Over the past few years, India’s online food delivery market has grown significantly, with both Zomato and Swiggy going public and the number of cloud kitchens increasing. Meanwhile, startups working on home-based services, such as on-demand housekeeping platforms such as Urban Company, Snabbit, and Pronto, are gaining popularity.
Human Archive, a Silicon Valley-based startup, is partnering with these companies to outfit employees with special hats equipped with cameras to collect first-person video data of their daily tasks that can be used to train robots.
Without naming specific partners, the startup said it works with companies in the home services, hostel and restaurant sectors to collect self-centered data, and said it has deployed more than 1,000 active headsets across multiple locations.
On the back of that traction, Human Archive announced Tuesday that it has raised $8.2 million in funding from Wing Venture Capital, NVP Capital, Y Combinator, and angels from OpenAI, Nvidia, Google, Mercor, AfterQuery, BAIR, SAIL, Brad Boa, and Meta.
The startup was founded by two students from Berkeley and Stanford: Samay Mani, Lucille Agarwal, Shlok Patel, and Raj Patel. The latter two are cousins. (Raj Patel is CEO.) All four have research backgrounds spanning robotics, hardware, and haptic data.
The company’s creation is a direct bet on where the AI industry is headed. Robotics labs and cutting-edge AI companies face significant bottlenecks as they race to build machines that can perform physical tasks in the real world. There is a dearth of high-quality, real-world training data that shows humans performing daily tasks. The Human Archive’s bet is that the workers staffing India’s burgeoning gig economy are an untapped and scalable source of just such data.
Human Archive is working with multiple partners, but has been turned down by many Indian home service companies, including Pronto and Urban Company, the company said.
The company’s rejection by major companies became public information over the weekend when Indian media outlet Entrackr reported that Pronto was actively seeking a partnership to collect worker data for robot training, and that Snavit was in early talks with Human Archive before the project fell apart.
Urban Company CEO Abhiraj Singh Bahl told X that the company would not entertain such a deal, and Patel countered that Urban Company would soon be forced to reconsider or risk losing relevance due to customer churn. Co-founder Lucille Agarwal was even more blunt, posting that when Pronto founder Anjali Saldana brought up the idea of a data partnership, she laughed at him and called him a “stupid.” Pronto acknowledged the conversation but said he chose not to move forward.
Across the country, other startups are also collecting self-centered data from a variety of work environments, including factory floors. To differentiate itself, Human Archive uses and develops additional devices such as haptic gloves, full-body motion capture suits, and wrist cameras that sync with RGB-D (color images paired with depth information in real time) to capture data including coordinated movement and tactile forces, which it sells to AI labs. The startup believes that video data alone is not enough and that combining it with other sensor data can further increase its value.
Initially, Human Archive used makeshift setups or off-the-shelf rigs to collect data. We are currently working on custom hardware that works together to capture different types of data. More than 50 different devices have already been deployed to collect various data points.
“To collect data, we started with an iPhone and then built our own custom rigs and caps. We now have over seven different hardware products that we use interchangeably across different modalities. After collecting data from different devices, we worked on synchronizing the data from all these different sources,” Patel said by phone.
The company said it is developing ways to fine-tune its AI models with its own data and test them on robots to assess their effectiveness on tasks. By doing this, startups can demonstrate the quality of their data to potential customers and their internal models after training.
Wing VC partner Zach DeWitt said the startup has the unique advantage of collecting data from multiple sensors.
“No other company in the world is able to synchronize and collect data from headset RGB-D, force feedback, whole-body motion capture, and chest and wrist cameras at scale. They are doing internal model training on this data, and the novelty of the sensor and the scale of the new dataset that will soon be released has all major labs and universities interested in running the experiment,” he told TechCrunch.
Data collection and expansion plans in India
Despite rejection from prominent companies in the home services industry, Human Archive partnered with a small startup to offer discounted services to customers. Once the employee arrives at the home, consumers are offered the option through the app to either pay a discounted price in exchange for consenting to data collection or pay full price for an unrecorded visit.
Patel said customers are happy to choose the former because quality-of-service disputes are common and video recordings can help resolve them.
The company pays workers a base rate of $1 an hour for their participation in self-directed data collection. Other companies pay between 250 and 400 rupees (about $2.63 to $4.20) an hour, the Economic Times reported. Patel said competitors pay higher salaries than Human Archive, but the company’s presence in India allows it to keep its fees lower.
“The Human Archives network provides immediate and flexible revenue opportunities around the world, lowering barriers to participation in the AI economy. We see this as a critical bridge to funding immediate livelihoods while building the infrastructure for a safer, more productive future,” DeWitt said.
Beyond wage payments, there are privacy concerns with data collection through video recording. It’s unclear what information Human Archive provides its employees about how the footage will be used. The company said its commercial agreements are compliant with India’s Digital Personal Data Protection Act (DPDP) and display a privacy policy notice along with consent information detailing the purpose of data collection and how it will be processed. The company said all data is anonymized and faces are blurred in the recordings. Last week, Moneycontrol reported that India’s Ministry of Electronics and Information Technology is investigating the consent mechanisms and data collection practices of startups that collect self-centered data through home-based service workers.
Human Archive primarily collects data in India, but has also begun expanding to Southeast Asia and the United States. The company is also working on building a platform where anyone can participate in data collection and earn money. It also hopes to offer services like cleaning and cooking to U.S. customers in exchange for data collection by participating workers, but these programs are only in the early testing stages.
Several well-funded startups are competing to develop physical AI. To achieve this, we need large amounts of training data that shows humans at work. The Human Archive is one of the companies racing to meet that demand. The scalability of that approach will depend on the partnerships we build and the uniqueness and amount of data we can collect to meet the demands of physical AI labs.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
