“The things we are doing together with causaLens is about how we get all the data from the sensors, processors & devices - a lot of raw data - and predict all the potential benefits”

When people think of AI, they think of massive data centers, racks of GPUs, and cloud-based APIs. But a quiet revolution is underway—and it’s not happening in the cloud.

It’s happening on the edge.

In a forward-looking fireside chat at the cAI Conference, Qualcomm LATAM President Luis Tonisi joined causaLens CEO Darko Matovski to talk about their strategic partnership and a future where AI agents run directly on devices—from drones to smartphones to industrial sensors.

“Today, we still imagine AI as something that lives in a cloud. But for many real-world problems, you can’t afford the latency, the cost, or the privacy risk,” said Tonisi. “Inference is coming to the edge. And agents are the new user interface.”

 

From Vision to Action: The Edge Use Case Boom

Luis didn’t speak in abstractions, he brought examples from the frontlines. Qualcomm’s partnership with Aramco, one of the largest oil and gas companies in the world, is already putting AI to work on the edge.

Here are just a few of the use cases he described:

• Surveillance agents that monitor restricted zones and trigger alerts in real time, without needing to replace legacy cameras.
  
• Drone dispatch agents that send unmanned vehicles to investigate anomalies when security is breached.
  
• Field service agents that let a technician snap a photo of a machine and immediately receive diagnostics, documentation, and repair instructions, automatically.

These aren’t ideas. They’re live or in development. And all of them share a common constraint: they must happen in real time, in remote or privacy-sensitive locations, with or without cloud connectivity.

That’s the edge. And it’s where agents thrive.

 

Why Agents + Edge Is a Breakthrough

What’s new isn’t just that AI is moving closer to where the data is, it’s who the AI is.

Luis made a clear prediction: the user interface of the future isn’t apps, it’s agents. “You won’t be opening ten apps on your phone anymore,” he said. “You’ll ask your agent, and it will figure out which app to call, what data to fetch, and how to act.”

And crucially, these agents won’t rely entirely on the cloud. In many cases, due to bandwidth, cost, or compliance, they’ll run inference locally.

This is where causaLens comes in. The company’s AI Data Scientists—autonomous agents that manage entire analytics workflows- are now being deployed in environments that previously couldn’t support advanced AI. Together, causaLens and Qualcomm are removing the need for complex analytics stacks, no databases, no ETL tools, no external modeling platforms. Just one agent, running locally, doing the work of an entire data team.

 

Industry Implications: The New Intelligence Layer

The implications go far beyond oil & gas. Across industries; logistics, agritech, automotive, smart cities, AI agents on the edge offer a leap forward in:

• Latency-sensitive applications (e.g., fraud detection, emergency response)

• Data privacy and sovereignty (e.g., healthcare, defense, critical infrastructure)

• Offline operations in rural or remote environments

• Cost efficiency through cloud compute reduction

“We’re just at the beginning,” Tonisi said. “The real challenge now is to create a seamless agent experience across devices and form factors, your phone, your car, your factory.”

 

A New OS War Is Coming

One of the most compelling moments of the conversation came when Luis spoke about Qualcomm’s work with device manufacturers building “hyper operating systems”, agent orchestrators that run across smartphones, tablets, and even vehicles.

“The real fight,” he said, “will be over who owns the agent.”

Just as operating systems once controlled the device experience, agent platforms may soon control the enterprise experience, navigating apps, making decisions, and managing everything from infrastructure to insights.

In that world, agent platforms like causaLens don’t just deliver analytics. They deliver intelligence infrastructure.

Final Thought: The Cloud Doesn’t Die—But It Shifts

Neither Tonisi nor Matovski believe the cloud is going away. Training will still happen in large data centers. Central systems will still orchestrate fleet-wide learning. But inference, doing things, will increasingly happen at the edge.

“It’s not about ‘cloud vs. edge,’” said Luis. “It’s about the right compute in the right place, with the right agent making the right decision.”

For companies building the next generation of AI experiences, that may be the most important insight of all.