we pivot is we couldn’t figure out how to make the unit economics of our product and business model make sense to us anymore. And I think the reason why is because we had not been entirely - I don’t wanna say honest with ourselves -we had overestimated the TAM of our first product. What it really was, we didn’t overestimate the TAM. We were impressed by a TAM that wasn’t really big enough.

Here’s my advice to early, first-time founders, you’ll hear from people that you need a TAM of at least a billion to raise venture capital. Nonsense; 10 billion or 100 billion, if your TAM isn’t 10 to 100 billion - orders of magnitude more than 1 billion is too small. Because you’re not gonna get that much of it at first. I’m happy to dive into it. What we went through is actually not so dissimilar from a lot of my colleagues who are in a similar space.

A lot of us started out trying to monetize our platforms using a SaaS business model, signing up either drug companies or diagnostics companies or even academic centers as subscribers. We still make money from that business model and that first product. The beauty of SaaS, of course, is once you get sticky customers, it’s pretty passive income at that point. It’s a high margin. That’s why people love it. That’s why investors love it. It’s why acquirers love it.

Again, go back to the mid-teens, you’ll remember venture firms like Andreessen Horowitz had just gotten into the biotech game. They had made an absolute killing building the first generation of SaaS companies in other verticals. But in other verticals that were either B2C, where you had billions of potential users, or B2B but much more general business models where you had millions of potential customers. There are only 5,000 pharma companies in the world. Now, a lot of them have a lot of money. Today, a lot of them don’t cause the market’s pretty crappy. But the point is that it’s not a huge market in terms of total number end. You’ve gotta have a real premium price if you’re gonna make the SaaS business model work there.

A few companies have done it well. There are some unicorns. There are some really well established legacy companies that have made a good living and built really valuable companies selling SaaS products into biotech. I would argue that it’s not necessarily harder than other verticals. But you’ve got market physics against you.

Roland Siebelink: What’s the alternative? Is it traditional enterprise software? Is it a package of consulting and software? What have you found works better in this context?

Rafael Rosengarten: I will say that I think it’s still evolving for most companies. I think the business model around how to monetize, how to capture value around machine learning in that space is not totally figured out. But there are a number of things that we’ve seen. In our case, we funded the pivot with technology-enabled services. And from that, we built a new technology product that we monetize through a combination of licensing. You license this technology, but it’s a premium price product. It’s a zero-to-one thing. We have the SaaS component. And we’ll do consulting on top of it, so it ends up being a very mixed revenue stream. Now what’s interesting is after a while the companies like Andreessen Horowitz that were really eager to push SaaS business models into biotech recognize that you leave a ton of money on the table if you don’t do those consulting services.

I think it just took a while for startups and the venture community to realize that that was gonna be true for this new generation of tech bio platforms. The evolution since then has really moved toward owning IP and becoming what we call a NextGen biotech itself. So, 2019 into late 2021, the biotech market was super frothy, super overvalued. But it was an opportunity for a lot of these technology companies to start owning some of their discoveries themselves and basing their valuation around that intellectual property.

Now we do that too. It’s not four, it’s really six revenue streams or six components to the valuation. But our key focus - and I know one of the things you like to dig into is what’s the key things for us to achieve in the next quarter, six months, a year. We are building what I think are gonna be important diagnostic models that we are building on our own, on our own dime, using proceeds from our last financing, using our global network of academic and clinical research partners. And with those models, we are gonna take them to both the pharmaceutical industry and also the diagnostics industry, and partner them out as essentially digital assets.

Roland Siebelink: Moving back a little bit to the development of the business, how big is the team now and how have you decided to split up resources between different key functions of the business model?

Rafael Rosengarten: We’re about 30-ish, and we’re growing. We’re hiring for a few key senior management positions. In business, we just this week brought on a new director of operations to build out our growing Boston office, and so forth. We’re hiring some in R&D as well, so I anticipate we’ll be 35-ish by Q3.

I don’t view headcount as a meaningful indicator of success. In fact, there are folks - I think it’s Paul Graham, he gets credited with a lot of quotes. Paul Graham of Y Combinator who has an essay on this; there’s some quips on this. You actually want to have an inverse relationship between your headcount and your impact. You want people to come to your office and be like, “How the heck did you build this with only 12 people?” Instagram was a 13-person company when Facebook bought them. Something like that. That’s the goal.

Roland Siebelink: WhatsApp was 29 people, I believe, when they were bought. Both similar stories. Absolutely.

Rafael Rosengarten: The point is I love our team and one thing we haven’t talked about is company culture, which is something Genialis really emphasizes and puts a lot of effort into. Growing big in headcount is fine if the business needs it, if the business supports it, if that’s what’s required to have the kind of impact you want. But if we can figure out a way to achieve massive scale of the business without having a massive scale of the company, all the better.

