Generative AI is generating plenty of buzz, but there is often a gap between the hype and real-world implementation. A recent conference, hosted by Dylogy, set out to answer not "should we go there?" but "how to actually make it work." Three experts in the insurance sector shared their concrete experiences moving from experimentation to deployment.
Causal Interpretation of LLMs for Risk Management
Abdallah Arioua, Chief Data Officer at Relyens, explained how going beyond correlation to understand causation changes risk management. By applying causal reasoning to large language models, insurers can better identify root causes of risks rather than just patterns.
A Conversational Agent for Internal Knowledge
Olivier Claeys, Head of Digital Factory at GG Vie Individuelles, detailed the deployment of an AI agent in production. He covered technical choices, obstacles encountered, and results obtained from building a system that helps employees master internal knowledge.
Managing a Portfolio of GenAI Use Cases
Waswate Ayana, Lead Innovation at Convex Insurance, discussed how to structure, prioritize, and steer generative AI projects to move fast without spreading too thin. Her approach helps insurers select the right use cases and manage them efficiently.
The conference included a Q&A session and covered chapters on what GenAI means in insurance, causal interpretation, deploying a conversational agent, and defining AI projects.