Primordiant is building robotic experimentalists that let scientific models test hypotheses on real lab equipment. The systems are designed to observe a setup, form a hypothesis, run the experiment, and use the outcome to decide the next step. The launch is for R&D labs in materials, biotech, and related physical sciences, where models can already draft many ideas and progress still depends on testing them as protocols and instruments change. Fixed lab automation can repeat a known workflow at high volume, then stalls when the question shifts, a new instrument arrives, or the procedure has not been written. Primordiant is building a system that retrofits into existing workflows and can take on new tasks as the science moves. Previously called SuperRadiant, the company is in Y Combinator's Fall 2026 batch and based in San Francisco. The founders are Owen Tower (CEO), Cooper Niu (CTO), and Benedict Quartey. Tower and Niu hold physics PhDs from Brown, with research experience at NASA JPL, CERN, and Fermilab. Quartey holds a computer science PhD from Brown and previously built robot learning systems at the Boston Dynamics Robotics and AI Institute. This launch treats adaptable physical testing as the open gap after models learned to read papers, write code, and propose experiments.