A Seattle-based startup is revolutionizing the landscape of scientific research by harnessing the power of artificial intelligence to automate experiments. This innovative approach promises to significantly accelerate the pace of discovery, reduce costs, and enhance the reproducibility of results, addressing some of the most pressing challenges in the scientific community.
The company, named Potato, was founded in 2023 by neuroscientist Nick Edwards and technologist Ryan Kosai. Potato secured $4.5 million in seed funding to further develop its AI platform. The core of Potato's innovation lies in its "closed-loop, autonomous science" model. This involves AI agents capable of handling every stage of the scientific process, from formulating hypotheses to designing and executing experiments, analyzing data, and even drafting manuscripts for publication. The company's name is a playful reference to the potato battery experiment.
Potato's current platform functions as an AI research assistant. It can generate hypotheses, optimize experimental protocols, summarize scientific literature, critique journal articles, and draft manuscripts. The system uses large language models (LLMs) fine-tuned with retrieval-augmented generation (RAG) to ensure outputs are grounded in verified scientific literature. The platform offers additional capabilities, including automated review of uploaded papers – highlighting methods, evaluating controls, and suggesting follow-up experiments – as well as visualizing workflows by generating diagrams from text-based protocol descriptions. Researchers can also query the system on specific biology topics and receive chain-of-thought answers grounded in published work.
Beyond its current capabilities, Potato is actively developing robotic systems capable of performing actual lab benchwork, aiming for a fully autonomous research cycle where AI agents plan, run, and interpret experiments with scientists remaining in the loop. By automating routine tasks, scientists can focus on more meaningful and complex tasks. This allows scientists to become more creative with their research and work more efficiently.
Reproducibility is a central tenet of Potato's mission. Edwards has described Potato as a "reproducibility company," aiming to tackle the persistent issues of inconsistency in scientific research by automating literature review and standardizing protocol design. The company believes its tools can reduce both the time and cost associated with generating reliable scientific results, particularly in underfunded or neglected research areas. Potato's platform has already been deployed in labs across major institutions, including Stanford, Harvard, MIT, UC San Diego, UC Berkeley, Scripps Research, and the University of Washington.
Other companies are also using AI to streamline scientific research. Bridgetown Research, another Seattle-based startup, raised $19 million in a Series A round to build AI agents that crawl the internet and analyze datasets for various research use cases. FutureHouse, a non-profit AI research lab in San Francisco, is building AI agents to automate research in biology and other complex sciences.
The integration of AI and automation is transforming laboratory research by enabling entire workflows to be automated end-to-end. AI systems are becoming more sophisticated, helping scientists acquire high-quality data in the shortest possible time and avoid mistakes that would result in time wasted redoing the experiment.
While AI offers tremendous potential, it's essential to maintain rigorous standards for data quality and to ensure responsible AI implementation. This involves transparency in how AI systems make decisions, allowing researchers to understand and trust the outcomes. By maintaining a balance between AI, automation, and human oversight, the scientific community can leverage the strengths of AI while mitigating its limitations, ultimately enhancing the quality of research and upholding ethical standards.