Roland Siebelink: You mentioned your company culture as a very differentiating factor as well, Rafael. Talk to us about how consciously you build a company culture and what makes it different from some of the other company cultures you see out there?

Rafael Rosengarten: Our overall attitude within the executive leadership but also the old team is that we want to build a great company before we build a great business. Or to say it differently, a great business will follow from a great company. We want to build a place where people really enjoy getting up and going to work or enjoy getting up and staying home to work, or enjoy taking their RV on a road trip where they can get their work done from the far corners of wherever.

We were hybrid before Covid. We’ve embraced a remote culture all along. We have two key geographies, although one is much bigger than the other. I mentioned earlier my co-founder’s from Ljubljana, Slovenia. A couple years after we started the company in Houston, we went in different ways. My wife got a job in the Bay Area, so I went there with one young kid and planning to have another. My co-founder had a young kid who was born in Houston and wanted to raise his kids closer to home, so he went back to Slovenia.

I am a trailing spouse in our relationship, so I figured I would always be on the move and it didn’t make sense to anchor the office around me, so we anchored it around Miha. We’ve got a big team in Ljubljana and what I would call a satellite team all over the world but growing in Boston. Boston is our commercial hub. And that’s the office I call my home office, although I like to joke that it’s really the star alliance lounge at whatever airport I happen to be transiting. That’s the first thing; we have these offices but we also encourage people very strongly to figure out how to best live their lives. And work is a part of it, but it’s just a part of it.

We have four key company values. Most companies have values and a lot of startups go through the exercise because it feels good or you want to have something to say when you have to hire people. But we really impress these and we try to link all of our strategic initiatives to these as well. When we do recognitions at team events, we always link the recognitions to people who embody the values. Those are - in no particular order, but kind of a particular order - people first, so putting people first. Again, these can be felt or construed in different ways by different people. But the key idea here is building a company where people feel empowered to live their best lives, whatever that means for them. And then the others are ownership, so we want everyone to own their corner of the company; you are the domain expert in the thing you do. You’re trusted to be that expert and to deliver, and you’re trusted to deliver with the interest of the company in mind. Constructiveness and honesty are the last two.

There’s a really smart paradigm in coming up with company values or any values that they’re not useful unless you could take the opposite tack and argue in favor of that. It has to be something that you have to argue for.

Roland Siebelink: It’s a real choice that way. There should also be other companies that make an opposite choice, and that could be equally valuable.

Rafael Rosengarten: That’s right. Honesty is one that skirts that line where loads of companies talk about integrity. Of course, no one’s gonna say we’re gonna lie. No one says dishonesty. But there are other contrary positions to honesty. Some of those are things like diplomacy could be contrary to honesty. The customer’s always right is a great one. That’s a huge one in the service industry. The customer’s always right. Nonsense, especially when what you’re doing requires scientific integrity.

Roland Siebelink: Exactly. Or do whatever it takes. There’s a value that many companies have that also can go against honesty.

Rafael Rosengarten: Absolutely. We really live those. It’s challenging as executive leaders to be able to walk the walk sometimes and say, “I’ve got some time off now and I’m gonna take that time off and please don’t schedule meetings during that time.” The business keeps going, but you have to set that example.

Roland Siebelink: Exactly. Because in the end, it’s a marathon, not just a sprint that you want people to be ready for, right.

Rafael Rosengarten: It’s gotta be. And it’s also thoughtful work. It doesn’t make sense to me to grind all the time. You have to take time to reflect.

Roland Siebelink: Exactly. I love that. Ending, people who have followed this episode all the way through here must be very interested in Genialis. How can they help you? Where can they find out more? Is there something specific you’d want them to download or to look at?

Rafael Rosengarten: Obviously, check out our website:( We’re building a lot of content that’s mostly going out on LinkedIn these days. I’m not entirely sure about the status of Twitter for our audience anymore. But we’re really growing the audience on LinkedIn, so you can find us there as well. Pay attention to our travel and our talk schedule. We also have a podcast, it’s called Talking Precision Medicine. You can find it on all the major podcast services. That’s Talking Precision Medicine. Those are the best ways. Take a listen and shoot me an email. My email is just my first name and. I am hopelessly addicted to my inbox, so I’ll probably respond.

Roland Siebelink: That’s great. And of course, anyone interested can also get an intro through me if they want to get to know Rafael. Thank you. Rafael Rosengarten, the CEO and co-founder of Genialis. It was such a pleasure having you on the podcast.

Rafael Rosengarten: Roland, thanks so much. This was great.

Roland Siebelink talks all things tech startup and bring you interviews with tech cofounders across the world.