Biopharma AI startup 10x Science is aiming to solve one of the biggest problems created by modern AI in healthcare: too many potential drug candidates and not enough efficient ways to evaluate them.
As artificial intelligence generates more possible treatments, researchers face a growing challenge in determining which molecules are truly worth advancing for testing and production. That is where 10x Science wants to make an impact.
Founded in December 2025, the company has announced a $4.8 million seed round led by Initialized Capital, with participation from Y Combinator, Civilization Ventures and Founder Factor.
Biopharma AI Startup Targets the Characterization Gap
David Roberts, one of the company’s co-founders, said drug developers already have strong prediction tools that can generate many candidates. The real challenge comes later, when every candidate must go through characterization and measurement.
That process is especially important for biologic medicines, which are developed in living cells and designed to target specific diseases or conditions. Understanding protein structure is a critical part of that work.
The founders of 10x Science — David Roberts, Andrew Reiter and Vishnu Tejas — previously worked together in the Stanford lab of Nobel Prize winner Dr. Carolyn Bertozzi, where they studied cancer cell and immune system interactions.
How Drug Characterization AI Works
The company focuses on mass spectrometry, a method used to determine molecular structure by measuring particles in an electric field. While highly accurate, the technique creates complex data that often requires specialized expertise and significant time to analyze.
10x Science says its platform combines chemistry- and biology-based deterministic algorithms with AI agents that interpret spectrometry data. The company also said it worked extensively to train its models and make outputs traceable for regulatory use cases.
That could make drug characterization AI more accessible for companies that need faster answers.
Early Users Say It Speeds Up Research
Matthew Crawford, a scientist at Rilas Technologies, said he has been using the platform for several weeks and found it accelerated his workflow.
According to Crawford, the system was able to explain conclusions, locate relevant data for analysis and adapt to different molecule types. He also said it made practical assumptions compared with other AI tools he had tested.
He described one example in which the software identified a likely protein based on a file name and then searched online databases for the matching sequence automatically.
Growth Plans and Bigger Ambitions
The startup said it is also working with multiple large pharmaceutical companies and academic researchers. It plans to use the new funding to hire engineers, improve the platform and expand its customer base.
Roberts said the longer-term vision goes beyond evaluating proteins. He hopes the platform can eventually combine protein structure data with broader cellular information to create what he called a new way to define molecular intelligence.
For investors, the appeal is that the company’s success does not depend on one drug candidate winning approval. Instead, it could become an important software layer across the wider AI drug discovery ecosystem.
Don’t miss out on our latest news—follow us for the latest AI news, breakthroughs