Research

We study what happens when a real company hands its judgment and its work to machines; how faithfully a model can stand in for one person, what agents actually do when the stakes are real, and how much of it can run on hardware you own. We want the leverage this technology creates to stay within reach of small teams and individuals.