Modernizing the Lab with Nathan Clark Cofounder of Ganymede

 

Kelly Stanton: 0:18

Hello, and welcome to another episode of From Lap Launch by Polio. We're at almost 70 of these interviews now with the movers in Shakers and life sciences. It's been a delight to share behind the scenes stories from some of the most innovative and advanced companies from around the world. I'm Kelly, your host. Before we jump in, just a reminder to please rate the show and share it with your friends for science nerds just like us. We know you have some, and also check out the show notes if you have a story or a product you want to showcase on our on our podcast with us. All right, so today we're talking with Nathan Clark, co-founder of Ganymede one of the fastest growing startup biotech's, software startups in biotech that helps automate and integrate basically everything in a lab. Before Ganymede Nathan was a product manager at Benchling, which is another darling startup in biotech we've had our eyes on, as do many of our customers. You can read his full bio in the show notes, but we're excited to have him on today to talk about the lab of the future and how that will look and much more. Thank you for joining me today, Nathan. Tell a little about Ganymede

Nathan Clark: 1:19

bit about Yeah, thank.

Kelly Stanton: 1:22

That's an unusual name.

Nathan Clark: 1:24

Yeah. Well, so I love space things and every company I've been at as a result is littered with various projects named after space things, you know, project, project in drama all sorts of things. So, uh, when I started Ganymede, it basically had to be a space thing. Ganymede's a nice one cuz it's a little bit more niche, one of the Jupiter's moons. And if I went to the.com of it, there had to be nothing there. So, that was the selection criteria. And here we.

Kelly Stanton: 1:49

That's awesome. But uh, you didn't start your career in life sciences, so tell us a little bit about your path, from doing machine learning at a firm to biotech now.

Nathan Clark: 2:01

Yeah, my background originally was in finance. I started out actually originally as a bond trader. And then went to firm, the Buy Now pay later company originally on the capital market team there and switched over to the product management side working on data products and financial products. And so in a way there's a lot of overlap, certainly with The medical tech, biotech life sciences fields, both very regulated both very data driven. And there's also some interesting contrast, but I have always loved biotech. I, I love finance academically, but I've always felt drawn to biotech. So after a firm, I made the decision. That I just wanted to get into the space. I wanted to learn about biotech and I had done some study in biology and volunteering on the side, but there's no better way to do that than to just go work at a company in the space that's great than so Benchling was a, a, a great place for me to go to learn get experience and I really enjoyed it.

Kelly Stanton: 2:53

So you've talked some about digital transformation in biotech before. Share how you believe labs today will adapt to new.

Nathan Clark: 3:01

Yeah, it's a good question. I think there's a few different angles that especially at the earlier r and d stage, people are trying to digitally transform through. There's been a lot of investment in the last few years in robotics, liquid handler lab automation. When people say, Lab automation they oftentimes mean liquid, liquid handler based work cells, scheduling software for that tying different systems into those work cells in an automated way. And that's that is a big area of focus right now. And I think it's seen a fair amount of success. But what we've seen is fairly neglected at Ed. The rest of the lab, you know, what are scientists still manually doing? Where are they data entering things? Where are they interacting physically with analytical lab instruments? You know, what are they doing in Excel? That is the, the, the kind of the glue that holds together the rest of the biolab between automated robotics and the rest of the actual operations that people have to do of qc, manually pipetting things. And so, In a way, a big inspiration for GME is to say, well, what would it actually take to integrate that and automate that? Like the things that scientists still have to do with their hands. That's a case where the robotics exist. There's just nothing that you can't program the software to figure out how to actually replace that because scientists are evolving too quickly with their experiments. It needs to be something where you're integrating the data. So, long way to say, I think that labs will become more and more data centric because you need. Unravel the software side of the data and really understand what you're trying to achieve and structure from a data perspective in order to bring automation earlier or earlier into r and d and also broader and broader in the late stage environments where there's some automation but ends up in these islands that are not glued together. And ultimately you end up still having these humans involved to actually make the loop close. I think bringing a more data centric approach to labs is really e. Definitely. Yeah.

