The next evolution of AI in biopharma is already here - systems designed not just to analyze data but to make decisions. These “agentic AI” models can set goals, adapt strategies, and generate hypotheses with a degree of autonomy that could reshape discovery and development. What is at stake is more than efficiency - it is a shift in perception. AI is no longer only a supportive tool but is beginning to be viewed as a potential teammate in scientific problem solving. The central question is whether these systems can be trusted to shape strategy. This panel will examine what standards of evidence are required before industry relies on autonomous input in hypothesis generation and experimental design, what safeguards are needed to ensure reliability and transparency, and how much human oversight will remain essential. Just as importantly, it will address how organizations will need to adapt their structures and mindsets if AI is to be treated as a collaborator rather than a calculator. Attendees will leave with a clearer understanding of what distinguishes agentic AI from today’s tools, the hurdles that must be overcome before adoption, and the conditions under which autonomy could credibly influence discovery.