Jean-Baptiste Kempf built Kyber to run remote robot control in real time

Jean-Baptiste Kempf is doing what a certain kind of engineer always ends up doing eventually: taking the boring parts of “it works” and turning them into infrastructure that everyone else can build on. Kempf, a French serial entrepreneur and open-source legend, has been building Kyber. The goal is straightforward but high-stakes in practice: Kyber is an infrastructure layer to control remote devices in real time.

That phrase, “control remote devices in real time,” matters because real-time control is unforgiving. In many software products, lag is annoying. In robotics and remote operation, lag can be dangerous, expensive, or just plain unusable. Kempf’s pitch here is essentially to make the plumbing dependable enough that remote systems can behave like they are connected at the speed that the task requires. In other words, he is taking the kind of smooth performance work that helps users feel frictionless interactions and applying it to the nervous system of robotics control.

To understand why this is interesting beyond one founder’s project, it helps to zoom out to how remote device control usually gets built. Teams can create custom pipelines that move commands from a controller to a device, receive telemetry back, and try to close the loop quickly. But every new deployment tends to recreate the same problems: networking variability, latency spikes, synchronization challenges, and the operational burden of making all that reliable across different environments. When those pieces are handcrafted for each robot or each site, scaling becomes a tax. An “infrastructure layer,” like Kyber is described, is the attempt to shift that tax from every product team to a shared layer.

Kempf’s background also signals the approach. The source frames him as an open-source legend. In practice, that typically means two things for how Kyber could influence the ecosystem. First, infrastructure layers in open-source-friendly form can reduce integration effort. If developers can rely on a standardized control substrate, they spend less time reinventing connection logic and more time on application logic, like safety behaviors, task planning, or user interfaces. Second, open infrastructure can accelerate adoption because it gives other builders a clear target. They can build against Kyber rather than building everything around a single proprietary stack that might not travel well.

There is another reason remote real-time control becomes a strategic board-level conversation: governance and compliance. Real-time control systems often intersect with regulatory and liability questions, especially when robots operate outside a lab and into warehouses, hospitals, logistics hubs, or public-adjacent spaces. While the source does not specify regulators for Kyber, the broader pattern is that the more a system influences physical-world outcomes, the more stakeholders care about reliability, auditability, and safety. An infrastructure layer that standardizes how remote control is executed can also standardize how logs are captured, how failures are handled, and how systems behave under degraded network conditions. Even when regulation is not explicitly named, the direction is the same: the closer you get to real-world action, the more compliance considerations become infrastructure requirements.

Now zoom in on the second-order implications for decision-makers. The “remote devices” framing suggests Kyber is not only about one robot on one network. It is about controlling devices across environments where connectivity can vary. That expands the market from robotics vendors to the broader category of organizations that want remote operational control, including industrial automation teams and platform builders who may not want to become networking experts. If a reliable infrastructure layer exists, partnerships can become easier: robot manufacturers can integrate it, system integrators can standardize deployments, and enterprises can evaluate solutions with a clearer picture of what is handled by the platform layer versus what is unique to the application.

For executives watching this space, Kyber also highlights a recurring shift in tech markets. Early wins often come from application features that look impressive. But the enduring advantage tends to migrate to the infrastructure layer, the part that makes everything above it stable, composable, and scalable. Kempf’s project is an example of that migration happening in real time-control robotics. If Kyber can deliver dependable remote control, the companies that benefit most might be the ones that can plug it in quickly and focus their competitive differentiation on the workflow layer, not the network layer.