McKinsey argued that biopharma companies won’t fully capture AI’s value unless they reorganize R&D around a closed-loop operating model that compounds learning. In a new report, the consultancy said current AI deployments may improve individual steps but often fail to create systematic feedback across stage-gate decisions. McKinsey proposed five connected decision points—from understanding patient and disease biology to improving real-world impact of approved therapies—so outputs from pivotal decisions can refine subsequent choices. The report highlights that techbio firms already pursue feedback-driven workflows, and it cites multiagent systems and AI research tools as examples of iterative hypothesis-testing loops. The takeaway for industry leaders is structural: scaling AI in drug development may require changes to governance, data capture, and decision architectures, not just tooling upgrades.