Kelly Stanton: 4:51

As someone who started her career in the laboratory and remembers when our data sheets were carbon copy printed with numbers in the upper right hand corner. goodness. Yeah. In your now I've been away from the lab for a few years. Now, and obviously I also remember, of course, the Excel transformation, right? Like, here's this power group tool right there in the lab on everybody's computers, which of course then became a very big regulatory thing, right? Had to lock your spreadsheets and all that kind of stuff. But based on what you're seeing now, how much of lab work today do you see a still manual with things like Excel spreadsheets and, and manual handling and that sort of thing? I would guess the majority of it. But what, what,

Nathan Clark: 5:33

say, Yeah, I definitely agree. Certainly the majority. I think that there, although, you know, that work on robotics and automated systems has pushed into some areas still, you know, there's so many angles that cause things to still be manual, even within those companies, even if you're doing the most standard like. Small molecule drug discovery or genomics driven workflows. Even if there's robotics, you're still gonna have a ton of scientists actually then taking the targets that are generated by these screening operations and validating them, working through assays to study them under training the parameters. And then anything that doesn't fall into that standard mold, especially in areas like biologics. Biologics like things I like to say, like lab grown. Or cpg kinds of things or other ingredient production. All of those things also become way too complex to actually. Lockdown. And so it inevitably leads people to fall back to Excel or even sometimes paper. I think companies in the early r and d area, companies like Benchling have done a great job offering a really great Elan, so at least people are typing on structured data into that. But it can be a challenge to update the data structures. For capturing structured data quickly enough to capture the evolving process of r and d. And I think the same applies even at the higher scale manufacturing side, electronic batch records manufacturing execution systems qms, it feels like, You can set it and build it and get the data structured, captured. But then as processes evolve, there's this continuing challenge of trying to say, okay, well, you know, you restructured your manufacturing line. How do I now update my batch records to reflect that? Mm-hmm. yeah, So yeah, Excel's always the fallback. And I think things need to get into those structured tools. It's essential that stuff gets into apps like mes, qms, e lens limbs. And one of the things GME hopes to do to help that is to power these integrations by saying, Hey, rather than just falling back to Excel to do your manual calculation and then putting it into the app, maybe we can offer a place to build effectively an integration and more quickly adapt the actual logic running in the cloud. To reflect these processes as they change. We wanna be the perfect tool for updating these integrations and writing them in and be that infrastructure for integrating apps together. And of course lab instruments, manufacturing machinery as well. Cuz that's oftentimes the data source. But we see that the challenge needs to be solved by having really flexible tools for connecting things together so people don't feel like they have to fall back into.

Kelly Stanton: 7:57

Definitely, definitely. Well, and the other thing that's sort of spinning in my brain now too, cuz again, we're an electronic quality management tool here at Klio. But people's resistance to Making these processes more automated because of change management, right? Like in r and d, you know, I've always joked you know, you're writing in pencil, you're not writing in pen. Once you get into design control or into your like i n d development phases, now you're writing in pen right now. Things have to be locked in under change management because we always have to justify why this change doesn't have any negative impact on blah, blah, blah, blah, blah. Right. And so from an industry perspective, it's kind of scary, but it sounds like you guys are really thinking about that already. You know, and so I guess, you know, can you walk us through a little bit about exactly what GME does, you know, to kind of solve some of those issues? You were talking about it some there with the integrations, but can you expand on that a little bit for those who aren't?

Nathan Clark: 8:52

Yeah, absolutely. So GME ultimately is a cloud platform for integrating data from complex data sources that involve physical processes, which you know, there's no place that's more complex and more physical processing than the life sciences. And so we really focus on that right now. Getting that data in, storing it in a structured database like format and storing it forever which helps a ton with compliance and gxb and then pushing it to where it needs to go. So we try to be, you know, we really studied how big companies as they grew, would end up building out their. Data backbone and sort of their production software backbone, that would actually unite the many apps that they have. Cuz as you grow, you end up developing so many different apps that need to be able to talk to each other. So many different data sources in your lab. These companies will end up building their own data lake and their own ways to kind of merge and move around data and transform it to make it work. So GME is meant to be that in a box and hopefully bring that to earlier stage companies and make it easy for them to not worry about the infrastructure. So they can just say, Hey, you know, I have data scientists that are writing code for my. NGS pipelines maybe, or for my lab automation systems now they can also just write the same code to automate the wet lab and connect their data together into their wet lab tools similarly for manufacturing. And so yeah, that's, that's what we do is we provide the cloud platform, but it's, the cloud platform is really meant for developers. Whether those are our developers doing it as services or client side developers. They can write that logic to connect directly to their lab instruments in our platform. And then so we oftentimes what that's used for is lab instrument integrations integrations into tools like quality systems, ELNs, et cetera and moving data between them. We have a pretty big library of those that we have out of the box at this point. But the secret sauce is the fact that it's fully self-service if you want. And so that helps. A, because then you can write your own integrations for free. And then b. When inevitably you need to adapt the integrations we have out of the box to work for your setup, or, you know, the instrument changes over time because the manufacturer releases something new, then it's literally just a one line of code to update it rather than having to call us up.

