More Biology From Every Microscope featuring Ivana Mikić

How did butterflies inspire seeing science more clearly? Find out today. Welcome to The Connect Current, the flow of innovation in San Diego. We bring you the founders, investors, and leaders building the future across life science, tech, climate, defense, AI, and more. This is where San Diego's next chapter takes shape. Welcome to the Connect Current, the flow of innovation in San Diego. I'm your host, James Zanewicz, and this episode is being recorded at Kiln in Rancho Bernardo. When you're looking for coworking space, look no further than Kiln. Today I'm joined by Ivana Mikic, PhD, CEO of Sarpeda, and one of Connect's 2026 Cool Companies. Welcome to the show.

Well, thank you for having me. We're excited to be included.

So great to hear more about you. And I want to give people a little bit of background because Sarpeda is building AI-powered software for fluorescence microscopy that helps researchers extract more biological information from microscopes and workflows that are already in use. The company's platform is designed to increase the information capacity of biological imaging, improve signal quality, and reduce really common imaging challenges like background noise, autofluorescence, and channel bleed-through. This makes it a great Connect Current conversation because Sarpeda sits at the intersection of AI, life sciences, imaging, which is also tech, research infrastructure, and commercialization. You hit all our sweet spots. And it's not just about building a more advanced tool, but it's about helping researchers see more clearly, use existing systems more effectively so you're cost-effective, and moving discovery forward with less friction. And as a 2026 Cool Company, you're part of the next wave of startups being highlighted to investors, showing how deep technical insight can become practical, scalable innovation. Are you ready to dive in?

Wonderful summary.

All right, let's go. So, Ivana, for listeners meeting you and Sarpeda for the first time, what are you building? What problem are you solving? And why does that matter right now?

So, as you beautifully described, we build AI software for fluorescence microscopy and help scientists pull far more biological information from microscopes they already own. And why does it matter right now? Biology is entering a new era of scale. AI can dramatically accelerate life science research, but only if you feed it massive amounts of high-quality data. And biology has never had that. That's changing. We can now grow highly relevant human biological models Organoids, organs-on-chip using patient-derived tissues. And companies are scaling these systems to test thousands of perturbations and drug candidates in parallel. But there is a massive bottleneck in that sensing. If you test hundreds of cancer treatments on hundreds of tissue samples, you have to actually see what happened. And fluorescence microscopy is one of the few ways to do that. And it has hard limits, a handful of channels, heavy background noise. And that's where we come in. Our AI cleans up the noise and dramatically increases the biological information you extract. No hardware changes, no new reagents. So labs can scale biological discovery much faster using infrastructure they already own.

So when you talk to scientists or researchers, where do they feel this bottleneck most acutely, kind of in their day-to-day work? Like, how does it affect them negatively?

So they have to make hard choices about what to capture because they can't capture all that's in there, or they have to repeat their experiments and do an image, image different things in different replicates, which increases the cost and time.

Yeah.

We've just talked to a CRO who says that they prepare 2 pieces of tissue for every piece of tissue that they actually need to image just so that they can calibrate exposure time to try to minimize some of the background noise. And we just eliminate that friction.

That's fantastic. You're eliminating friction for imaging the same way Connect is trying to eliminate friction in the ecosystem and make it more efficient. But you spent a whole lot of your career— your background is at the intersection of AI, imaging, and biology. So you were ahead of the curve where biology is all about AI tech and data now, but you were always doing that. So what did you see over time that made you believe microscopy had that major information gap waiting to be solved?

Yeah, so Pieter is a combination of my life's work. I've spent over 20 years at that intersection of AI, computer vision, and biology. I had 2 successful exits along the way.

Uh-huh.

So I led— most relevant to this discussion, I led AI teams in digital pathology, where we image and analyze tissue samples day in and day out. Then I led an AI team in an organ-on-a-chip company, where we grew, imaged, and analyzed lab-grown tissue. And I lived the problem Serpida solves every day for years. Not enough channels.

Mm-hmm.

Too much background, and really fluorescence microscopy leaving biological information on the table that was in that biology, and we just couldn't get it imaged. And through Define, my consulting company, I saw the AI system that our research team had developed for a different problem, and I realized it could solve these 2 problems. that I had run into for years. The photons hitting the sensors carry far more information than we capture, and AI can recover it. And so I realized that this technology can solve this problem and revolutionize fluorescence microscopy and decided to start Serpida.

Well, and it's really interesting because from digging in and doing my research and the research our amazing Chief Brand Engagement Officer Trey did for us, your approach does not ask labs to replace their microscope. buy expensive new hardware, or rebuild their workflow. But your technology unlocks a lot of potential. How does it make those systems more efficient? How does it work?

Yeah, so our technology lives between the microscope and any downstream analysis. And so you don't need new hardware, reagents, or to change your workflows. And it rests on 2 insights. One is that the biological markers are sparse. Only a small fraction of pixels in the images actually have a meaningful value. We historically devote entire channels to each of these markers. And so our product, SarpedaPlex, basically allows you to put multiple markers into the same channel and then separate it computationally. And the other insight is that the background signal, as disruptive and strong as it is, is actually predictable. And biologists every day capture what we call negative controls, which are images of only the background signal before they stain them, just to calibrate their own eyes to what is real and what is noise. And we realized that we can train an AI system on those negative controls to clean up the systems. And so we just sit between without requiring any changes to hardware or workflows.

So if someone's non-technical out there or they're not an imaging person, how should they think about the difference between simply improving an image or extracting more biological data from it. What is that difference?

