When venture capitalist Tim Draper backs a startup, it usually hints that something potentially transformative is brewing. His latest move: a $4.5 million investment in Potato, a company developing autonomous AI scientists to accelerate research through agent-driven workflows. Potato’s ambitious goal is shifting scientific discovery from painstakingly slow lab work toward what Potato CEO Nick Edwards, Ph.D., envisions as “runaway knowledge production.”
Founded in 2023 by Edwards, a neuroscientist with experience at NIH, BCG, and Illumina, and veteran CTO Ryan Kosai, Potato is explicitly focused on making scientific processes not only automated but reproducible—a notoriously elusive challenge in research. “Our goal is to accelerate scientific research,” Edwards stated plainly. He elaborated on the core problem: “The big challenge in life sciences is that research is incredibly cost-prohibitive, expensive, and slow, largely because it’s very manual. It takes scientists years to understand the literature context, identify problems, and learn experiments.”
Aiming to make science more automated and reproducible
Nick Edwards, Ph.D. with Ryan Kosai
To address this, Potato positions itself as a “reproducibility company.” Edwards explained: “Reproducibility has got to be one of the guiding features, perhaps the North Star for Potato. You want to be able to trust things. It’s a big challenge in science… Reproducibility is a key, if not the key, component. There’s an opportunity to build tools that help with that, pulling hard-to-find information across literature and papers. That’s a big part of what we do.”
In addition to reproducibility, Potato is taking aim at addressing the overwhelming volume of published literature and persistently high research costs. The manual process of synthesizing existing research is a major bottleneck. “[Manual literature review] It’s painstaking,” Edwards noted from experience. “It’s a challenging process, easily taking days to find the right information and compare papers.” To tackle these problems, Potato’s agents automate scientific literature reviews, generate precise experimental protocols grounded in peer-reviewed knowledge, and employ computational tools alongside lab automation technologies to execute experiments more efficiently. “To me, the vision here is closed-loop, autonomous science with humans in the loop as AI collaborators,” Edwards summarized the operational goal.
This emphasis on trustworthy workflows is already in action. Last October, at the Frankfurt Book Fair, the scientific publishing giant Wiley announced Potato as the inaugural collaborator in its AI Partnerships: A Co-Innovation Program, granting Potato crucial access to Wiley’s extensive library of verified scientific content. Potato is also part of this year’s Merck Digital Sciences Studio cohort.
Initially focused on biotechnology, Potato plans to rapidly expand into additional fields such as materials science and chemistry. In addition to the support from Draper Associates with significant participation from Dolby Family Ventures, the funding round includes the participation of Boost VC, Ensemble VC, Silicon Badia, Alumni Ventures, Defined, and The FounderVC, as well as strategic angel investors Michael Liou and Geoff Entress, Potato is poised to scale quickly.
Scientific feedback loops
“We’re expanding quickly,” Edwards confirmed, discussing the company’s trajectory boosted by the new funding. “The additional funding allows us to scale up and build new capabilities.” He noted the investors’ alignment with the company’s grand vision: “They’re [Draper Associates] excited about this idea of runaway knowledge production,” Edwards shared.
Thinking about closed-loop science—hypothesis, plan experiments, generate data, analyze data, feeding into the next hypothesis iteration—can we shorten that loop?
Under Kosai’s technical leadership, Potato will further refine its AI agents toward fully autonomous computational and wet-lab experiments. Kosai brings 20 years of experience as a principal engineer and software leader, including roles as former CTO of Pioneer Square Labs, founder and CTO of Attunely, and principal engineer at ExtraHop Networks.
A virtual lab meeting
The vision remains collaborative, integrating human guidance: “The human-in-the-loop is essential to that closed-loop process,” Edwards emphasized. “Where it gets more interesting is taking an agentic perspective… You might have one agent with expertise in neuroimmunology, another in single-cell proteomics. It’s like a virtual lab meeting.”
The potential benefits are significant, particularly in neglected research areas: “Think about rare diseases not being investigated because science is expensive, slow, and information is hard to get,” Edwards noted. “[Agents can] pull out some of the mundane aspects or free up scientists for more creative pursuits. There’s an amazing opportunity for that.”
Tim Draper underscored the transformative potential of Potato’s approach: “It is becoming clear that AI agents will create new productivity by wielding the same tools currently used by humans,” he said in a press release.
The opportunity to accelerate science and enable runaway knowledge production is bigger than the Internet.
Potato’s tools are already in use at leading academic and research institutions, including Stanford University, California Institute of Technology, Massachusetts Institute of Technology, University of California Berkeley, The Scripps Research Institute, and Harvard University, among others.
Edwards concludes with an eye toward the future: “A big theme for Potato and scientific discovery is that AGI (Artificial General Intelligence) is coming, with significant implications for science.”
Tell Us What You Think!
You must be logged in to post a comment.