• BNP Paribas is now deploying an internal LLM as a Service platform, designed to provide the Group’s entities with unified and secure access to large-scale language models (Large Language Model – LLM).
• This shared infrastructure, operated by the Group’s IT teams, is set to accelerate the development of generative artificial intelligence (AI) use cases to better serve employees and clients.
• This initiative is part of the Group’s technology strategy, which leverages data and AI as key drivers to enhance customer personalization and operational performance.

To facilitate the development and deployment of generative AI use cases in its businesses and functions, BNP Paribas has designed and deployed an internal LLM as a Service platform.

Operated by the Group’s IT teams, this technology, allows each entity to access a selection of language models (LLM) tailored to its needs via a standardized and secure interface. These models can be directly integrated into tools or processes.

The platform is based on a shared infrastructure hosted in the Group’s data centers, equipped with specialized computing capabilities (GPU). It provides access to a selection of open-source models, models from Mistral AI, a partner of the Group, and soon models trained on internal datasets, according to specific needs.

This architecture allows the Group to have high processing power to efficiently execute language models, while sharing and limiting costs. It ensures secure deployment and enhances the coherence of generative AI usage across businesses and functions.

« With LLM as a Service, we are creating a common technological foundation that allows entities to focus on business use cases, while relying on a shared and secure infrastructure. It is a common lever to innovate and improve the performance of certain applications. » Marc Camus, Group Chief Information Officer, BNP Paribas

Several generative AI use cases are already in production or experimentation within the Group’s businesses, such as internal assistants, document generation, and information retrieval in documents.

The implementation of the platform allows for scaling up by facilitating the deployment of these use cases within a common framework and accelerating the implementation of new projects. The platform enhances the businesses’ ability to quickly integrate generative AI into their existing processes, with shared technical and security standards.

LLM as a Service was deployed in an initial test phase across various entities of the Group, such as Hello bank! and the Group’s general inspection, and is intended to be progressively deployed across the Group according to an industrialization process. Additionally, generative AI already enables developers to create code as well as easily and centrally access development tools, AI technologies, and DevOps chains.

LLM as a Service builds on the feedback from the first user entities and complements the initiatives already undertaken around generative AI within BNP Paribas. This gradual deployment allows for the consideration of local specificities while ensuring a high level of security, compliance, and operational efficiency.

Source: BNP Paribas