Facts

• Segment: Data Infrastructure
• Country: United States
• Founded: 2026
• Website: https://www.humanarchive.ai/

Profile sections

• Summary
• Market context
• Facts
• Funding
• Context / Market Position
• Events
• Team
• Sources

Summary

Human Archive collects and labels multimodal human activity data for embodied intelligence. Its data infrastructure covers video, sensor, audio and long-horizon activity data across homes, hotels, restaurants, agriculture, industrial, construction and retail environments.

Market context

Human Archive matters because embodied-AI systems need broad real-world human-interaction data, not just narrow factory datasets. A data supplier focused on scale, quality, and multimodal capture is part of the robotics infrastructure stack.

Facts

• Website: https://www.humanarchive.ai/
• HQ: San Francisco, California, United States
• Founded: 2026
• Segment: Data Infrastructure
• Known funding rounds: 1

Funding

• 2026-05-27 - $8M - Human Archive raises $8.2M for physical-AI training data - source: pulse2.com

Multimodal robotics dataset provider

Human Archive sits in the data layer for robotics and world models, focusing on large-scale real-world human-interaction datasets.

• Target environment: Embodied-AI and world-model training workflows that need diverse human activity data from real environments.
• Deployment model: Data-capture and processing infrastructure with QA, anonymization, and annotation before delivery to model-development teams.
• Customer context: Frontier labs and robotics companies that need large real-world datasets instead of narrow task or factory-only data.
• Workflow context: Collects human sensorimotor data in real environments, cleans and structures it, then feeds it into robotics training and evaluation loops.
• Commercial maturity: Founded in 2026 and publicly launched through Y Combinator in Winter 2026.
• Market position: Positioned as a robotics-data supplier focused on multimodal real-world behavior capture at scale.
• Adoption constraints: Adoption depends on data quality, privacy controls, breadth of environments covered, and evidence that the datasets improve downstream model performance.
• Adjacent context: Comparable context includes embodied-AI dataset companies, robotics data marketplaces, and world-model training infrastructure providers.
• Source confidence: high

Events

• 2026-05-27 - Funding - Human Archive raises $8.2M for physical-AI training data - The funding round totaled $8.2 million.
• 2026-03-08 - Capability - Human Archive builds multimodal robotics dataset infrastructure - HA-Multi combines vision, stereo depth, tactile gloves, IMUs, and wrist cameras.

Team

• Samay Maini - Founder
• Shloke Patel - Founder