AI is spitting out more potential drugs than ever. This start-up wants to figure out which ones matter.

AI has made significant strides in the scientific community, particularly with Google DeepMind’s groundbreaking work in predicting protein structures, fundamental to cellular processes. However, a new challenge has emerged in the biotechnology sector: efficiently characterizing the multitude of potential drug candidates generated by these AI models. Enter 10x Science, a startup founded in December 2025, which has just secured $4.8 million in seed funding led by Initialized Capital, with additional support from Y Combinator, Civilization Ventures, and Founder Factor. The team behind 10x Science comprises David Roberts and Andrew Reiter, both seasoned biochemists, along with Vishnu Tejas, a veteran in computer science and AI. Roberts notes that while biopharma companies have advanced prediction tools at their disposal, the bottleneck remains in the characterization of drug candidates. "You can add as many candidates as you want to the top of the funnel, but they all need to pass through this characterization process. Everything needs to be measured," he explained. Understanding protein structures is crucial for developing biologic drugs, which are crafted to target specific diseases using complex designs. A prime example is Keytruda, a Merck drug that aids the immune system in identifying cancer cells. The founders of 10x Science collaborated in the laboratory of Nobel laureate Dr. Carolyn Bertozzi at Stanford, where they encountered frustrations in deciphering molecular interactions between cancer cells and the immune system. The startup employs mass spectrometry, a sophisticated technique that determines the atomic structure of molecules, which generates intricate data requiring expert interpretation. To tackle this challenge, 10x Science’s platform integrates deterministic algorithms with AI agents designed to analyze this data. The founders invested considerable effort in training their models with spectrometry data to ensure regulatory compliance and traceability in their analyses. Matthew Crawford, a scientist at Rilas Technologies, has been utilizing the 10x Science platform and reports a significant increase in efficiency. He highlighted the model's impressive capability to autonomously find relevant data and adapt to various molecular evaluations, a feature he attributes to the deep expertise of its creators. "I ran a particular protein through it, and it just kind of figured out what the protein probably was based on the file name," he said. Beyond aiding startups, 10x Science is collaborating with major pharmaceutical companies and academic researchers. The seed funding will be used to hire more engineers and refine their model further. Looking to the future, Roberts hopes the company will enhance its understanding of biology by integrating protein structure data with other cellular information. With 10x Science, investors are presented with a promising entry into the biotech landscape that isn’t solely reliant on the success of a particular drug. Should the founders’ vision materialize, the platform could become an invaluable asset in drug development, streamlining processes irrespective of market success. "This is a SaaS platform that pharma has to pay for, every single month, to navigate through all of these potential candidates," noted Zoe Perret, a partner at Initialized. She emphasized the founders' unique expertise as a competitive edge in this specialized field. Crawford anticipates that this innovative software will empower researchers by simplifying complex methodologies and enhancing their research capabilities.