Fusing NLP and Graph: Building a Conversational Agent for Enriched Financial Data

Session Track: Graph Memory & Agents

Session Time: 12:00 pm - 12:30 pm April 15

Session description

In this session, we share our experience of building an interactive, context-aware knowledge graph by combining structured corporate data with unstructured financial reports. We begin with relational corporate data, such as company hierarchies, and enrich it using a Natural Language Processing (NLP) pipeline that extracts novel entities and relationships from textual reports. The final stage introduces a conversational AI agent that uses this enriched graph as its semantic long-term memory, enabling intuitive exploration of complex financial data.

Through this presentation, we’ll discuss the challenges, insights, and outcomes from fusing structured and unstructured data in financial analysis, and how this approach can push the boundaries of financial intelligence.

Speakers

Alix de Cremoux

Machine Learning Engineer, Banque de France

Alix de Cremoux is a Machine Learning Engineer at Banque de France, where she develops secure and scalable AI solutions to support compliance, document processing, and digital transformation in the financial sector. She holds a Master of Science in Computer Science with a focus on Machine Learning from Georgia Institute of Technology. She has led projects involving the deployment of an intelligent sovereign assistant powered by Mistral LLM and the integration of AI solutions on Azure. Additionally, she represents Banque de France as an Employer Brand Ambassador, mentoring others and contributing to internal innovation through hackathons and workshops.

Gabriel Laffitte

Machine Learning Engineer, Banque de France

Gabriel Laffitte is a Machine Learning Engineer at Banque de France, developing AI-driven use cases for finance.