Documents bedevil the lives of customers and the organizations that rely on them. However, in some vertical markets, the document is a necessary evil if accurate and compliant services are to be delivered. Insurance is just one example; the sector is document-rich or document-burdened, depending on your point of view. The complexity of the sector and the individual needs of customers make documenting fundamental. Surely, artificial intelligence (AI) is ripe for managing documents for the benefit of the insurance provider and the customer. The short answer is yes, as data leader Sébastien Conort, the chief data scientist of BNP Paribas Cardif, revealed.
With over 80 million customers in 30 nations and more than 500 distribution partners it is easy to see how BNP Paribas Cardif has a forest of documents to manage. The Paris headquartered business sells insurance as a white-label product through the automotive industry, banks, retailers, telcos and insurance brokers in Asia, Europe and Latin America. In 2023, BNP Paribas Cardif had written premiums to a value of €30.3 billion and has been actively growing as a business, acquiring 51% of Italian insurance firm BCC Vita last November and began talks to acquire Neuflize OBC from ABN Amro in May.
For financial services companies, documents form the core of their relationship with the customer, especially in insurance, as Conort explains:
In the insurance business, we have two important customer touchpoints: the sale and when they have to open a claim. These are infrequent.
Yet, across a business of the scale of BNP Paribas Cardif, those infrequent interactions mount up, and the chief data scientist says the business processes 100 million documents a year. Conort adds:
Being excellent at document processing is a differentiator; it impacts the customer and lowers the operational cost of the business.
Coping with high demand and seeking to improve customer services led BNP Paribas Cardif to realize the documents and the processes around those documents were a core business value. Automation would enable the business to derive increased value, such as data, from its document processing, but that automation had to be cost-effective. Since 2021, BNP Paribas Cardif has operated Intelligent Document Processing (IDP), an automated document processing platform that uses a raft of technologies, including AI. Conort says of the platform:
IDP augments our collaboration with operational departments, and it relieves them from tasks that are not added value, so they can concentrate on having more interactions with customers.
Using optical recognition, natural language processing, information extraction, data matching and a data quality assessment set of technologies, IDP reduces manual tasks for teams that are either dealing with a claim, managing relations with third-party retailers or selling insurance to a customer. Conort says:
IDP means we have faster processing for our customers, as we are using AI to make faster decisions.
The addition of AI modelling has enabled BNP Paribas Cardif to model its claims and discover which types of claims require less documentation, making the claim process less stressful for the customer and faster for the business. AI is not the only way the insurer has reduced its reliance on documentation, it has assessed when its administration partners offer documents via application programming interface (API) systems, this again reduces the need to contact the customer and ask for further documentation.
However, he adds:
We never refuse a claim based on AI. Claims that are not automatically accepted by AI go through usual human review.
Embedding AI into IDP has given the business confidence to increase its experimentation with AI, Conort says:
This is one of the first use cases of AI for us, and it proves that it brings value, and we will try to think of areas that AI can bring more value and improve services.
BNP Paribas Cardif has publicly stated its AI strategy is to turn customer data into value for the customer through three pillars: data quality, data protection and data exploitation, and that the Corporate Analytics team that Conort is part of is the center of building, operating and reusing algorithms.
Corporate Analytics is involved in the Scikit-Learn open source skills community, and works closely with supplier Domino as the international platform for what the business calls industrial data science management. These are central to that AI strategy.
As machine learning engineering director and chief data scientist, Conort worked closely with IT to develop IDP, with the development being run as a joint venture between his analytics division and IT. A key technology partner has been machine learning operations provider Domino. In the BNP Paribas Cardif data science department 100 team members use the Domino Enterprise AI Platform to share their work, which reduces duplication and creates a better community within data science.
Two years in, and IDP is delivering business benefits, Conort says:
In Spain, we have managed to reduce the claims process from seven days to just one, and in Chile, we have reduced the number of claims documents from six to three.
In all the AI hype, use cases are going to become increasingly important to digital leaders. Incorporating AI into a wider automation programme, with clear benefits to the customer and, therefore, the business, is exactly the type of step change that AI can and will deliver to organisations. BNP Paribas Cardif has clearly understood customer and business challenges, as well as concerns about AI and deployed AI in a way that cuts manual processes. What will be interesting to see over the longer term is how this usage of AI permeates across the business, as well as how the technology impacts the energy usage of a major international financial services firm.