So a clever experimental setup allows you to extract more information. So we're basically telling biologists, look, you can put the same fluorescent molecule on multiple markers where you had to, in the past, only put one per marker. You can now put it on multiple markers, and our AI can separate them. So it truly gives you— allows you to pack way more information into a single channel. And when you think about it, we've been cramming our wires and air with so much more signal.

Yeah.

And here in microscopy, we give this precious channel— we only have 3 to 5 in most microscopes— we give a whole channel to one signal. And so now we are not just improving an image a little bit. We are truly packing more information in there. And for background removal, It's important also to know that this is not some small improvement to background subtraction algorithms that are out there. This is a true separation of a signal that's in there that can be recognized.

That's incredible. And it's in large part because of AI, which is being talked about everywhere. It's the buzzword of all things. But in science, it has to be more than just a buzzword. It actually has to have applications. So in your view, what does responsible— or what does a responsible, useful AI tool look like? when you're applying it to imaging and microscopy?

Yeah. Well, for us, there are 3 things that we think really matter. The first one is no hallucinations. So most generative AI tends to invent plausible-looking pixels, but in biology, plausible is not the same as real. And the class of AI that we use that's called unpaired image translation has historically been very prone to hallucinations. Missing real structures, inventing fake ones. And we have designed our system through many architectural and training innovations to enforce biological consistency, and we quantitatively validate our performance. The second is that we are integrating, not replacing the workflow. And by keeping their tools, their habits, their quality controls, the scientists feel much more— they trust the system much more than if it was a black box that just gives them an answer. And the third one is performance tracking. So our system monitors its own performance. And if the incoming data starts drifting from the data the model was trained on, it would alert the user that they need to trigger a retraining session. Yeah.

What do you think has to be true for a scientist to trust the AI that it's actually doing what you say it's doing? How do you convince them?

It's validation data. If they see that you're getting the same results as if they use the technique they trust, even though it took them hours of tuning to be able to pull out the information, then if they see the numbers and statistics, they—

It shows them the money, more or less. Like, show them the truth.

Exactly. And then we have some stunning, impressive images that they can see by eye. are preserving what they see as a signal, right? Yeah.

Well, and as we talked about before, Sarpeda is one of Connect's Cool Companies this year for 2026. And Cool Companies is really about highlighting startups with strong potential that are at that late seed Series A and really helping them build visibility, relationships, and additional momentum. So as you look to the next stage for your company, what kinds of partners, customers, investors, or collaborators are most useful for you right now? What are you looking for in the moment?

Yeah. Well, like any business, customers and partners are most important. And we bring value to anyone doing fluorescence microscopy. We can think about tissue-based diagnostics where we are working with precious patient samples that need true co-expression data from a single acquisition, in spatial biology where autofluorescence contaminates RNA reporter readouts, cycling drives cost and time, in drug discovery imaging assays where marker-rich phenotypic screens need reliable quantification, and in organoids and microphysiological systems where 3D matrices have high background and every sample is expensive to generate. And San Diego has world-class examples of all 4 of these applications. We also want to talk to reagent companies, microscope vendors, CROs. We are a software layer that scales best through partners who already have sales channels, right? And to anyone listening who is wrestling with autofluorescence and crosstalk or channel limits, please reach out. We're actively signing up teams who want to transform the value of their microscopy workflows.

And you're working on a raise right now, I imagine, right?

Yes, but our—

Always, always hustling, right?

But our priority is growing our customers, and we are currently self-funded and I'm very optimistic that we will not need a lot of money.

That you'll be able to grow your customer base and do that.

Yes.

So you're open to conversations, but not looking. I like that. All right, so we have our one last classic Connect Current question. It's a very San Diego question. What drives your innovative flow, or what stirs your creativity?

I think it's working at the intersection of AI and biology, where you are learning about the struggles, understanding them deeply, and knowing what technology can do is where cool ideas come from. And there is such excitement when you figure out a solution to the problem that once you figured it out, it seems so obvious, and you can't believe you or somebody else hasn't thought of it sooner.

Right. Yeah.

Our company name, Serpida, comes from a special butterfly, Graphium serpidin, who has 15 different types of color receptors in their eyes. We only have 3.

So it sees better.

So it symbolizes vision beyond human limits. And that's kind of what good microscopy should feel like, right? And yeah, and the other thing that is inspiring and motivating always is good people. So PETA has brilliant physicists and AI scientists who I just love working with. And then the ecosystem of San Diego, cool companies. I'm also a member of Women in Bio and Biotech CEO Sisterhood, like amazing organizations that keep inspiring me as I hear other people's successes.

Yeah, San Diego is just, it's a special ecosystem. And I'll say it's so great. And it's so great to get to have conversations like this here in San Diego. So, I want to thank you for coming on and having such a thoughtful conversation with us. It's a great reminder that innovation, it's often about seeing something more clearly, literally sometimes, as well as figuratively. And Sarpeeta's work shows how AI can help researchers get more from tools they already have, reducing that friction and discovery, and unlocking biological information that might otherwise stay hidden or lost. And as they talked about with the Milken Institute for years, time equals lives, and you're enabling faster science, which could save lives sooner, which is so important. And that's one of the many reasons why we're really excited to have you as one of our 2026 Cool Companies, and really appreciate you joining us.

Thank you very much. Thank you. We are very grateful.

Thanks for listening to The Connect Current. If you want to learn more about Connect and explore resources across San Diego's innovation ecosystem, follow us on LinkedIn or visit connect.org. We'll see you next time on The Connect Current, where San Diego's innovation story keeps moving forward.

This podcast was edited and produced by Resonate Recordings.