Kelly Stanton: 11:04

Nice. Well you touched on this a little bit earlier too, but what are some of those challenges that you're seeing with bringing the technology in? I mean, obviously there's the regulatory side of it and oh my gosh, we have to validate all of this. Right. But what other challenges are you seeing around

Nathan Clark: 11:17

this? Yeah, I think certainly the regulatory one is a, a big and interesting one, and there's of course a lot of movement right now on the FDA side around how they think about computer systems and data systems in a, a regular context. Yeah. New G version. Yay. Yes. And I think that that's great for us in a sense because we do try to have an opinionated stance of, Hey, we'll store everything forever. But on the flip side I think a challenge that we see and others see is, okay, well, you know, you start recruiting so much data, how do you make sense of it? And that's where inevitably I think you can bring as much automation and great data tooling to bear as you want, but it won't help unless companies have an understanding themselves from a business perspective of how they wanna actually structure and think about their data. Like what is your sample management process? That's always the golden question somewhere around the, the 50 person stage when you're flipping to that imd kind of filing mindset that you mentioned. And I think. Be a lot of pain to, for companies to figure that out. I would love to see more standardization in the industry there and see sort of that mindset of it. I don't know if you wanna call it quality or data structuring it. There's this broad thing it feels like of how companies mature and trying to bring that a little bit earlier and have people more map their processes into that rather than, Going wild in the wet lab in the early r and d days. That feels really necessary. And I feel like that's definitely probably the biggest challenge that I see right now is the tools are getting there. I like to think that we're a good tool for it. But the flip side is companies will have to learn how to use them and learn how to adapt their business processes to be usable in that format. Where

Kelly Stanton: 12:53

we have that same challenge, you know, with that whole CSA update, right? Getting people to think about their processes and then how this tool supports those processes. Because ultimately that's how you determine whether or not your tool is fit for intended purpose, right? Yeah. You have, there's this expectation that you have this base understanding and so, In the startup space, right? You know, it's like, well, you need dock control and you need training, and you need change management and you need supplier management. And, and if you ask 10 QA people, you might get 10 different answers about what that looks like. But the reality is that the required elements are the same no matter, you know, whether you're a 10 person company or a hundred person company, where you are in that process. So it's interesting that you, you look at it from that process perspective, because that really is. I think the goal of the FDA's thinking about trying to shift us away from, did you document this? Check the box. Right? It's like, so what? Yeah. What's your process behind it? Because yes, the output is inspectable that you've checked this box from a regulatory perspective, but really if your process is well built and it really can enable you to take all of that data to the next level so quickly, I feel like people miss that.

Nathan Clark: 14:08

Yeah, absolutely. And I feel like in a way you know, there was this big trend five years ago or so where everyone and again, this is kind of my r and d world view, but everyone was so focused on these workflow based limb systems. Trying to say, oh, let me just breathe out into the ether what my scientist processes are, and then I'll like no code, stitch it together and have my exp my protocol just reflected in this nice flow diagram. And that's an honorable notion. Like, yes, it would be nice if we could achieve that, but I feel like what people have really learned in the last few years and a wave that we're riding is the more data-centric view of, it's not about. The actions that people are taking. It's about, like you said, like what are the outcomes? What's the, the process and the framework that the actions are happening within. Like what is the actual data structure of all of this? And it feels like on the r and d side, people are moving a little bit more towards this. What's the, I wanna set up the backbone of my data and figure out how to plug into that mindset. And it sounds like from what you're saying, that's. Definitely a big trend on the regulatory side as well. That it's not about check the box, did you do X? It's more like, what was X? You know, uh mm-hmm. what does that mean to you? How do we make semantic understanding of that? And it really goes together. It

