Deploying AI on underwater and maritime platforms sounds straightforward until you’re actually doing it. Teams run into the same friction points: integrating sensors from different vendors, building custom middleware, managing data pipelines, and trying to run inference on hardware with no reliable cloud connection. Most of that work has nothing to do with the problem they’re actually trying to solve.

This talk covers how edge-native AI is changing that equation in working commercial systems today. Drawing on real deployments with VideoRay for automated underwater inspection, OceanAero for maritime threat detection on autonomous surface vessels, and WESMAR for intelligent commercial fishing sonar, we walk through what these integrations looked like, what broke, and what made them work. All three teams used NEPI (Numurus Edge Platform Interface) to skip rebuilding the sensor connectivity, edge inference, and data pipeline layers from scratch, freeing them to focus on the actual application.

Attendees will leave with a practical framework for evaluating edge AI for their own systems and a clear-eyed look at where the real integration challenges sit.