From Reactive to Predictive: Vision AI in Action at VA SYD and EMASESA

Why we wrote this paper

For decades, utilities have relied on scheduled manual inspections to monitor remote assets — a model that is expensive, reactive, and leaves most assets without visibility between visits. Battery-powered cameras, embedded connectivity and ultra-light AI training have changed the economics. Continuous visual monitoring is now possible at the scale of a full network, not just at the most critical sites.

Our new white paper, From Reactive to Predictive, looks at what actually changes in operations when inspection becomes continuous. It draws on two live deployments — VA SYD in Skåne, Sweden, and EMASESA in Seville, Spain — and the operational results their teams are seeing today.

The thesis: continuous inspection, condition-based action

The combination of battery-powered sensors with embedded connectivity and ultra-light AI training collapses the cost and complexity of remote monitoring. Utilities can now deploy continuous visual monitoring at the scale of their network — not just at their most critical assets. The result: fewer truck rolls, faster response, and condition data across infrastructure that was previously inspected only when something broke.

Three numbers from two live deployments

The paper highlights three figures that summarise what changes in practice when inspection becomes continuous:

• −66% site visits — VA SYD's deployment eliminated two-thirds of scheduled inspection trips.

• 1.2M consumers served — EMASESA, Spain's 4th-largest water utility, runs continuous visibility at scale.

• ~15 days to actionable data — a handful of training images per location is enough to go from install to operational insight.

What's inside

From Reactive to Predictive walks through the operating shift step by step:

• The legacy problem — why scheduled inspection has been the default for 50 years, and the real cost of that model.

• How Waltero works — battery-powered W-Sensors, embedded connectivity and the Mimir cloud platform turning images into condition data.

• Fast deployment — a handful of training images per location is enough to reach operational insight.

• Case: VA SYD — how the Skåne utility cut scheduled inspection trips by two-thirds across thousands of water inlets.

• Case: EMASESA — how Spain's 4th-largest water utility, serving 1.2 million people in Seville, scaled continuous visibility across its network.

• Reactive to condition-based — the operating model shift, and what it means for cost, response times and asset risk.

• A multi-modal future — where vision AI fits alongside other sensing modalities in tomorrow's utility operations.

Who it's for

If you operate distributed infrastructure — water inlets, stormwater grates, pumping stations, meters, overflow chambers — and your monitoring still depends on a technician arriving in a van, this paper is for you. It's written for utility operations leaders, asset managers, and digitalisation teams evaluating whether continuous, AI-assisted monitoring can replace scheduled inspection at scale.

Get the white paper

Read the full white paper, From Reactive to Predictive, for the deployment detail behind the headline numbers — including how VA SYD and EMASESA structured their rollouts, what their teams measured, and where they're heading next.

Download the white paper (PDF)