Omnisent builds distributed acoustic AI networks - ultra-low-power sensor mesh designed to make sound a persistent, passive intelligence layer for defense and critical infrastructure. The Munich-based startup, founded in 2024, has raised a $3 million pre-seed to commercialize what it calls a Large Acoustic Model (LAM): a foundation model trained on non-speech audio that runs on-device for real-time detection, classification, and localization of acoustic events. The pitch is deliberately anti-radar: silent, jam-resistant, zero-emission sensing that works where GPS and electronic warfare make traditional approaches brittle.
The operational constraints are specific. Each node operates on battery power for over a year. The network deploys in under ten minutes. The company claims 99.9% uptime across its deployments, targeting use cases from perimeter defense to critical infrastructure protection where acoustic signatures - vehicles, drones, anomalous mechanical events - matter more than visual ones. It's a contested-environment bet: the value proposition scales directly with the jamming and spoofing threat model.
Technically, the stack spans acoustic AI, large acoustic models, distributed sensing, and ultra-low-power edge compute. The company is a participant in the Cambridge Accelerate Plus venture program and is currently hiring across engineering disciplines. Based in Munich, Omnisent is early-stage but operating in verticals - defense, homeland security, critical infrastructure - where procurement cycles are long and technical credibility is non-negotiable.
