Ensemble is excited to announce our investment in Potato, a platform building autonomous science agents that help researchers plan, run, and iterate experiments using AI. Potato was founded in 2023 by Nick Edwards, a neuroscientist with research experience at Brown University, NIH, and University of California San Diego, and Ryan Kosai, a technology executive with a background in data science and engineering.
Potato collaborates with leading academic institutions and biotech companies to lower the cost of discovery and expand access to underexplored scientific questions, from rare diseases to materials chemistry. Beyond specific use cases, however, Potato plans to amplify scientific headcount and push a new paradigm of discovery by bringing agentic scientists into the engine room of innovation.
- Potato’s agents automate core research workflows: conducting literature reviews, generating experiment protocols, and integrating with lab tools to run analyses — all while learning from experimental feedback.
- The company is already trusted by labs at Stanford, Caltech, and MIT, and has partnered with Wiley and Merck to bring agentic science into the heart of biotech and pharma R&D.
Ensemble is proud to join Tim Draper (Draper Associates) and Adam Draper (Boost VC) in supporting the company’s Seed round, alongside Dolby Family Ventures, Silicon Badia, Alumni Ventures, The FounderVC, and strategic angels Michael Liou and Geoff Entress.
We sat down with CEO and co-founder Nick Edwards to talk about the limits of human cognition, what it means to hire an AI agent, and how the future of science might unfold.
If a cure for cancer is discovered in a room full of AI agents — but no human knows the secret yet — have we cured cancer?
That’s just one of several inspiring and (until recently) outlandish questions that came to mind as Ensemble spoke with neuroscientist Nick Edwards, founder of Potato. With a notable scientific research career to back up his vision, Nick isn’t just trying to speed up scientific research, but reimagining who does science and who’s in the know when machines begin to think like scientists themselves.
Across biotech labs and academic institutions, the limits of human cognition are becoming a bottleneck to progress. Urgent questions, promising molecular targets, and neglected diseases are piling up, and there simply aren’t enough researchers with the bandwidth to pursue them. Paradoxically, as we make new discoveries, the volume of knowledge we consider to “still be out there” increases rapidly. In the age of the LLM, the once-vaunted human mind, with its limited “context window,” becomes the great constraint.
The Crisis in Science: Cost, Speed, and Scalability
Before founding Potato, Nick Edwards spent a decade inside the world of academic science, publishing research at Brown, NIH, and UC San Diego. He saw firsthand how slow, expensive, and exclusive the process of discovery could be, even at top institutions. The best ideas could sit idle for years, buried under grant applications, protocol design, and institutional friction. Rare diseases were often ignored. Cross-disciplinary hunches got left on the whiteboard. Not because the science wasn’t possible, but because the system isn’t built to pursue it.
“There’s just a lot of science that’s not being done in the world right now that could be… if it were faster, less expensive, and less manual.”
If the core bottlenecks were time, bandwidth, and the limits of human cognition, then scaling human labor wouldn’t solve the problem. Something more flexible and more capable of reasoning across vast technical domains was needed. As large language models began demonstrating early signs of tool use and memory, Nick started to see a new kind of scientific teammate emerge — not one you hired, but one you deployed.
AI as Headcount: Redefining the Scientific Workforce
This is where the mind-bending questions come into play: Is science necessarily a human endeavor? Are the human inputs – deciding which questions to ask, which experiments to fund, which cures to pursue – a roadblock rather than the engine?
Humans burn time navigating literature, designing protocols, and switching between brittle software tools. At biotech startups, he saw how constrained discovery could be by hiring cycles and budget approvals. With the imminent arrival of AGI, Nick wondered: What is the role of humans in the scientific endeavor? The resources exist to deploy AI at a scale many orders of magnitude greater than what all the humans in the world could achieve.
“We see a world in which we sell agents as headcount… capable of a lot of that work.”
Those insights, learned over years with “boots on the ground,” shape Nick’s vision around AI in the lab. Potato isn’t selling a dashboard or a pipeline tool. It’s selling a synthetic teammate with access to domain knowledge, protocols, literature, and lab tools, ready to contribute like a human scientist. Except it never sleeps, never forgets, and doesn’t need a grant to keep going. And it has a nearly infinite context window for handling inputs.
The implications of truly scalable scientific labor are massive.
From Assistant to Collaborator to Autonomous Scientist
When Nick first started building Potato, he was trying to give time back to scientists.
As the platform matures, the agents have gotten smarter — integrating with computational tools, lab automation systems, and proprietary datasets. They stopped just organizing information and have started acting on it. Planning experiments. Running them. Analyzing results. Learning from each cycle and feeding it into the next.
“We're just massively increasing the number of intelligences that can collaborate on things.”
What Nick envisions next is something even more ambitious: autonomous scientists — agents that can move through the full scientific loop on their own. They’ll know where to look, what to test, and how to improve. This isn’t a distant sci-fi fantasy. It’s already visible in the way Potato’s agents operate inside the lab.
“We think there’s an opportunity to build autonomous AI scientists… to speed up experimental discovery and flood the world with science.”
Building the Future While Shipping the Present
When you hear Nick Edwards talk about autonomous AI scientists, it sounds like something out of a DARPA playbook or a sci-fi novel. But for the Potato team, the focus is practical: ship tools that solve real problems for researchers today. The team is already working with top-tier academic labs and pharma partners, helping them streamline literature review, generate protocols, and analyze experimental data.
“We’re already solving meaningful problems in real labs,” Nick says. “But we’re doing it in a way that ladders up to something much bigger.”
About Potato
Potato is developing autonomous AI scientists and was founded in 2023. The company’s tools are used by major academic institutions and labs including Stanford University, California Institute of Technology, Massachusetts Institute of Technology, University of California Berkeley, The Scripps Research Institute, and Harvard University, as well as leading biotech and non-profit labs.
WSJ: Saronic Corsair locates and rescues downed U.S. pilots
The Journal reports the crew had spent two hours in the water off the coast of Oman. An unmanned vessel (Saronic Corsair) found them and brought them to shore safely.
Critical Infrastructure is the Foundation of the New AI Economy
AI can only go as far as the underlying physical infrastructure allows.
Q2 Update: Ensemble VC’s AGM features world-leading founders, Ensemble data breakthroughs
Ensemble's 2026 AGM brought together founders, LPs, and investors for a deep dive into the interplay between AI and the critical infrastructure it's built on.
Systems That Learn: What We Look For in Physical AI Companies
Physical AI and the Reindustrialization Window: What Separates the Companies That Will Matter From Those That Won't
Dear VC Partner: Your Firm is Falling Critically Behind. And It's Probably Because of You
In this article, Dr. Gopinath Sundaramurthy argues that while venture capital firms are aggressively funding AI platforms, most have failed to adopt AI meaningfully within their own organizations. Drawing on Ensemble VC’s decade of experience building an AI-native investment platform, the paper explains why superficial AI usage—tools, pilots, and isolated workflows—does not translate into durable competitive advantage.
