Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Learn how to implement live hallucination detection in production generative AI applications using LaunchDarkly's AI Configs in this 51-minute conference talk from AWS re:Invent 2025. Explore the critical importance of accuracy when deploying generative AI applications to production environments and discover how LaunchDarkly's AI Configs provide the flexibility to modify your generative AI applications at runtime for more effective model performance measurement in real-world settings. Follow along with a practical demonstration featuring a simplified RAG (Retrieval-Augmented Generation) implementation of an insurance chatbot built using an Amazon Bedrock knowledge base. Watch as the chatbot utilizes LaunchDarkly AI Evals to detect hallucinations in production environments, and gain insights into collecting essential metrics such as latency and cost. Understand strategies for building guardrail logic directly into your runtime configuration to ensure reliable and accurate AI application performance. This session, presented by AWS Partner LaunchDarkly, provides hands-on guidance for implementing robust monitoring and evaluation systems for production generative AI applications.
