🗓️
2026
Rebis
Introducing Rebis, the platform that allows you to earn with every move.
Rebis aims to provide the multi-modal data layer for Physical AI. The first envisioned product is a consumer-grade visuo-tactile glove that captures how people touch, grip, and handle objects, turning everyday work into robot-training data at scale.
The problem we are solving:
Robot foundation models are starved for manipulation data. There is no internet-scale corpus of physical interaction the way there is for text and images.
Every existing source falls short. Teleoperation is slow and expensive, simulation struggles with contact physics, and egocentric video records what hands look like, not what they feel.
The few wearables that capture true touch (grip, pressure, contact) are expensive, specialist hardware tied to a single robot hand. None are built to scale across thousands of collectors.
Our hypothesized target audience:
Physical AI labs and robotics OEMs training dexterous manipulation policies.
Data-collection vendors looking to add touch to their vision- and motion-capture offerings.
On the supply side, gig workers and skilled tradespeople who can earn by recording work they already do.
We are in the process of validating the thesis through field research funded by Northwestern’s Levy Inspiration Grant, including interviews with Physical AI experts across big tech and leading robotics startups.
