Everyday Chores Filmed for Mere Pennies an Hour Are Helping to Refine AI Technology

To feed a wave of robot-training datasets, Indian workers like Nagireddy Sriramyachandra film everyday actions with head-mounted smartphones for about 250 rupees (roughly $2.40) an hour. Companies including Objectways and Bangalore-based Humyn Lab turn this egocentric footage into training labels as investors size a humanoid-robot market at about $38 billion by 2035.

• Key Takeaways:
• Objectways pays about $2.40/hour as India supplies AI robot-training data.
• Goldman Sachs sees humanoid robots hitting $38B by 2035, boosting AI demand.
• Ravi Shankar’s Objectways faces privacy and pay-equity debates as AI scales.

In Tamil Nadu, a headband camera catches Nagireddy Sriramyachandra slicing mangos, and the footage becomes training data for robots at about $2.40 an hour. Those handheld chores feed an egocentric video pipeline that runs from textile floors to annotation shops like Objectways and on to US tech corridors where CEO Ravi Shankar courts the next wave of AI demand. The money is modest, the market forecast is not, with Goldman Sachs pegging humanoid robots at $38 billion by 2035. Between those numbers sit thorny questions about who benefits from India’s push to be a data hub, and what constant recording means for workers’ privacy.

Filming daily life to train tomorrow’s AI

AI needs human context, and right now that context often looks like a smartphone strapped to a worker’s head, capturing the ordinary. In southern India, people film themselves slicing mangos, tying shoes, or making coffee. The footage helps US companies refine robots and assistants that learn by watching. The pay is low, the stakes for global AI development are high.

Earning pennies for human-like data

For a typical home task, workers earn about $2.40 per hour, recording first-person footage that shows exactly how hands move through a chore. This is called egocentric video, and it is prized because it maps intent, motion, and environment in one stream. The clips teach imitation, which is crucial if robots are expected to fold laundry or prep food without step-by-step coding.

The business of AI annotation

One hub in this supply chain is Objectways, a data-labeling company founded by Ravi Shankar that works closely with US tech clients. Employees film hundreds of micro-tasks inside staged homes and factory mockups, then colleagues annotate frames into machine-readable steps. Per the company, the output feeds teams building household robots and warehouse systems at major platforms, including firms like Amazon.

The economics are hard to ignore. Training humanoid and mobile robots demands mountains of clean, real-world behavior data. That is fueling a services market Wall Street is tracking closely. According to Goldman Sachs, spending tied to humanoid robots could top $38 billion by 2035, assuming hardware costs fall and general-purpose models continue to improve.

Challenges and ethical questions

This work raises thorny issues that follow the data upstream to Silicon Valley. Privacy is the first concern, since footage often comes from kitchens, living rooms, and factory floors. Some workers avoid filming bedrooms or family members altogether. Others want clear rules on retention, licensing, and whether their content will be fed into future commercial models without ongoing compensation.

> > 🚨 India’s Workers are Training AI Robots to Take Their Jobs. > > > For ₹250 an hour, workers record everyday activities to help global tech companies train AI-powered robots to move like humans. pic.twitter.com/Y5DYgGcMTh > > > — Bharat Tech & Infra (@BharatTechIND) June 11, 2026 > >

There is also the matter of pay equity. The distance between a robot trained on low-cost labor and a premium product sold in the US invites scrutiny from policymakers and customers. Should datasets that enable high-margin robotics command higher wages for contributors? The question shadows the AI boom much like debates around ride-hailing and content moderation did a decade ago.

Still, scale keeps pulling the market forward. US teams need diverse hands, lighting, and environments to avoid brittle models. For now, the cameras keep rolling, one mango slice and towel fold at a time, turning everyday gestures into the next training set for machines that want to learn how we live and work.