A question often asked in the field of tech is: Who takes the blame when AI fails? And while this might not be completely clear in sectors like autonomous driving, an Israeli startup seems to have cracked the code for the liability question in AI and infrastructure.
When people go out onto the street, they expect roads, bridges, and infrastructure in general to work, but in reality, they need constant maintenance to keep working. With this in mind, the Israeli startup Dynamic Infrastructure created a platform that uses artificial intelligence to speed up and improve the accuracy of this process.
“Our AI platform helps civil engineers process great amounts of data in order to know which infrastructure needs to be prioritized,” Saar Dickman, CEO of Dynamic Infrastructure, told The Jerusalem Post.
Dickson addressed the liability topic by explaining that AI is not yet developed enough to be used in a fully autonomous way. While it’s impossible to know what will happen in the future, the current approach to addressing liability is to add “human checkpoints” in the system.
“The information processed is there. Someone certified, an inspection engineer, or a certified contractor collected the information, and someone paid him for this information. So, from this moment on, the infrastructure owner is liable because he paid for the inspection service. He's supposed to know what's happening once the inspector provides the results. The platform only helps in going through that information,” Dickson explained.
“Additionally, our system is not completely run by AI. There are points in the analysis where civil engineers come in and revise the work, adding an extra layer of reassurance to the final result,” he added.
He also explained that the company doesn’t aim to replace civil engineers, but rather to become a helping hand when processing massive amounts of data. “I don't know what will happen in 10 years, but right now, artificial intelligence cannot fully replace engineers just support them,” he pointed out.
How does the virtual engineer work?
The company, founded in 2019 by Dickman and Amichai Cohen, has recently announced that Arkansas, United States, will be the latest state to adopt its technology for infrastructure analysis, with governments from 13 US states already using the platform.
The company reported a 100% contract renewal in 2025, and Dickson told the Post about its plans to expand to the Australian and European markets, with the objective being “to provide every public or county engineering and maintenance department with an AI-based ‘virtual engineer’ that works alongside professional teams and delivers unprecedented force multiplication.”
“In a world where infrastructure is aging, and budgets are limited, our system enables authorities and state transportation agencies to gain clear visibility into the condition of their assets and manage them efficiently and proactively,” he said.
The company reported $1 million in revenue in 2025, with projections indicating the capacity to triple it in the coming year.
From not recognizing people to pointing out individual bricks
Dickson also said that one of the challenges in developing the system was explaining the difference between modern structures (which have 30 to 40 years of data from their construction to the present) and antique structures hundreds of years old.
“It took a while, but once you train the system, and then you use our unique IP, it goes very smoothly. We had to train the difference between a brick falling from a modern brick and one falling from a medieval arch built 400 years ago,” he explained.
But the advantage they had was that they developed the system with a team of civil engineers, not just programmers. This way, they knew what the AI needed to learn, where to find mistakes, and how to correct them.
Dickson gave an example of how this training journey was: “When we began, the system was in early development, and we used a photo from one of our clients based in Greece that featured a red-haired woman standing on a bridge. And the system, again, many years back, identified the red-haired lady as a rust on the bridge. You don’t see that today, but it still remains both a learning experience and a funny story.”