Large‑language models (LLMs) are now being evaluated at every rung of the drug‑development ladder—from candidate‑selection brainstorming to post‑marketing signal detection. The focus of this article is on productivity accelerators rather than high‑stakes, fully autonomous decision making. These are