Saddlepoint Labs is a pre-seed company whose mission is to help experimentalists easily and consistently produce accurate data.

Why Saddlepoint Labs?

Saddlepoints are solutions to optimization problems that are locally optimal for a restricted set of directions but that can be improved by looking in new directions.

Our logo is a contour plot of a saddlepoint: along one diagonal, contours fade and value decreases as we move away from the current solution at the center; along the other diagonal, contours get darker and value grows as we try new solutions.

In the chemical and life sciences, lots of work has gone into AI methods that predict properties and propose new molecules. Unfortunately, these in silico approaches are only as accurate at the data given to them. Much of this data was collected using hard-to-reproduce manual processes or poorly supervised robotic systems.

At Saddlepoint Labs, we feel that the biggest opportunities for AI in the chemical and life sciences are not in predicting properties or proposing new molecules. They are in changing how the data itself is collected to reduce experimental variability.

Our mission is to address the reproducibility challenge in both manual labwork and laborotary automation, helping experimentalists to generate accurate consistent data that they can trust.

Meet the Team

Saddlepoint Labs was founded by Cornell Professors Peter Frazier and Nick Abbott, who bring experience from multiple prior startups, 9 yrs of data science/ML at Uber, and 10+ yrs of research in AI/ML for the chemical sciences.

They are joined by Founding Engineer Henry Robbins (ex-Uber, Cornell double-major in computer science and operations research).