Blotdle
Backed by Y Combinator

Environments for
scientific discovery.

Blotdle is an applied research company building environments and evals for scientific discovery and recursive self-improvement.

We are living through the fastest acceleration of discovery in human history. In just eight months, AI went from math contests to resolving hundred-year-old conjectures. Today, models can read every paper ever written and propose millions of hypotheses. Yet they still cannot test those ideas or build on what they learn.

While digital domains like math and code scale rapidly because they possess cheap, instant, and deterministic verifiers, empirical science is tethered to the physical world. Moving from generating millions of hypotheses to actual scientific breakthroughs requires closing the loop between digital reasoning and physical reality.

Scientific discovery runs as a long, iterative loop of hypothesizing, experimenting, evaluating, and adapting. Current models can execute any isolated step, but they lack a veteran researcher's judgment across the full cycle: knowing which experiment to run next, when to doubt an anomaly, and when to change course. That intuition comes only from years of real experiments and compounded failures.

Blotdle is building the physical verifier to unlock this capability. Starting in biology, we run closed-loop wet labs operated directly by AI models, pairing automated workcells with real-world execution.

Every experiment generates proprietary failure data, telemetry, and training environments. We use this loop to teach models how to conduct autonomous research-accelerating discoveries that advance humanity.

We invite you to partner with us.