A structural engineer once noticed that his used car carried dozens of sensors, and the earthquake-exposed building he worked in carried zero. Andy Thompson decided to fix the building.
Andy Thompson runs Safehub, a company whose entire premise fits on a Post-it: put a small, cheap sensor on a building, and let the building tell you - in real time, in the seconds and minutes after an earthquake - whether it is fine or whether it is in trouble. No guessing. No waiting for an inspector to climb the stairs three days later. The structure reports for itself.
That sounds obvious right up until you realize almost nobody does it. Most of the world's earthquake-exposed buildings are silent. When the ground moves, the people who own, insure, occupy, or depend on those buildings are left squinting at regional maps and rough models, trying to infer what happened to a specific address from data measured kilometers away. Thompson's bet is that a building should not be a black box. As cofounder and CEO of Safehub, he has spent a decade turning that conviction into hardware, software, and - more recently - insurance contracts that pay out based on what the sensors actually felt.
Today Safehub's sensors sit in thousands of buildings across dozens of countries. The customer list includes the kind of names that do not buy unproven technology lightly: Amazon, FedEx, and other Fortune 500 companies. The product has grown from "is my building damaged?" into a platform called ShakeNet that blends Safehub's own dense sensor network with government seismic feeds to produce some of the highest-resolution shake maps available anywhere. And in 2025, that platform became the trigger for a new generation of earthquake insurance.
Through our existing sensor networks, and our ability to quickly deploy into developing markets, we are able to provide the highest-resolution regional shaking maps anywhere in the world. - Andy Thompson, on the ShakeNet partnership with Liberty Mutual Re
The slow road to an obvious idea
Thompson did not arrive at sensors as a gadget person. He arrived as an engineer who had seen too many disasters up close. He studied structural and earthquake engineering at UC Berkeley, then spent 16 years at Arup, the global engineering consultancy, where he built the firm's Catastrophe Risk & Insurance practice from the ground up. The job put him in front of corporations and governments that needed to understand catastrophic risk - how to mitigate it, how to respond when crisis hit, how to keep operating, how to insure what was left.
It also put him in the field. Thompson did post-disaster work in Chile, Haiti, Japan, India, the United States, and elsewhere. Walking through cities after the shaking stops teaches a particular lesson: the gap between "an earthquake happened" and "here is exactly what it did to this building" is enormous, expensive, and dangerous. Decisions that should take minutes - evacuate or re-occupy, pay the claim or dispute it, send crews here or there - instead take days, because the information simply is not there.
Two events in 2001 sharpened his focus more than any seminar could. The Bhuj earthquake in India and the September 11 attacks, unrelated in every way except their scale of loss, convinced Thompson that engineering was not just about making structures stand. It was about saving lives and making whole societies more resilient. That idea - resilience as the goal, not just safety - became the through-line of everything he did next.
A cofounder he'd known for nine years
Thompson did not start Safehub with a stranger from a startup mixer. He cofounded it in 2015 with Doug Frazier, the former Chairman and CEO of EQE International and EQECAT - at one point the largest catastrophe risk management company in the world. The two had been working together since 2006, when Frazier consulted at Arup. By the time they launched, the partnership was less a leap of faith than the natural next step for two people who had spent careers staring at the same problem from slightly different angles: one from engineering, one from risk and insurance.
How the sensor actually thinks
The technology is elegant in the way good engineering usually is. Every building has a natural frequency - a rhythm at which it likes to sway. Safehub's sensors listen to that frequency continuously, building a vulnerability profile for each individual structure and establishing a baseline for what "normal" sounds like. When an earthquake hits, the sensors compare the shaking against that baseline. A shift in frequency can reveal damage that is invisible from the street, buried in places an inspector cannot easily reach.
This matters because earthquakes are not democratic. The same quake can leave one building untouched and crack the one next door, depending on age, design, soil, and a dozen other variables. Regional shake data treats a neighborhood as a single dot. A sensor treats each building as itself. That difference - from neighborhood to address - is the whole game, and it is what lets Safehub tell a property owner not "there was an earthquake near you" but "your building is fine" or "your building needs to be looked at, now."
Listen to advice; but listen to your own instincts more. - Andy Thompson, on building a company
From data feed to insurance trigger
The most interesting turn in Thompson's story is what happened when the sensor data met the insurance industry. Traditional parametric insurance pays out automatically when an event crosses a threshold - say, a quake of a certain magnitude within a certain radius. It is fast, but it suffers from basis risk: the gap between what the model says happened and what actually happened to your specific building. You can get paid when you were fine, or get nothing when you were wrecked.
Safehub's sensors collapse that gap. Because the trigger can be tied to the shaking measured at the policyholder's actual location, the payout tracks reality far more closely. In 2024, Safehub and Liberty Mutual Reinsurance announced an exclusive partnership to bring a building-specific, sensor-based parametric earthquake product to global clients. In August 2025, they launched ShakeNet Parametric, designed explicitly to reduce basis risk using the highest-resolution regional shaking data Safehub could produce. It was the moment Thompson's engineering instinct and Frazier's insurance pedigree finally clicked into a single product.
The drummer who reads skylines
For all the catastrophe modeling, Thompson is not a grim person. He worked his way through college as a drummer in a band, and he still plays. There is a certain logic to an engineer who spends his days thinking about how structures resonate also being someone who keeps time for a living - frequency, rhythm, and resonance are the same vocabulary in two different rooms.
He is also, by his own account, the kind of person who cannot turn the engineering brain off. Friends and colleagues describe a man who reflexively analyzes how buildings respond to extreme loads while simply walking or driving through a city. Where most people see a skyline, Thompson sees a set of structures quietly negotiating with gravity and the next big shake. It is the occupational hazard of someone who has made buildings his life's work, and it is probably why the sensor idea felt less like an invention and more like an obligation.
That mindset shows up in his other work, too. He coauthored a book called Peace of Mind in Earthquake Country, and he serves on the governing board of the Global Earthquake Model, a nonprofit effort to make earthquake risk understandable and shareable across the planet. These are not side hobbies. They are the same mission as Safehub, expressed in print and in policy instead of in hardware: get good information to the people who need it before the ground starts moving, not after.
What he's really building
Strip away the funding rounds and the Fortune 500 logos and Thompson's project is simple to state and hard to do: give every earthquake-exposed building a nervous system. A way to feel, to remember its baseline, and to speak up the instant something changes. Multiply that across thousands of structures and dozens of countries and you get something larger than any single building - a living map of how the world shakes, building by building, updated in real time.
His stated aim is to increase society's resilience to natural disasters. That is a big sentence, and the kind of thing many founders say. The difference with Thompson is that he has spent 26 years in the specifics - in the rubble of real disasters, in the risk models, in the insurance fine print, and now in the sensors themselves. He is not trying to predict the next earthquake. He is trying to make sure that when it comes, nobody has to guess what it did.