The gap was there and Benny Bennet Jürgens took advantage of it. After ten years with the insurance company Generali, he left his stable job and founded the start-up Nect in 2017 with the business economist Carlo Ulbrich. Nect wants to fill the void that Jürgens has encountered on several occasions. During his last years in insurance, he worked on digital applications. They have made it much easier for customers to purchase insurance than before. But there was a problem. Insurers need to verify that clients are who they say they are. Official: An identity check is due. “Most insurers therefore send an activation letter,” Juergens explains. “But customers don’t want to wait three days for a letter.”
There are options like Video-Ident or Post-Ident, but they are expensive. With Video-Ident, employees must be available to verify via video call whether an identifier is real and valid and whether the face also matches the image. With Post-Ident, Swiss Post employees get the job done and charge insurers a fee. And sometimes clients have to go to an agency for this, again defeating the goal of getting a deal done quickly and easily. Indeed there is also the online function of the identity card, but it is complicated to use.
Jürgens’ idea, on the other hand, is to have the identity check carried out without human intervention. With devices that many people already have today: with smartphones. Nect calls it selfie-ident or robo-ident, but does that work too? How to prevent fraudsters from holding false documents in front of the camera? Or that someone is playing a video on the system?
Open detailed view
This is how the existing Video-Ident system works: the customer holds her ID card and face into the camera – employees then have to verify if the two are genuine and go together.
(Photo: Andrea Warnecke / dpa)
“We have taken the technology to such an extent that we can meet almost any compliance requirement,” says Jürgens. Almost? “Bureaucracy slows us down,” explains the founder. “We want the law on money laundering to react more quickly to innovations in the fight against fraud. Where this law applies, Nect cannot yet be used. At least Jürgens thinks the weakest point in bank control is the human factor. “I firmly believe that our technology cannot be corrupted.”
The city of Hamburg has already used the company’s technology for corona aid applications
The Nect system has already passed the acid test. When the city of Hamburg needed a solution in 2020 to verify the identity of people wishing to file Corona aid requests, it quickly became clear that it needed to be automated. “Hamburg has shown courage,” Jürgens says today, “there has not been a single case of fraud that we have not uncovered”.
The more Jürgens talks, the more he lists safety barriers that he and his now 84 employees have taken into account. “I can talk about it now,” said the founder calmly, “because we have patented and patented various technologies that are used in the Robo-Ident process”.
Image recognition algorithms play an important role in this regard. For example, they check how the security features of ID documents change when a customer films their ID. “The optically variable security elements are printed by different printing machines, so we check several elements,” explains Jürgens. A forger may have obtained one of the machines, but it is extremely unlikely that he will have more than one. Nect works with security authorities to keep abreast of counterfeits. The company is also on the warning list of the Federal Criminal Police Office, so it will be notified if any counterfeits are in circulation.
Customers have to film themselves with their smartphones and are invited to hold hands in a certain place, for example at the bottom left. This verifies that it is not a pre-recorded video. The system also recognizes so-called deepfakes. Deepfakes are videos in which, for example, one person’s face is superimposed on another using artificial intelligence. It feels deceptively real to human viewers when it’s done right. But machines are not so easily fooled. “You can’t make the little wrinkles on your lips, for example.” Deepfakes are pushed upwards, says Jürgens, “but until now it was still possible to detect manipulation”. He has yet to meet a deepfake. But he also admits: “The whole thing is a race of rabbits and hedgehogs.”
“98 to 99% of fraud cases are very easy to spot. “
Another safety barrier: While the candidates are recording videos, the smartphone emits ultra-high frequency sound which the mobile phone also records in the recording. Its exact frequency is however variable, so that it can be checked whether the pitch is correct. Other algorithms also check if the mobile phone moves according to the movements on the video. Otherwise, it would indicate an attempted video fraud. “It’s all in the combination,” Jürgens explains, “one algorithm alone does not create sufficient security, but several significantly reduce the likelihood of fraud.” All the sophisticated security measures are seldom necessary anyway. “98 to 99% of fraud cases are very easy to spot. For example, the check digit is not calculated correctly on identity documents. “Or the counterfeiters use the wrong font.”
It was therefore not easy to develop a system which eliminates the risks as much as possible. There were also technical hurdles, for example the large number of cameras used in Android phones. Or the different seasons with their different lighting conditions.
The biggest challenge, however, was another. “We already had experience and knew what we were doing,” explains Jürgens, “but it was still a long way to find the first customer. The problem is regulations. It is very complex. “It meant a lot of talk,” says Juergens, “we didn’t have a compliance department at the time either.”
The company lacks a qualified workforce: “This is already slowing down our growth”
Everything changed. “We are now growing very quickly,” says the founder, “it is getting harder and harder to find good people, and this is already holding back our growth.” Above all, the young company lacks specialists capable of operating its own data center. “We run our own hardware, so you need people who know what they’re doing.” Many, however, did not want to change jobs yet.
According to Juergens, Nect is growing on its own and the business is profitable. Over the past year, money has been raised from investors including TV well-known Carsten Maschmeyer, but this should only be used when it comes to expanding into other country.
Jürgens himself has experienced a rather unusual development. After graduating from high school, he began an apprenticeship as a computer scientist in an insurance company. “I fought up there,” he says, “I’m not a computer 1.0 student.” Most recently, he was mainly responsible for the implementation of the digital strategy. Computer science is also a lot of craft, he said, you need hands-on experience. Scientists with doctorates also work in his company. With automated image recognition, for example, it is very useful to have a scientific background.