signal/raiseTrack rounds
Series AAI · Bengaluru

Humyn Labs raises $47m Series A from Accel and Lighthouse Canton for robot training data

The round, announced six months after Humyn Labs began operating, is made up of $26m of equity led by Accel and $21m of venture debt led by Lighthouse Canton. The company says it has already passed $50m in annualised revenue.

s/rFunding deskIndia · 3 min read

Humyn Labs, a startup that collects real-world data to train and test robots, said on 18 November 2026 that it had raised $47m in a Series A round. The total has two parts: $26m of equity led by Accel and $21m of venture debt led by Lighthouse Canton.

Prosus Ventures, BeeGlobal, Finserv Intelligence (through Bajaj Finance), the Bajaj Alts AI Opportunities Fund, Dallas Venture Capital, Jump Trading and Lighthouse Canton took part in the equity portion, together with an angel investor the company did not name. Entrackr and Robotics 24/7 identified the angel as Anita Pramoda, a former chief financial officer of Epic. Trifecta Capital joined the debt facility. Inc42 put the total at about ₹439 crore. No valuation was disclosed.

Humyn Labs was founded in 2026. It said it had crossed $50m in annualised revenue within six months of starting operations, a figure it did not break down further.

“We bring the model lab and the site operator onto that floor together, so every deployment teaches the next model, and the world's factories, farms and warehouses become the training set.”Ishank Gupta, Co-founder, Humyn Labs

What Humyn Labs does

The company records how people carry out jobs in factories, warehouses, farms and homes, using capture devices it builds itself to each customer's specification. A processing system it calls the enrichment engine turns those recordings into labelled training datasets covering sight, sound, motion and touch, with a record of where each piece of data came from.

A second line of work, which Humyn Labs calls deployment readiness, puts robots into the same kinds of sites together with model developers and site operators. Each deployment is measured against how people do the job, and the gap is fed back into the next version of the model. The company argued that the main obstacle for physical AI is now less the model itself than getting a robot that works in a demonstration to do the same task reliably over full shifts on a real site.

Its customers include frontier AI labs, humanoid robot makers and developers of world models and voice AI. Humyn Labs said it runs data capture and processing in more than 20 countries across Latin America, East Asia and South and Southeast Asia, and its website mentions 36 languages. It also publishes BRIDGE, a benchmark series for physical AI with editions on voice, vision and touch.

Who is behind it

Humyn Labs was started by Manish Agarwal, an IIM Ahmedabad graduate, and Ishank Gupta, a graduate of the Wharton School, both described by the company as serial entrepreneurs. According to Entrackr and Inc42, Agarwal was previously chief executive of Nazara Technologies. Entrackr reported that Gupta earlier co-founded KGeN, a platform for building networks of verified users.

The company's website lists a team of 25 working in research, data science, infrastructure, engineering and field operations, names Bangalore as its headquarters and mentions a US subsidiary. Entrackr also described the company as Bengaluru-based, while Inc42 reported that it is headquartered in San Francisco.

Pratik Agarwal, a partner at Accel, said in the announcement that data and deployment were where a lasting lead in physical AI would be built, and that Accel had led the round to back the founders over the long term as they expand abroad.

What the money is for

Humyn Labs said it will use the funds to set up deployment-readiness infrastructure in Brazil, South Korea and India, to scale its capture hardware and enrichment engine, and to grow its teams in San Francisco, London and Seoul.

Manish Agarwal said the money would also go into research, building trust with labs and enterprises and a wider international presence. He pointed to what he described as a $10bn opportunity in data infrastructure and said the company aims to reach $1bn in annual revenue.

Entrackr reported that in April 2026 the company had announced a $20m commitment to expand data collection in India, Southeast Asia, Latin America and West Asia. That was a spending plan rather than a funding round. None of the sources gave details of an earlier equity raise.

A growing market for robot training data

Inc42 cited a July 2026 report from Counterpoint Research estimating that data collection and annotation for physical AI could become a $10bn revenue opportunity in India by 2030.

Inc42's article also noted concerns raised by this kind of work, including privacy, informed consent, who owns the data, how workers are paid, and the risk that the people being recorded help train machines that could later take over their jobs.

Company website: humynlabs.ai