The Assurtech Zelros has developed a computer vision solution that allows documents to be recognized and analyzed.. The tool en allows the managers of the insurance broker AssurOne to save time on the processing of gray cards and could soon be used on other official documents.
In 2017, insurance broker AssurOne is looking for a solution that can authenticate on a large scale the thousands of documents received by its clients. But he can’t find what he’s looking for, the publishers he meets doing only classic OCR [Optical Character Recognition]. The technology is not considered satisfactory and the project is suspended. He was reinstated two years later.
At the time, Zelros was working on a RAD-LAD solution – document recognition and automatic reading – that is not yet deployed to its insurance clients. Created in 2016, this start-up relies on computer vision technologies to support employees of insurance companies in two areas: sales assistance and claims management. 3 MINUTES TO 10 SECONDS
It is on the first point that it proposes a solution coconçed with the teams of AssurOne, and this in less than two weeks. The principle: a solution that automates the authentication of grey cards. “This step used to be done manually,”says Pierre Barochez, CTO of AssurOne. A verification process that can take up to 3 minutes when done by hand, while several hundred grey cards are sent each month by customers.
The solution, called Zelros for Customer Service and offered in SaaS, hijacks the white areas, authenticates the document and extracts the relevant data to validate the subscription process in less than 10 seconds.
It uses computer vision and machine learning techniques to first recognize the type of document and then read its contents. “We individualize documents and then transform them into a language recognizable by the machine,” explains Paul-Henri Chabrol, “Head of Customer success” at Zelros.
20,000 DOCUMENTS PROCESSED SINCE 2019
In the event of a problem, such as a poorly scanned document, the invalid field is highlighted and the manager can quickly regain control. AssuOne therefore verifies internally the accuracy of the information transmitted by Zelros’ software by comparing it to those on the customer’s contract. In the event of an error, the manager intervenes knowing exactly where the inconsistency is. 20,000 documents have been processed since last year, and 25% of grey card flows are already validated “end-to-end” by the Zelros system, i.e. with all fields validated without human intervention.
The first benefit is to speed up the insurance underwriting process. “These, Pierre Barochez are repetitive tasks, with no added value for our employees, but they are nevertheless essential for customers who want to be covered,” recalls Pierre de Barochez. But it’s also rewarding for teams because the management rules have been set with managers.”
NEXT STEP, THE DRIVER’S LICENSE
In the medium term, the two partners plan to significantly increase the rate of grey cards validated end-to-end by the system and, on the other hand, to extend automatic recognition and reading to driver’s licenses. The first results obtained with the new version of this official document “are promising since they have a data extraction rate of 95%”,the two French companies welcome. Zelros, which develops features specific to the world of insurance, wants to participate in the digitalization of the sector by offering insurance policyholders the experiences offered in retail or tourism, which digital has greatly contributed to improve. The idea is to speed up the subscription process, with the customer’s only obligation to send the supporting documents and validate the information on the subscription form, without having to fill it out yourself.