Skild AI Unveils Groundbreaking Progress in General-Purpose Robotic Brain Development
Just one year after emerging from stealth mode, Skild AI, a pioneering AI robotics company, has released a first-of-its-kind video showcasing its remarkable progress in developing a scalable foundation model for robotics. The footage provides an exclusive look at the capabilities of Skild Brain, the company’s general-purpose robotic brain designed to operate across diverse robot embodiments, environments, and tasks. This milestone underscores Skild AI’s commitment to advancing Physical AI—a shared, adaptable system that can tackle both simple and complex challenges with unprecedented versatility.
Addressing Moravec’s Paradox with a Unified Approach
Robotics has long been constrained by Moravec’s paradox, which observes that tasks easy for humans—like climbing stairs or assembling objects—are often incredibly difficult for robots, while tasks challenging for humans—such as performing precise dance moves or martial arts routines—are relatively straightforward for machines. Deepak Pathak, CEO and Co-Founder of Skild AI, explained this dichotomy: “A lot of current robotics models focus on tasks that are hard for humans and easy for robots, such as dancing or kung-fu, because they involve free-space actions and don’t require much generalization.”
What sets Skild AI apart is its ability to tackle both ends of this spectrum. Pathak elaborated: “Skild AI models can not only solve these simpler tasks but also handle everyday hard tasks like climbing stairs under adversarial conditions or assembling fine-grained items, which demand vision, reasoning about contact dynamics, and adaptability.” This dual capability highlights the robustness and flexibility of Skild Brain, positioning it as a transformative solution in the field of robotics.
A Truly General-Purpose Robotic Brain
Unlike many competitors who adopt a vertical approach—focusing on specific robots or tasks—Skild AI emphasizes generality. Abhinav Gupta, President and Co-Founder of Skild AI, emphasized this point: “Skild AI showcases the general-purpose nature of our brain: the same AI model can be adapted to different tasks and robot morphologies. From quadrupeds to humanoids, table-top arms to bimanual autonomous mobile robots (AMRs), and home tasks like loading dishwashers to physically demanding feats like climbing slippery slopes, Skild Brain is a one-stop AI model for all forms of robotics.”
This adaptability is a game-changer. Whether deployed in industrial settings, households, or outdoor terrains, Skild Brain demonstrates consistent performance across varied scenarios. Moreover, the system’s robustness ensures it can safely coexist with humans. Thanks to end-to-end AI training methods that prioritize low-force interactions, Skild Brain excels in safety and adaptability, seamlessly adjusting to disturbances and human interference—a feature often lacking in other systems.
Solving the Data Challenge Through Innovation
One of the most significant hurdles in creating a robotics foundation model is the scarcity of large-scale robotics data. Collecting real-world data using physical hardware is slow, expensive, and logistically complex. Skild AI has overcome this obstacle through innovative strategies: leveraging large-scale simulations and human videos available online to pre-train its foundation model. By combining these vast datasets with targeted real-world data during post-training, Skild AI delivers tailored solutions to meet customer needs.
The company’s approach expands the scope of training by incorporating human data, enabling models to work across various morphologies—from quadrupeds and humanoids to tabletop manipulators and mobile systems. This diversity in training data equips Skild Brain with the ability to generalize effectively, making it a truly universal physical foundation model capable of adapting to any task or environment.
A Legacy of Robotics Breakthroughs
Behind Skild AI’s groundbreaking advancements is a team of world-class researchers and engineers whose contributions have shaped modern robotics. Members of the Skild AI team have been instrumental in key breakthroughs such as self-supervised learning, curiosity-driven exploration, large-scale imitation learning, sim-to-real locomotion, parkour navigation, dexterous manipulation, and learning from internet-scale video. Their expertise spans leading organizations including Anduril, Tesla, Nvidia, Meta, Kitty Hawk, Google, Everyday Robotics, and Amazon, bringing unparalleled depth and breadth to the project.
This collective experience has enabled Skild AI to build production-ready systems that push the boundaries of what’s possible in robotics. With a foundation rooted in cutting-edge research and practical applications, the company is poised to redefine how robots interact with the world.
Backed by Visionary Investors
Skild AI’s ambitious vision has attracted support from some of the most prominent names in venture capital and technology. The company is backed by investors including Lightspeed Venture Partners, Coatue, SoftBank Group, Jeff Bezos (through Bezos Expeditions), Felicis Ventures, Sequoia, Menlo Ventures, General Catalyst, CRV, Amazon, SV Angel, and Carnegie Mellon University. This strong backing reflects confidence in Skild AI’s potential to revolutionize the robotics industry.
Paving the Way for a New Era in Robotics
The release of Skild AI’s video marks a pivotal moment in the evolution of robotics. By addressing longstanding challenges like Moravec’s paradox and the lack of scalable data, the company has laid the groundwork for a future where robots are no longer confined to narrow use cases but can seamlessly adapt to any task or environment.
From navigating complex terrains to performing delicate assembly work, Skild Brain exemplifies the power of a unified, general-purpose AI model. Its ability to integrate safety, adaptability, and scalability positions it as a cornerstone for next-generation robotics applications. As Skild AI continues to refine and expand its platform, the possibilities for innovation are virtually limitless.
In a world increasingly reliant on automation and intelligent systems, Skild AI is charting a bold path forward—one where robots become indispensable partners in solving humanity’s greatest challenges. With its visionary leadership, cutting-edge technology, and unwavering commitment to progress, Skild AI is set to transform industries, empower businesses, and improve lives worldwide.
About Skild AI
Founded in 2023 by two pioneers in the field of self-supervised and adaptive robotics, Skild AI is building a scalable foundation model for robotics that serves as a shared, general-purpose brain across diverse robot embodiments and a variety of real-world applications. The company has offices in Pittsburgh and the San Francisco Bay Area.
