[Korean Medicine AI Integration Task Force Meeting] From Silicon Valley to Biotech Valley - How Agentic AI is Reshaping Drug Discovery

Abstract

Agentic AI is rapidly moving from proof-of-concept to practice, redefining how we discover, design, and validate medicines. This talk traces that shift “from Silicon Valley to Biotech Valley,” connecting foundation-model breakthroughs (transformers, multimodal LLMs) and agent architectures to concrete gains in hypothesis generation, experiment planning, and iterative decision-making. We’ll situate today’s momentum in the broader arc of AI—rule-based systems to deep learning to diffusion and agents—and frame why the current wave is different in speed, scale, and enterprise adoption.

We then zoom into biomedicine: how AlphaFold-era structure prediction, physics-aware modeling, and data integration across genomics, proteomics, imaging, and clinical streams shorten the path from target to lead. Case patterns will illustrate where agents add leverage—automating literature triage, orchestrating wet-lab workflows, and closing the loop between in-silico and in-vitro to increase the yield of promising candidates while reducing cycle time and cost.

Finally, we discuss building blocks for deployment—data quality and bias, validation and safety, regulatory readiness, and the compute and workflow stacks that make agentic systems reliable. Attendees will leave with a pragmatic blueprint: what to adopt now, what to prototype next, and how to architect teams and platforms so AI agents become trustworthy collaborators in drug discovery rather than standalone demos.