1 Startup · $10.8M · Industrial Adaptive AI

What they're building

Luffy builds what it calls neuroplastic AI: sparse neural networks trained in simulation — no large training datasets required — then refined against the real machine, where the company claims up to 400x greater efficiency than conventional deep learning. The models are small enough to embed in existing hardware and self-refine from live feedback, so there's no cloud dependency or constant retraining. First target: the electric motor. Around half the world's electricity is consumed by electric motors, most running inefficiently, and Luffy is deploying its Adaptive Neural Controllers into motor control and variable-frequency-drive applications — industrial pumps, fans, conveyors — so a motor can tune itself to its load in the field. Longer term, the same stack targets positioning control for robotics and drones, thermal process control, and broader physical AI.

Why this matters

The AI boom has mostly stepped around the factory floor because LLM-style approaches need data, compute, and connectivity that a pump or conveyor doesn't have. Luffy attacks exactly that gap, and BGF's investor framed it as disrupting an industry norm that has stood for a hundred years — replacing specialist commissioning engineers with self-commissioning control. The founders, Dr Matthew Carr (CEO) and Dr Alex Meakins, are former nuclear physicists from the UK Atomic Energy Authority, and the company sits on the Culham fusion campus — deep technical credibility for a control-systems problem. The syndicate mixes UK growth capital (BGF) with Munich deep-tech money (MIG Capital AG), which matters for landing German industrial customers. With energy efficiency regulation tightening and electricity demand spiking, plug-and-play efficient motor control has a real commercial 'why now'.