McKinsey argued that biopharma companies need “structural redesign” to get full value from AI in drug development, warning that incremental workflow efficiency won’t create “compound learning.” The consultancy’s report says AI can speed decisions, but programs often still run through linear stage gates without closed-loop feedback that drives systematic improvement. McKinsey proposed reorganizing around five connected decision points—from understanding patients and disease biology to improving the real-world impact of approved therapies. The analysis frames the closed-loop R&D model as the mechanism for continuous learning, not just automation of existing steps. The report contrasts traditional operating models with tech-driven approaches already used by some AI-first companies, including closed-loop platform descriptions from firms such as Recursion Pharmaceuticals.