Instawork Robotics Lab Debuts Instacore — a Wearable System Built to Scale Real-World Robot Training Data

Lightweight five-camera system balances the data requirements of leading research labs with the comfort and usability requirements of real-world operators.

SAN FRANCISCO — Instawork Robotics Lab (IRL) today announced Instacore, the company’s wearable hardware system that captures real-world training data for general-purpose robots and physical AI foundation models.

Developed with direct input from Instawork’s workforce of more than 10 million skilled Professionals (Pros) and robotics partners, Instacore is purpose-built for deployment in real commercial environments — kitchens, warehouses, hotels, and light manufacturing facilities. It is the first-of-a-kind data capture platform designed specifically for the humans wearing it.

Instawork is currently collecting hundreds of thousands of hours of real-world task data every month through its Instawork Lens platform; Instacore is built to significantly accelerate data volume and quality.

Designed for Humans, Not for Robots

Because data diversity is only possible at scale if workers can actually do their jobs while wearing the system, Instacore was built in the Bay Area and tested through dozens of iterations. Instawork Pros wore early versions at home and while working to ensure the system can be worn during entire shifts.

“The value we provide to general-purpose robotics companies is in the diversity of data. Diverse tasks. Diverse physical environments. Diverse lighting conditions. Diverse operators,” said Aaron Bromberg, Head of Robotics and Physical AI at Instawork. “Instacore was built to capture that diversity at scale in the exact commercial environments where robots will actually be deployed.”

How Instacore Works

• Five cameras with hardware time synchronization — head, chest, and two wrists — are connected to a power-efficient compute backpack built for a full work shift.
• Stereo head camera provides 3D depth perception at arm’s length.
• Wide-angle chest camera captures the full working environment, giving models the spatial context robots need when operating outside a lab.
• Wrist cameras with tracking markers allow computer vision models to calculate the exact 3D position and angle of each hand, utilizing multiple IMUs for SLAM and pose estimation — translating directly into robot joint angle training data.
• Modular by design: the system can be customized with UMI grippers or alternative camera configurations based on partner requirements.

The Data the Robotics Industry Is Missing

The training data required to teach robots real-world physical dexterity does not yet exist at scale. The industry collected roughly 100,000 hours of robot training data in 2024, growing to approximately one million hours in 2025. Projections reach 20 million hours in 2026, which is still less than 0.04% of what researchers believe is ultimately needed.

Industry Scale

Goldman Sachs forecasts the humanoid robot market could reach $38 billion by 2035. Grand View Research projects the global data collection and labeling market at $17 billion by 2030. Deloitte estimates nearly 2 million skilled physical AI jobs will go unfilled by 2030.

Learn more about Instacore and Instawork Robotics Lab at instawork.com/robotics.

---

About Instawork

Instawork’s mission is to create economic opportunities for businesses and hourly workers across the globe. As the leading AI-powered marketplace for hourly labor, Instawork connects light industrial, hospitality, retail, and robotics companies to skilled workers, turning staffing agility into a competitive advantage. Instawork helps more than ten million workers earn on their terms while developing valuable skills.

Backed by leading investors including Benchmark, Craft, Greylock, and Spark Capital, Instawork is redefining how businesses stay resilient and how people work.

Media Contact

Amanda Pires

Head of Communications, Instawork Robotics Lab

apires@instawork.com | 650-208-3728