The Silicon Valley startup Human Archive has raised $8.2 million to gather real-world data for training robots by leveraging India’s rapidly expanding local service industry. The company says it has deployed over 1,000 head-mounted devices at multiple locations to capture first-person perspectives of tasks in settings such as housekeeping, food service, and dormitories.
Funding comes from multiple AI-focused institutions.
This funding round was led by Wing Venture Capital and NVP Capital, with participation from individual investors affiliated with Y Combinator, OpenAI, Nvidia, Google, Meta, and other institutions.
Human Archive was founded by four students from Berkeley and Stanford, with the team’s research spanning robotics, hardware, and haptic data. The company recognizes that embodied intelligence and robotics development are accelerating, but high-quality, real-world training data remains scarce—this is their opportunity.
Synchronize video motion with haptics
Unlike solutions that rely solely on video capture, Human Archive is also developing haptic gloves, full-body motion capture suits, and wrist-mounted cameras to simultaneously record movement, force, and RGB-D data. The company states that over 50 different devices are currently in use, and more than seven types of interchangeable hardware products have been developed.
The company stated that video alone is insufficient to support more complex robot training; combining it with tactile and depth information would significantly increase data value. In addition to selling data to AI labs, the team is also experimenting with fine-tuning models using their own data and testing task completion performance on robots to demonstrate data quality.
The Indian model faces privacy scrutiny
Human Archive has not disclosed the names of its partners but states that it has collaborated with companies in the home cleaning, dormitory, and food service industries. Its approach involves having service staff wear camera-equipped devices during visits, allowing consumers to choose between two options in the app: consent to data collection in exchange for a discount, or pay the full price without being recorded.
Raj Patel said some consumers prefer the former option because service disputes are not uncommon, and video recordings help reconstruct the process. The company pays workers involved in data collection a base rate of $1 per hour, though the report notes that other similar companies offer higher hourly rates.
However, this model has raised concerns about privacy and informed consent. The company states that its business contracts comply with India’s Digital Personal Data Protection Act, displays privacy policies and data usage disclosures, and anonymizes data, including blurring faces. India’s Ministry of Electronics and Information Technology is reportedly reviewing the consent mechanisms and data collection practices of relevant startups.
Expanded into Southeast Asia and the United States
Despite some leading Indian home service platforms refusing to cooperate, Human Archive has partnered with smaller platforms and begun expanding into Southeast Asia and the United States. The company is also building an open platform aimed at enabling more people to participate in data collection and earn compensation.
According to its vision, U.S. users may in the future exchange services such as cleaning and cooking for price discounts, while agreeing to have their data collected by service providers. However, these programs are currently still in early testing phases.