Despite breathtaking technological advances, R&D productivity in life sciences has been paradoxically declining for decades—a phenomenon known as the Eroom’s Law. This chapter explores how generative AI offers a powerful new toolkit to potentially reverse this trend. We review areas of its application across the value chain, from using Retrieval-Augmented Generation (RAG) to synthesize scientific knowledge and generate novel hypotheses, to deploying foundation models that design new proteins and small molecules from scratch. The second half of this chapter pivots from the promise of the technology to the practical realities of its implementation. Drawing on direct, hard-won experience, we distill bitter lessons on why a brilliant model is not enough. We confront the real-world barriers of fragmented data, organizational friction, and the complex legal and regulatory landscapes. This chapter provides a pragmatic guide for leaders on how to build the interdisciplinary teams, robust data strategies, and governance frameworks required to translate algorithmic intelligence into tangible scientific and clinical impact.