Intelligence Factory
About
About Intelligence Factory
Intelligence Factory builds a data and model layer for robot manipulation that is independent of one hand or gripper design. Its data factories use custom gloves to capture what a person sees, does and feels while completing physical tasks. Human hand trajectories and contact signals are then retargeted across robot hardware, allowing the same demonstration source to train five-finger hands, simpler grippers and other manipulators.
The company uses those datasets to develop general-purpose manipulation models for work in warehouses, grocery operations and data centres. A deployment pairs the model with the customer’s robot, perception stack and task environment, then fine-tunes behaviour from real demonstrations and operating feedback. This separates manipulation intelligence from the machine body: robot manufacturers can change hardware while retaining a common skill and data pipeline, and end users can collect additional examples inside the workplace where the model will operate.
Team
- Yash Sinha - Co-founder and CEO
- Jalaj Shukla - Co-founder and CTO
Timeline
Events
- 2026-05-30 - Capability - Intelligence Factory discloses its human-demonstration manipulation pipeline
The system collects human demonstration data through gloves. The pipeline retargets human motion into robot motion. Target environments include warehouses, grocery stores, and data centers.
Source
Context
Comparable profiles
- Flexion
stack adjacency - Embodied AI Model Layer
- MindOn
stack adjacency - Embodied AI Model Layer
- XYZ Embodied AI
stack adjacency - Embodied AI Model Layer
- ACE Robotics (大晓机器人)
stack adjacency - Embodied AI Model Layer
- Delta Intelligence
stack adjacency - Embodied AI Model Layer
- ForgeAI
stack adjacency - Embodied AI Model Layer
- LiberAI
stack adjacency - Physical AI Training Data
- Aether AI
stack adjacency - Embodied AI Model Layer
- Lupa Robotics
stack adjacency - Physical AI Training Data
- Generalist
stack adjacency - Embodied AI Model Layer
- Algomatic Dynamics
stack adjacency - Physical AI Training Data
- Striding AI / 正行创新
stack adjacency - Embodied AI Model Layer
Shared workflows
- Physical Intelligence
workflow adjacency - Cross Embodiment Robot Learning
- RoboScience / 机器科学
workflow adjacency - Cross Embodiment Robot Learning
- Reward AI
workflow adjacency - Cross Embodiment Robot Learning
- Lingsheng Technology
workflow adjacency - Cross Embodiment Robot Learning
- Dream Labs
workflow adjacency - Cross Embodiment Robot Learning
Context tags
- Manipulation foundation model - Capability - The company explicitly develops general-purpose manipulation intelligence for robots.
- Cross-embodiment robot learning - Workflow - The company retargets human motion across five-finger hands and grippers.
- Embodied AI model layer - Stack Layer - Its stated product is an intelligence layer that transfers across robot hardware.
- Physical AI training data - Stack Layer - Custom gloves capture human demonstrations used to train its models.