MachineGenes Information

The MachineGenes Group has developed entirely new paradigms of evolutionary ML and adversarial AI, with significant advantages over Artificial Neural Networks (ANNs). Using our MachineGenes AI platform we are developing applications that enable users to maintain stable control of complex dynamic systems. We use tiny fragments of partial sensor data to create a digital twin of the systems: a bionic human organ or a turbine engine, that simulates that complex dynamic system. We then apply an adversarial AI to the digital twin that attempts to predict how the ‘real’ system is going to behave so it can generate and recommend strategies to keep it operating in a desirable state. This has four key advantages over (ANNs) critical shortcomings; - Explainable: All recommendations are full explainable from the ground up – vital in mission critical applications - Controllable: Human users can be in meaningful dialogue with the AIs, requesting a change in strategy at any point - Lightweight: AI models are build using tiny 1st party data sets (~100kB), no need for third-party data such as massive LLMs - Efficient: AIs engage in adversarial processes on isolated Edge devices, no need for resource-hungry uplinks to the Cloud The first application of the MachineGenes platform is Diabetes Neuromathix (DNx), a precision medicine start-up enabling the real time adaptive control of insulin levels of people with highly unstable Type 1 diabetes to enable them to live productive fulfilling lives. With patents granted in multiple markets and negotiations underway with investment, research and commercial partners we are coming out of stealth mode and are excited to bring the power of evolutionary AI to this critical industries and many more.