The article "Bedside to Bench — AI and the New Science of Medicine" describes the use of artificial intelligence (AI) in medicine, which is changing from a tool to help doctors to an active partner in their clinical care. Stanford research introduced the MedAgentBench platform, which tests AI agents' abilities to perform everyday clinical tasks in real-world electronic health record settings. Newer versions of large language models (LLM) show significant improvement in the success of these tasks, suggesting that AI may soon take over basic administrative and organizational tasks in healthcare. The goal is to reduce the workload of doctors and nurses and thereby alleviate burnout in the medical staff. At the same time, the OpenEvidence platform, trained on 30 years of New England Journal of Medicine archives, provides physicians with up-to-date and validated knowledge to support clinical decision-making right at the patient's bedside. This AI platform is already used by 75% of doctors in the US during clinical practice and is free for verified doctors. The developers plan to expand the team of experts and create specialized models for various medical fields, such as oncology, neuroscience and cardiology. The rapid increase in medical knowledge, doubling approximately every 73 days, highlights the need for such AI tools to keep medical information current.