Skild AI builds physical AI for robotics, centered on a unified, omni-bodied brain designed to control any robot for any task. Its positioning emphasizes bringing intelligence out of cyberspace and into real-world environments where robots must perceive, move, and act under messy, changing conditions.
Skild AI frames its approach around a robotics foundation model that generalizes across robot types and task types rather than relying on separate, narrowly engineered control stacks. The company pairs that long-term model ambition with real robot applications intended to deliver tangible operational value in areas such as navigation, manipulation, inspection, and repetitive physical work.
Offerings, Capabilities, and Integrations
Skild AI’s core capability is generalized robot control across embodiments such as quadrupeds, humanoids, tabletop arms, and mobile manipulators. Its architecture combines higher-level manipulation and navigation policies with lower-level control that converts intent into joint angles and motor torques, and it is trained using large-scale simulation, internet human video, teleoperation, and deployment data.
Beyond model development, Skild AI is building deployment pathways with robotics and simulation partners. It has described partnerships with ABB Robotics, Universal Robots, and Mobile Industrial Robots to extend its intelligence into industrial fleets, and it works with NVIDIA Isaac Lab, Isaac Sim, and Cosmos to support simulation, synthetic data generation, and sim-to-real transfer.
Products and Services
- Skild Brain: Skild Brain is Skild AI’s unified robotics foundation model, designed to generalize across robot hardware and tasks and operate machines ranging from quadrupeds and humanoids to tabletop arms and mobile manipulators.
- Security/Inspection Robot Platform: Security/Inspection Robot Platform applies Skild Brain to robots that navigate unstructured and potentially dangerous environments so organizations can reduce constant manual inspection work.
- Mobile Manipulation Platform: Mobile Manipulation Platform uses Skild AI’s control stack for skills such as grasping, handover, and navigation on mobile platforms, with those skills abstracted through an API-oriented interface for application building.
- Autonomous Packing: Autonomous Packing is a dexterous automation offering focused on highly precise, repetitive packing-related tasks.
Target Customers
Skild AI targets robotics OEMs, automation providers, and enterprises that need robot intelligence to transfer across changing hardware, workflows, and environments. Its recent partner activity points especially to manufacturers and industrial operators that want more flexible automation without task-by-task reprogramming.
Its application focus also aligns with organizations in security, inspection, warehousing, delivery, construction, data centers, and factory assembly, as well as environments where labor shortages, hazardous conditions, or repetitive physical tasks create a strong case for adaptive robotics. Skild AI also signals longer-term relevance for healthcare, hospitality, and home settings as its model expands from semi-structured to less structured environments.
Cloud Integrations and Marketplace
- NVIDIA Isaac Lab, Isaac Sim, and Cosmos: Skild AI integrates with NVIDIA’s robotics and physical AI stack, using Isaac Lab and Isaac Sim for simulation-driven robot learning and Cosmos for synthetic data generation and augmentation to improve robustness and sim-to-real transfer.
Key People
- Deepak Pathak: CEO and Co-Founder
- Abhinav Gupta: President and Co-Founder
- Aditya Raghunathan: Chief of Staff
- Dominic Haury: Creative Lead
Key Facts
- Headquarters: Pittsburgh, Pennsylvania, United States
- Employees: Approximately 92
- Annual Revenue: Approximately $30M
- Parent Company: None
- Subsidiaries: None
- Publicly Listed: No
