Human Archive funding, valuation and investors

Overview

Human Archive is a Silicon Valley data lab building foundational multimodal datasets for robotics and physical AI. Founded by Stanford and Berkeley researchers Raj Patel (CEO), Shloke Patel, Samay Maini (CTO), and Rushil Agarwal, the company deploys custom wearable rigs — head-mounted RGB-D cameras, tactile gloves, wrist cameras, and motion-capture sensors — across homes, factories, restaurants, and warehouses in India and beyond. Workers collect synchronized egocentric demonstrations of everyday physical tasks that frontier robotics labs use to train foundation models. The company raised $8.2M in seed funding in May 2026 led by Wing Venture Capital with Y Combinator (W26), NVP Capital, and angels from OpenAI, Nvidia, Google, and Meta.

Why Human Archive is interesting

Human Archive synchronizes RGB-D, force feedback, and full-body motion capture at scale via 1,000+ deployed headsets — Wing-led $8.2M seed positions it as the 'Common Crawl for embodied intelligence.'

Product & use cases

Two dataset products: HA-Multi (aligned RGB-D, tactile gloves, body IMUs, wrist cameras with 3D hand reconstructions and pose labels) and HA-Ego (mono RGB with task-aligned atomic action labels). Internal QA, anonymization, and annotation pipelines deliver data to robotics foundation-model teams.

• Humanoid and general-purpose robot training on real-world manipulation data
• Frontier AI labs benchmarking multimodal embodied-intelligence models
• Physical-AI companies needing synchronized vision, tactile, and motion data at scale
• Research on cognition, prosthetics, and spatial computing using human sensorimotor corpora

Key facts

• Seed (May 2026): $8.2M led by Wing Venture Capital — TechCrunch
• 1,000+ active headsets deployed; 50+ device types across India and other markets
• Four co-founders dropped out of Stanford/Berkeley during finals week to go full time
• Most requested startup at Y Combinator W26 Demo Day per Wing VC
• Partners with India home-services, hotel, and restaurant gig platforms for capture

Funding history (newest first)

Competitive landscape

Edge: Rare ability to synchronize headset RGB-D, tactile force feedback, full-body mocap, and wrist cameras at scale — Wing VC cited no other team collecting all modalities simultaneously in production.

Physical AI's bottleneck is training data showing humans doing real work — not synthetic demos. Human Archive bets India's gig economy plus custom hardware creates a cost-scalable capture network competitors cannot replicate quickly. Data quality will be tested as customers post-train models on delivered datasets.

• Scale AI (robotics data) adjacent

  General data-labeling giant expanding into robotics; less specialized in synchronized multimodal wearable capture.

• In-house data at Figure AI / Tesla Optimus alternative

  Humanoid companies collect proprietary teleoperation data; Human Archive sells third-party datasets to labs without internal capture.

• Synthetic simulation (NVIDIA Isaac, etc.) alternative

  Simulated data scales cheaply but lacks real-world tactile and contact physics Human Archive captures.

• Teleoperation startups direct

  Various robotics-data vendors collect egocentric video; fewer combine tactile and mocap at Human Archive's modality depth.

Notable stories

• The four co-founders have known each other 20+ years and dropped out of Stanford and Berkeley during finals week to build Human Archive full time — then deployed hardware across San Francisco, Germany, China, and India in months (Wing VC blog).
• Wing partner Zach DeWitt said no other team synchronizes headset RGB-D, force feedback, full-body mocap, and wrist cameras at scale — every major lab wants to run experiments on the dataset (TechCrunch, May 2026).

Industries

Related funding articles

Venture Capital Tracker pieces that cover Human Archive's financing or category context.

• Human Archive's $8.2M Seed: India's Gig Economy Trains the Robots