Gig workers gain new gear to train the next generation of robots

The system, which weighs less than three pounds, consists of five cameras synced to a compute backpack. By tracking hand movements and depth perception, the gear allows AI labs to observe tasks ranging from stocking shelves to assembling hardware. Instawork, which connects hourly staff with employers, aims to leverage its 10 million-strong workforce to build a massive, diverse commercial data set. CEO Sumir Meghani pivoted toward this hardware after observing robotics companies increasingly using his platform to fill shifts, realizing that the intersection of physical labor and AI represents a significant untapped market.

Designing the hardware required overcoming ergonomic hurdles, as early prototypes proved too heavy or prone to overheating. After iterations based on feedback from workers, the current model balances data fidelity with comfort, though users note a learning curve regarding the attached cabling. While production currently takes place in Mountain View, the company intends to scale manufacturing overseas. This push comes as tech giants like Nvidia, Meta, and Tesla race to develop machines capable of operating in unstructured environments. By making data collection an opt-in process, Instawork hopes to position its laborers not just as manual workers, but as essential trainers for the future of physical AI.