by

To protect us against earthquakes, SeismicAI needs just three seconds

The company uses technology to predict earthquakes faster than any system - offering seconds that could save millions of lives

James Spiro 10:1530.10.21

As far as natural disasters go, earthquakes can be one of the worst. Today one-third of the global population lives in a seismic area and can be exposed to them at almost any time. And while they may differ in size and magnitude, the unpredictability of them can mean billions of dollars in damages and thousands of lives potentially lost should they choose to strike an unknowing population.

“No one can tell you you’re going to have an earthquake in two days - not even in two minutes from now,” explained SeismicAI’s CEO, Benjamin Sasson. Sasson originally joined the company in 2017 as CTO and earlier this year became its CEO. The company claims to have developed a system using a mix of algorithms developed at Tel Aviv University with AI to help countries identify when an earthquake is about to strike. And while the purpose isn’t to prevent one, SeismicAI can provide between 3-90 seconds of warning that might just be the difference between life or death.

SeismicAI CEO Benjamin Sasson. Photo: SeismicAI

“Of course we can not stop buildings from falling, but we can reduce a lot of the damages using automated preventive measures,” Sasson explained. Speaking with CTech, he describes a use case in Istanbul, Turkey, which is currently planning for the inevitable hit of an earthquake that could affect its 20 million citizens. “When we are talking to the gas companies in Istanbul, all they want is three seconds to reduce the pressure in the gas pipes. The scenario they have in Istanbul is that in the case of a strong earthquake, something that IS going to happen in Istanbul… the whole city of Istanbul will catch fire. It will burn down because the whole city is sitting on a mesh of gas pipes.”

Each country prepares for earthquakes in its own way, making it difficult for countries to share their systems with one another. Sasson admits that the United States, which has 143 million Americans exposed to dangerous earthquakes, has “an excellent system” to help prevent damages. However, most earthquake early warning systems rely on decades of empirical data specific to the region. To simply move one system overseas would still take years for it to catch up with the data required.

SeismicAI avoids this problem by integrating different types of sensors in its physical systems to obtain reliable and precise results, usually operational in only 2-3 weeks. It then uses low-magnitude earthquakes, usually around 2.5, and takes their data to help tweak algorithms and predict when larger ones of more than five might hit. Today, the company partners with academic institutions and government bodies in Israel, Canada, Turkey, and India to implement its services and help companies or citizens get early warnings for when danger may strike.

Related Stories:

“If an earthquake is detected in the Himalayan region, it will give about 90 seconds or more of warning time to the population in New Dehli. If we’re talking about something closer, if it happens in the Dead Sea, it will give Tel Aviv 10 seconds.”

Of course, the warning times depend on how far the population rests next to the epicenter of any given earthquake. “If you are sitting on top of the epicenter of the earthquake, then there’s probably not much we can do. The only thing that our system sells is time.” Sasson highlights that in Los Angeles, 12,000 people could die if an earthquake were to hit at any time, simply because rescuers wouldn’t be able to access those trapped in elevators. If SeismicAI were to be deployed by government or privately-owned buildings, then it could use a 2-3 second warning to quickly and automatically stop at the nearest floor and open its doors for those inside.

An example of SeismicAI tracking Istanbul, Turkey. Photo: SeismicAI

SeismicAI has two deployment models: the first is a private network similar to the one used in Istanbul. The company has complete control over the network and its alert systems which can be deployed via SMS, mobile app, or any other method. This module allows SeismicAI to replicate its model and partner with different regions and local companies who help maintain the system as ‘feet on the ground’. The second model is seen in countries like India, which are operating via a grant from the Israel Innovation Authority and India’s local equivalent. In this case, the customer is the government buying the service and is therefore responsible for the alerts.

SeismicAI currently has 12 employees and is looking for funding so it can expand its operations in the private model leg of its business. While the company can enjoy the fruits of the Indian deal it says cost the government $160 million, it calls the private networks the main driver for growing its business. “The government sector has huge potential but things go slowly,” he concluded. “We expect revenues from the government sector only at the end of next year. But we expect revenues from the private sector using the private network as early as Q2 2022.”