For the last decade, we taught machines to talk.

We connected billions of sensors, machines and devices and gave ourselves unprecedented visibility into the physical world.

But visibility was never the destination.

The real opportunity is what happens when machines stop simply telling us what is happening — and start deciding what to do about it.

That transition is already underway. Across factories, energy systems and industrial infrastructure, AI is moving beyond dashboards, predictions and copilots toward intelligent agents capable of making decisions and, increasingly, acting on the physical world.

But making a prediction is very different from making a decision.

The world's best industrial operators don't run factories by analyzing data alone. They combine data with experience, intuition, constraints and decades of accumulated knowledge about how a process actually behaves. Much of that expertise has never been written down — and cannot simply be scraped from the Internet or found in a training dataset.

So how do we teach machines to make decisions we are prepared to trust?

This keynote explores the next evolution of IoT: from connected things to thinking systems. It looks at how human expertise, operational data, digital twins, simulation and teams of specialized AI agents are coming together to create a new generation of autonomous industrial systems.

And it examines one of the most important questions facing our industry: before an AI is allowed to control something that matters, how does it prove that it is ready?

The next decade of IoT won't be defined by how many things we connect.

It will be defined by what we trust them to do.