Kelly Stanton: 15:19

does. It does. And it's a huge. That's, I feel like the big shift, you know, again, and, and it's interesting to hear your perspective on it from a research perspective, because I spent most of my time on the commercial side, but then some in my later years before I came to polio was in the development and the, and the product development space. And, and so I, I do, I think, you know, the. It's a, it's a philosoph shift, in how you look at it across the whole industry. Not just, you know, Hey, here's another software tool that's is gonna make your life easier. Like, you really have to shift how you think about the entirety of the process. And of course, you know, the FDA's focus being on product quality and patient safety. So if you're thinking about it from that perspective, then you really can scale up or down how much of that regulatory checkbox work you have to do. And so getting your brain over that hump of, you know, I'm not gonna check all these boxes, that's, that's what I see as the biggest challenge to the adoption of all of this kind of. New technology that's that we're seeing now. And it's interesting too, I, you know, just to tie in big buzzwords, right? Right now the hot thing everybody talks about is machine learning and ai, right? That's every, I swear, every and half. It's interesting. Half of that GAMP standard goes into, you know, some, some more sort of thoughts and thinking around those buzzwords, if you will. But are you seeing that sort of. Tying in here, or are you guys thinking about it from that perspective? Or maybe that's just a buzzword that doesn't encompass accurately what you guys are trying to do.

Nathan Clark: 16:53

I think it's actually it's a funny thing because I, I have not mentioned machine learning or ai throughout this conversation yet. Even though, you know, my background is very much from it. Working in finance, working in, I built out a machine learning team at Benchling. We I think it's very tough and the issue that I see in the space is that the data is so fragmented in biotech and pharma and in generally any, you know, manufacturing anywhere where you have these complex processes. In other industries like finance, where you do get pretty good application of machine learning, at least the data structure is already sort of known and you have to clean it up. But you know, if you're saying, oh, I'm gonna do machine learning for underwriting, someone for a credit card, you kind of know what a credit card charge is. You can anticipate that and what the data structure is. But here, gluing it together is, is so hard. And so I feel like that's, we do hope to, you know, we are very much focused on trying to enable AI and ML in the industry. And our, our platform certainly can and does run a lot of AI and ML models, but it's the kind of thing where the modeling is only like 10% of it. In the end, the 90% of it is how do you get the data there into the right shape? And so it all. We say machine learning and ai, but I feel like what that really means, it's the same thing again, of just how do I get the data in one place, in a harmonized, you know, fair workable format. And so I think that's our angle that we take, is we wanna, on the side, we wanna help get the data there. And then ideally, you know, because the data is so fragmented, there's a lot of low hanging fruit. Once it's there, it'll actually be quite easy to do machine learning on it. The grand challenge is the data, and I feel like every organization that I've seen that's investing in this, they come to the same conclusion where they say, Hey, you know, wouldn't it be great to have this. Machine learning, predictive dashboard. And yes, of course it would be great, but that takes like, a few days to build once the data's in place and the data in place takes months or years of digital transformation. So, yeah, as, as someone who's definitely been practicing a lot in this field, I feel like when people say machine learning and ai, what that means to me is data assembly and cleaning. And I think that is something that people are gonna have to get more comfortable with. Yeah, definitely,

Kelly Stanton: 19:00

definitely. We have the same challenge with quality systems, right? I mean, and, and I love, you know, this again, brain spinning from a QE perspective, but you know, it's, it's a similar challenge just in your whole quality system, right? Is is having, cuz you do have all these discreet data sources and they don't talk to each other. So, it's interesting too to think about data mining in a laboratory cuz that's, that's huge. I mean, there's, there's so much potential there. Wow. I, I get excited about that. Yeah. So I'm gonna pivot a little bit here. We love asking startup founders how they got the first hundred customers. So just curious, how are you finding customers and building awareness as a fast growing startup and, and especially in today's economy, things are a little challenging in the startup space. What are you guys doing

Nathan Clark: 19:45

around. Yeah, I would say especially given the industry now and the market conditions, there's really no substitute for a really strong. Ironclad enterprise sales backbone. We I think from day one have taken that approach of trying to build relationships really, really deeply understand what clients are doing. You know, I I know in pretty fine grain detail exactly what every single one of our clients is trying to achieve with ed. And I think that helps us find ways to offer the highest leverage solutions there. When the markets recover, I think it'll be a little bit easier to go out with a product. Kind of sales motion and say, Hey, you know, here's our self-service package. You have the money to budget. We don't need to persuade you you know, just you know, you're ready to go. But right now, I think when people are really in this hardcore cost benefit mindset, given constrained budgets, I think you really have to help them understand. Both the value of the tool, but also I think what their overall software stack should be and how they should think about investing their entire budget. So I think we've become very consultative on the sales side. We go to a ton of conferences. We do a ton of networking and also try to also do a good job with our existing customers to expand within those customers. And I'd say right now, yeah, there's no substitute for that. Spamming people and cold emails, all that kind of stuff ads and things, that's, that's not gonna cut it compared to building these lasting relationships and really helping people understand how the value that we deliver can be quantified.

