This project aims to:

• reduce entry errors in respect of data about people and risk descriptions
• reduce the time spent checking documents collected and entering data into the information system

In the short term, this project will make it possible to automate checks on the conformity of collected documents, and in the medium term, to make data entry more reliable and automated, generate productivity gains and help to modernise our image.

This in-house Groupama solution is based on the provision of automated document recognition and reading services that can be used in different business use cases. It is underpinned by new technologies and innovative methodologies leveraging artificial intelligence.

There are three steps to the process:

1. The collection of supporting documentation, which can occur over various channels: in-branch scanning using a scanner, uploading documents, photos on a mobile phone, or retrieving documents already in the EDM.
2. The use of document-recognition services (e.g. driving licence, vehicle registration document, identity card) and then OCR to read the necessary data on the document (e.g. surname, first name, date of birth).
3. Data verification and therefore automatic validation of supporting documentation in comparison with expectations.

The innovative dimension lies in the ability to recognise and read complex documents in a vast range of formats, often as photos on a mobile phone, and sometimes of very poor quality.

Note that the DALI project was based on a solution called “Document Teller”, developed in-house by the Perspectives and Digital Acceleration team, as conventional solutions on the market that usually process incoming mail were unable to meet business-line objectives.