Researchers have introduced TxBench-PP, a benchmarking tool for evaluating AI agent performance in small-molecule preclinical pharmacology, a crucial step in drug discovery1. This development is significant as it enables the assessment of AI agents' decision-making capabilities in a realistic and verifiable manner. TxBench-PP is part of a larger initiative, TherapeuticsBench, aimed at creating a comprehensive benchmark for therapeutics. The introduction of TxBench-PP marks a crucial step towards the practical deployment of AI in drug discovery, as it allows for the evaluation of AI agents' performance in a real-world setting. This has implications for the pharmaceutical industry, as trusted evaluation of AI agents is essential for their adoption. The development of TxBench-PP also highlights the need for careful consideration of the broader implications of AI in drug discovery, including policy, security, and workforce dynamics. This matters to practitioners as it enables them to critically evaluate the performance of AI agents in preclinical pharmacology.