Post by Dylogy

Configuring an AI platform for every client's specific context manually doesn't scale. So we built a way to automate it. In this third episode of our Behind the Scenes series, Christelle, ML Engineer at Dylogy, walks through how her team approaches one of the less visible but most consequential challenges in deploying AI for insurance: making powerful technology genuinely usable at scale. She talks about how Dylogy's ML team combines traditional algorithms, machine learning, and generative AI. Not out of technical habit, but to find the most efficient and cost-conscious solution for each problem. And she shares the thinking behind "env generation", a feature that automatically creates platform configurations directly from client documents, removing a step that used to demand significant manual effort. The technical detail is worth pausing on: old-school string distance metrics for deduplication, embedding models for semantic proximity, reranking models for relevance, each chosen deliberately, with a constant eye on cost, latency, and energy consumption. Small team, high standards, and a clear sense of where the real bottlenecks are. That's what delivery looks like at Dylogy. ⬇️ #AIinInsurance #GenAI #MachineLearning #LLM #Insurtech #BehindTheScenes

Video Content