Kelly Stanton: 21:11

Definitely, definitely. Yeah. We, we have a similar challenge in the startup space. It's hard to mm-hmm. help them see how the investment now is going to pay off in the. And having some of the infrastructure, you know, through either, you know, some, some good lab data management practices, a good quality system product, right? And having those things all in purpose-built tools from the beginning is really gonna speed them up later. It's, it's hard to, that's a hard argument sometimes to, to work through

Nathan Clark: 21:41

Yeah. It's so abstract. But it's the future, so I feel like people will get there eventually and. We're, we're both probably lucky that we can at least ride that wave and offer really strong, modern solution. Yeah, definitely. Definitely.

Kelly Stanton: 21:54

Well, if you could go back to the start of your career, what would you tell yourself based on what you know now?

Nathan Clark: 22:00

Oh man, that's a big question.

Kelly Stanton: 22:03

I love the answers to these questions, especially from people who pivoted from one industry. Then actually it's always so

Nathan Clark: 22:08

fascinating. Yeah. You know, I would actually say, although, you know, I was in, I was a bond trader so I was very comfortable in a way with taking risk and understanding risk. That's kind of the job. But I would actually say to take more risk with myself. I think I've always been interested in this direction of trying to help make an impact in the biotech and industry and healthcare and medicine. And it felt inaccessible to me in a way. But I became comfortable with that over time as I grew and I built out my network and I realized, hey, I could actually start building out technology here. I can code, I can go do what is needed. But I could have done that back then too. You know, why did I need to go into. Finance direct. I mean, I liked finance, but I think it's hard. I think it's hard to really anticipate the ability that people have to, I think to just go jump into a space and start making an impact. And have to be sustainable. You know, it's not like I'm quitting my job or derailing my career or to do what I want. Like this is my career now. It's a it's been great. And so, yeah, if I could, I would go back and say, Hey, you know, Focus a little less on just the, the career growth and focus a little more on the pursuing the dreams as, as trite as that might sound. You know, it can be your career and it can be fine. Like anything else, if there's something that you wanna make an impact and just go start working on it you know. Someone has to do it. So here we're, yeah. Yeah. Definitely,

Kelly Stanton: 23:27

definitely, definitely. Another fun question that we've we had somebody on a previous podcast ask and, and love this one, but if you walked into a bookstore or if I walked into Barnes and Noble, where would I find you? What section?

Nathan Clark: 23:43

You know what I really love? I don't know if this is a section so I may be cheating a bit on the question, but I've always loved those, those cut out cross section books, like the ones where you open them up and there are these big diagrams of ships or machines or I remember there was a Star Wars one where it had like a cross section of the, the death star and they just made up what was inside it. Yeah, they had these really fine greened. Models and details that they just completely made up for it. So I love that kind of stuff. It's so cool. It's like a where's Waldo, but with machinery or stuff like that. Oh,

Kelly Stanton: 24:12

definitely. And a physical manifestation of things that are maybe imaginary. And so it's a, it's, yeah, that's, I love those too. Those are a lot of fun.

Nathan Clark: 24:22

Yeah. Well, thanks so much. Check out the Star

Kelly Stanton: 24:23

Wars one. Yeah. I will have to check out the Star Wars one for sure. That's yeah, we're, we're total Star Wars nerds around here. Awesome. Well, thanks so much for joining us today. Where can folks go to connect with you and follow along on

Nathan Clark: 24:36

GA's progress? Yeah. We're at gme.bio. Feel free to reach out to us. We have our contact form there always. Of course. Email me. I'm Nathan gme.bio. We have a, a LinkedIn, our Twitter account, but, you know, just hit us up on email anytime. And I, I'd love to chat. I love to talk about the industry and learn from people. So, yeah, always looking to, to chat and very open to connect.

Kelly Stanton: 25:01

Awesome. Thanks. Thanks for joining us today.

Nathan Clark: 25:04

Yeah, thanks for hosting me.