Keeping BNP Paribas Cardif compliant
Faster, more accurate on-premises processing
With customers in 30 countries and products spanning automotive, telecoms, and finance, BNP Paribas Cardif must take a proactive approach to research and development – especially when it comes to AI use cases.
Challenge
Cardif's insurance workflows handle everything from ID cards to injury claim reports – vast numbers of varied, sensitive documents. It needed an on-premises solution to process these documents quickly and at scale to keep data safe, outpace competitors, and protect its brand reputation.
Solution
Using Domino as its in-house platform, Cardif built an intelligent document processing system that delivers:
Modular, future-proof architecture
The team can decouple IT services (like security and data storage) from AI infrastructure, letting them deploy new models without having to rewrite loads of code.
Compliant, on-premises processes
AI models can be packaged into portable formats like ONNX and run entirely on-premises, helping Cardif meet strict data-protection regulations across markets.
Improved model accuracy
Its generative AI models can scan documents without relying on Optical Character Recognition (OCR) or tedious, error-prone manual annotation.
The pipeline-first, on-premises approach helps speed up AI innovation, reduces operational costs by minimizing memory requirements, and improves model responsiveness with lower latency.
Results
Powered by Domino, Cardif’s in-house AI models now achieve 85% accuracy, significantly outperforming third-party tools. It can also keep critical customer data on-premises for added security, and rapidly roll out new AI capabilities to better serve its customers.
Generative AI models
with 85% extraction accuracy
On-premises deployments
enabling secure, compliant processes
New use cases built on
end-to-end generative pipelines
When we benchmark what we can do on-premises compared to external vendors, we do far better.
— Sébastien Conort, Chief Data Scientist & ML Engineering Director, BNP Paribas Cardif.
Source: Building future-proof in-house AI systems: An illustration with intelligent document processing session delivered at RevX London.
View more case studies
Saving Allstate countless hours
Allstate's team was losing precious time recreating models. With Domino, it can channel that time into creating value instead.
Empowering Direct Line's agents
Direct Line’s agents can focus on supporting customers when it matters most, using an AI model built on Domino.
Enabling governed AI innovation at USAA
To help it innovate safely, USAA created a solution to manage and purge experimental data sets.