What if you could communicate with any software using natural language? Imagine entering a command, and AI would translate it into machine-readable instructions, performing tasks on your computer or phone to accomplish your desired outcome.
This is the concept behind Rabbit, a rebranding of the startup originally founded by Sun Microsystems co-founder Vinod Khosla, known as Cyber Manufacture Co. Rabbit is developing a customized AI-powered user interface (UI) layer that sits between users and any operating system.
Jesse Lyu and Alexander Liao, the founders of Rabbit, are creating a platform called OS2, backed by an AI model that, as they claim, can perceive and interact with desktop and mobile interfaces much like humans.
According to Jesse Lyu, “The advancements in generative AI have sparked numerous initiatives in the technology industry to define and establish the next level of human-machine interaction. Our perspective is that the ultimate determinant of success lies in delivering an exceptional end-user experience. Drawing upon our past endeavors and experiences, we’ve realized that revolutionizing the user experience requires a bespoke and dedicated platform and device. This fundamental principle underpins the current product and technical stack chosen by Rabbit.”
Rabbit has secured $20 million in funding from Khosla Ventures, Synergis Capital, and Kakao Investment, valuing the startup at an estimated $100 million to $150 million. It’s not the first company to attempt a natural language interface on top of existing software, as Google’s DeepMind and other projects have explored similar ideas.
However, Rabbit aims to differentiate itself by taking a unique technical approach. While it may resemble robotic process automation (RPA) on the surface, Rabbit claims its core interaction model can comprehend complex user intentions and navigate user interfaces. Rabbit’s model can already interact with various consumer applications, and they plan to extend support to all platforms.
Rabbit’s model is designed to be versatile, capable of handling various software interfaces even as they change over time. To train the model, it requires observing a person using a software interface at least once through a screen-recording app.
Despite Rabbit’s ambitious goals, the model’s robustness in handling diverse real-world scenarios remains uncertain. The company is working on a framework for testing, observing, and refining the model to address these challenges. They also plan to release dedicated hardware to host their platform, although specific details are currently undisclosed.
Rabbit’s ultimate goal is to make their platform accessible to users of all skill levels, enabling them to teach the system how to achieve specific goals within applications. The startup believes it can sustain itself by licensing its platform, refining its model, and selling custom devices. However, it faces competition from established players and emerging challengers in the space. Rabbit remains confident in its vision, emphasizing its focus on end-users and quick deployment of safeguarded systems into production to realize the value of cutting-edge research.
