Portuguese Startup Uses AI to Predict Urban Flooding 24 Hours in Advance — and It's Already Saving Cities

When December 2022's floods paralyzed Lisbon, three engineers from Instituto Superior Técnico realized their sensors—initially designed to monitor Iberian lynx migrations—could do something far more valuable: predict urban flooding before it happens.

Today, their company, Greenmetrics.ai, operates intelligent sensor networks across 11 Portuguese municipalities, providing real-time flood warnings that have transformed emergency response from reactive guesswork into precision operations.

From Wildlife Tracking to Saving Cities

Tiago Marques, CEO of Greenmetrics.ai, remembers the pivot moment clearly. "We were monitoring golf course irrigation and wildlife when Lisbon flooded. We realized the same data architecture could save entire cities," he told Dinheiro Vivo this week.

The solution is deceptively simple: install intelligent sensors in critical urban "hot spots"—storm drains, sewer collectors, and urban streams flowing through densely populated areas. These sensors measure water level rises in real time, capturing millimeter variations that human observers and traditional statistical models miss until it's too late.

"There are no omelets without eggs: if you want AI, you need quality data," Marques explains. "We install the printers that generate that data directly at the source."

The Cold Email That Changed Everything

Without formal introductions or political connections, the startup sent a "cold email" to Lisbon's Civil Protection office. What could have been dismissed as spam caught attention.

"Suddenly we were in a room with 10 experienced Civil Protection operatives, hungry for solutions to manage uncertainty," Marques recalls. The pilot project proved the technology could withstand extreme humidity, debris, and the isolation of underground conduits.

That success opened doors across Portugal. Today, Greenmetrics.ai works with Lisbon, Oeiras, Cascais, Porto, Maia, Guimarães, Loures, Odivelas, Tavira, Loulé, and Torres Novas—adapting algorithms to each city's unique urban morphology.

From 1 Hour to 24-Hour Warnings

During the recent storm sequence that battered Portugal—Kristin, Leonardo, and Marta—the system proved its worth. In Cascais, the Ribeira das Vinhas alert reached operation centers 1 hour and 20 minutes before Largo de Camões flooded. That window allowed shopkeepers to protect their businesses and remove valuable inventory.

But the company isn't stopping there. "In a year, we want to predict floods 24 hours in advance," Marques says confidently. The key lies in AI's "snowball effect" and feature engineering. "Our AI learns more the more cities it's deployed in. It learns from Porto and applies knowledge in Leiria."

By transforming conduit and slope data into location-agnostic information, the system can anticipate water behavior in new cities based on what it learned in dozens of similar scenarios elsewhere.

From Brazil to the American Dream

With a market valuation already reaching €7 million, Greenmetrics.ai is expanding internationally. The startup already operates in Brazil and is preparing to enter the United States market—where urban flooding causes billions in annual damage and where predictive technology commands premium valuations.

"The American market is the dream," Marques admits. "The infrastructure challenges are enormous, and municipalities are willing to pay for solutions that work."

Why This Matters for Portugal

For a country increasingly vulnerable to extreme weather events, Greenmetrics.ai represents more than just a successful tech export—it's a homegrown solution to a homegrown problem.

Urban flooding costs Portuguese municipalities millions annually in infrastructure damage, business interruption, and emergency response. Early warning systems that actually work don't just save money—they save lives.

For foreign residents and expats, particularly those in flood-prone areas like Lisbon's riverside neighborhoods or Porto's historic center, the system offers something increasingly rare: predictable warning time to protect property and plan around severe weather.

The technology also positions Portugal as an exporter of climate-adaptation solutions rather than just an importer of foreign tech—a small but significant reversal in a country that has long struggled to commercialize its academic research.

As climate instability becomes the new normal, Greenmetrics.ai's sensors watching from beneath Portugal's streets might represent the kind of innovation-driven future the country needs to weather what's coming—literally.