Industrial IoT is usually discussed at the cloud or dashboard layer, but the hardest problems often live much closer to the machines.

This talk shares lessons learned building Dredger-IoT, an open-source Ruby-based telemetry agent designed to run at the edge, collect industrial data, and feed modern analytics and AI systems without rewriting existing infrastructure.

Rather than focusing on hype or vendor tooling, we will explore the practical realities of edge systems:

Interfacing with legacy equipment and protocols

Operating in unreliable, resource-constrained environments

Designing data pipelines that are resilient before they are “intelligent”

From there, we look at where AI actually fits — not as a replacement for good engineering, but as a force multiplier once reliable telemetry exists.

Topics include:

Why most AI initiatives fail before data ever reaches the model

Designing edge software that survives the real world

Using lightweight agents to bridge legacy systems to modern platforms

Open-source tradeoffs when building industrial software

This session is grounded in real code, real deployments, and real mistakes. It is aimed at developers and team leads who want to understand how edge data, open source, and AI intersect outside of idealized architectures.

Attendees will leave with a clearer mental model of edge-first systems and concrete ideas they can apply in their own environments.