TLDR : Dataiku and NVIDIA have launched FSI Blueprint, a new AI architecture for banks and insurance companies, facilitating the deployment of AI agents with built-in governance and compliance features.
This Wednesday, June 11, at the NVIDIA GTC Paris at Viva Technology, Dataiku unveiled FSI Blueprint, a new solution designed for banks and insurance companies. Co-developed with NVIDIA, this enterprise architecture aims to facilitate the deployment of AI agents while integrating governance, compliance, and performance imperatives specific to financial services from the outset.
Conceived as a response to the dispersion of AI initiatives in financial institutions, FSI Blueprint aims to transition organizations from the experimental stage to fully industrialized AI. This unified and secure framework enables the creation, testing, deployment, and supervision of autonomous agents, ensuring their compatibility with regulated environments.
A Technological Foundation for Industrializing Agent-Based AI
The offering relies on the integration of the Dataiku platform with the NVIDIA Enterprise AI Factory, combined with components such as NVIDIA NIM, NeMo, and Dataiku's LLM Mesh. The latter allows for fine orchestration of language models, whether open source, proprietary, or specialized, in a controlled environment.
Thanks to this technological framework, these institutions will be able to:
- Automate manual tasks at scale, by deploying agents capable of replicating complex workflows;
- Enhance fraud detection and anti-money laundering (AML) efforts, through models calibrated on trusted internal data;
- Improve customer experience, by deploying adaptive and contextual virtual assistants;
- Optimize risk analysis, through more refined modeling of operational or financial scenarios;
- Ensure centralized supervision, thanks to integrated safeguards like Cost Guard and Quality Guard, which allow for controlling usage, costs, and the quality of results.
Sector-Specific "Blueprints" to Structure AI in Regulated Industries