Kyber Aims to Revolutionize Remote Device Control with Real-Time Infrastructure Layer

Jean-Baptiste Kempf, the creator of the VLC media player, is developing Kyber, an open-source infrastructure layer designed for the real-time control of remote devices. According to a June 20, 2026, report from TechCrunch, the project applies Kempf’s expertise in low-latency data streaming to the field of “physical AI” and robotics to enable precise, remote hardware manipulation.

Kyber functions as a communication layer that sits between AI controllers and physical hardware. The system aims to solve the latency and stability issues that typically plague remote-controlled robotics, where a delay of a few milliseconds can result in mechanical failure or safety hazards. TechCrunch reports that Kempf is leveraging the same principles of efficiency and hardware abstraction that allowed VLC to run smoothly across thousands of different device configurations.

How does Kyber enable real-time device control?

Kyber focuses on the transport and execution of commands in real time, prioritizing speed and reliability over the bulk data transfer typical of standard cloud IoT platforms. While traditional IoT systems often rely on asynchronous messaging—where a command is sent and the device eventually acknowledges it—Kyber is designed for synchronous, high-frequency updates.

This infrastructure is specifically targeted at physical AI, which refers to artificial intelligence embodied in physical forms like robotic arms, drones, or autonomous vehicles. These systems require a constant, high-speed feedback loop between sensors and actuators. According to the project’s objectives, Kyber provides the plumbing necessary for these AI models to interact with the physical world without the lag induced by standard networking stacks.

Why is the creator of VLC building a robotics platform?

The transition from media players to robotics is a shift in application, but not in technical logic. Jean-Baptiste Kempf built VLC to handle fragmented codecs and diverse hardware, ensuring a seamless stream of data regardless of the device. Kyber applies this “universal” approach to robotics, attempting to standardize how remote commands are delivered to diverse hardware components.

The project mirrors the open-source philosophy of the VideoLAN project. By keeping the infrastructure layer open, Kempf aims to prevent the fragmentation of the robotics industry, where proprietary controllers often lock developers into specific hardware ecosystems. This approach allows developers to swap out hardware components without rewriting the entire control layer.

How does Kyber differ from existing IoT infrastructure?

Most current IoT frameworks are designed for telemetry and simple triggers, such as turning on a light or reporting a temperature sensor. Kyber is built for active control. The following distinctions separate Kyber from standard IoT layers:

• Latency Requirements: Standard IoT often accepts delays of several seconds; Kyber targets millisecond-level responsiveness.
• Data Flow: While IoT focuses on data collection (upload), Kyber prioritizes command execution (download) and real-time telemetry (bi-directional).
• Hardware Abstraction: Kyber seeks to create a common language for remote devices, similar to how VLC handles various video formats, rather than relying on vendor-specific APIs.

What are the implications for physical AI?

The development of Kyber addresses a primary bottleneck in the deployment of AI-driven robotics: the “last mile” of connectivity. Even the most advanced AI models cannot operate effectively if the command to stop a robotic arm arrives after the arm has already collided with an object.

By providing a stable, open-source layer for remote control, Kyber could lower the barrier to entry for companies building autonomous systems. Developers can focus on the AI’s decision-making capabilities rather than the underlying network engineering required to move a physical joint in real time.

Kempf’s history with VLC suggests a focus on cross-platform compatibility. If Kyber achieves similar ubiquity, it could lead to a standardized ecosystem where AI agents can control a wide array of third-party robotic hardware through a single, unified infrastructure layer.