Facts

• Segment: Foundation Models
• Country: United States
• Founded: 2026
• Website: https://intelligence-factory.com/

Profile sections

• Summary
• Market context
• Facts
• Context / Market Position
• Events
• Operations / Deployments
• Team
• Sources

Summary

Intelligence Factory develops foundation models for embodied agents and robotics. The company works on AI systems that can support physical-world perception, planning and action, with a focus on general-purpose intelligence for robots rather than a single narrow automation product.

Market context

Intelligence Factory matters because it is trying to build a body-agnostic manipulation intelligence layer rather than another single robot platform. That makes it relevant to the embodied-model stack, not just to one hardware form factor.

Facts

• Website: https://intelligence-factory.com/
• HQ: San Francisco, California, United States
• Founded: 2026
• Segment: Foundation Models

General-purpose manipulation intelligence

Intelligence Factory sits in the embodied-model layer, using human demonstration data to train manipulation intelligence that can transfer across robot hardware.

• Target environment: Robot-learning workflows for manipulation in warehouses, grocery settings, data centers, and other unconstrained environments.
• Deployment model: Develops model infrastructure and data-collection systems rather than selling a single robot embodiment as the core product.
• Customer context: Robot builders and embodied-AI teams that need manipulation intelligence capable of transferring across tasks, objects, and hardware.
• Workflow context: Captures what humans see, do, and feel during tasks, then uses that data to train and retarget policies across different manipulators.
• Commercial maturity: Founded in 2026 and publicly launched through Y Combinator in Spring 2026.
• Market position: Positioned in the embodied foundation-model layer with a manipulation-specific focus and in-house human-data capture.
• Adoption constraints: Adoption depends on data quality, model transfer across hardware, and proof that the approach generalizes outside narrow benchmark tasks.
• Adjacent context: Comparable context includes embodied foundation-model developers, robot imitation-learning companies, and human-data capture platforms for manipulation.
• Source confidence: high

Events

• 2026-05-29 - Capability - Intelligence Factory builds human-demonstration pipeline for robot manipulation - The system collects human demonstration data through gloves.

Operations / Deployments

• 2026-05-29 - Capability - Intelligence Factory builds human-demonstration pipeline for robot manipulation - The system collects human demonstration data through gloves.

Team

• Jalaj Shukla - Co-founder and CTO
• Yash Sinha - Co-founder and CEO