Uthana evaluates its human-motion foundation model at scale, and feeds what it learns back in.
Uthana builds a foundation model for human motion, whose output has to be high quality and feel real. It uses Rapidata both to evaluate the model at scale and to inform the inputs the model learns from.
The challenge
Whether generated motion feels real is a judgment people make. To stay best in class as it grows, Uthana needs that judgment on its model's output continuously and at scale, without it becoming the step that slows the team down.
How Uthana uses Rapidata
Uthana uses Rapidata to evaluate its model's output with people at scale. The same human feedback also shapes the inputs that go into the model, so evaluation and training improve together.
The result
According to Uthana's CEO, Rapidata helps the company remain best in class, and removes the risk of human evaluation stalling its growth.
“The output of our foundation model for human motion needs to be of high quality and feel real. Rapidata not only helps us evaluate the model at scale, but also informs inputs into the model that help us remain best in class. With Rapidata we do not run into the risk of stalling our growth.”
Source: Press release, February 19, 2026 · uthana.com