This article discusses the potential of AI agents in biomedical research, emphasizing their ability to collaborate with humans, integrate AI models and tools, and perform tasks that enhance scientific discovery. AI agents are designed to combine human creativity with AI's capacity to analyze data, navigate hypothesis spaces, and execute repetitive tasks. They can be used in various areas, including virtual cell simulation, programmable control of phenotypes, and the design of cellular circuits to develop new therapies. The article outlines the development of AI agents as systems that can learn and reason, using large language models (LLMs) and generative models to create structured memory for continual learning. These agents can also incorporate scientific knowledge, biological principles, and theories using machine learning tools. The article highlights the importance of ethical considerations, such as ensuring safety and preventing harm when AI agents make changes in environments. The article also discusses the evolving use of data-driven models in biomedical research, including databases, search engines, machine learning models, and interactive learning models. These models have advanced the modeling of proteins, genes, phenotypes, clinical outcomes, and chemical compounds through the mining of biomedical data. The article describes different types of biomedical AI agents, including LLM-based agents and multi-agent systems. It outlines the levels of autonomy in AI agents, ranging from Level 0, where ML models are used as tools, to Level 3, where agents can develop and extrapolate hypotheses beyond the scope of prior research. The article also discusses the challenges and limitations of biomedical AI agents, including the need for responsible implementation and the potential risks of overreliance on agents. The article provides examples of AI agents in genetics, cell biology, and chemical biology, illustrating how they can assist in research tasks, such as analyzing molecular interactions, designing new drugs, and providing chemical probes for biological systems. It also outlines a roadmap for building AI agents, emphasizing the importance of modules such as perception, interaction, memory, and reasoning. The article concludes with a discussion on the future of AI agents in biomedical research, highlighting their potential to revolutionize scientific discovery and the need for responsible development and implementation.This article discusses the potential of AI agents in biomedical research, emphasizing their ability to collaborate with humans, integrate AI models and tools, and perform tasks that enhance scientific discovery. AI agents are designed to combine human creativity with AI's capacity to analyze data, navigate hypothesis spaces, and execute repetitive tasks. They can be used in various areas, including virtual cell simulation, programmable control of phenotypes, and the design of cellular circuits to develop new therapies. The article outlines the development of AI agents as systems that can learn and reason, using large language models (LLMs) and generative models to create structured memory for continual learning. These agents can also incorporate scientific knowledge, biological principles, and theories using machine learning tools. The article highlights the importance of ethical considerations, such as ensuring safety and preventing harm when AI agents make changes in environments. The article also discusses the evolving use of data-driven models in biomedical research, including databases, search engines, machine learning models, and interactive learning models. These models have advanced the modeling of proteins, genes, phenotypes, clinical outcomes, and chemical compounds through the mining of biomedical data. The article describes different types of biomedical AI agents, including LLM-based agents and multi-agent systems. It outlines the levels of autonomy in AI agents, ranging from Level 0, where ML models are used as tools, to Level 3, where agents can develop and extrapolate hypotheses beyond the scope of prior research. The article also discusses the challenges and limitations of biomedical AI agents, including the need for responsible implementation and the potential risks of overreliance on agents. The article provides examples of AI agents in genetics, cell biology, and chemical biology, illustrating how they can assist in research tasks, such as analyzing molecular interactions, designing new drugs, and providing chemical probes for biological systems. It also outlines a roadmap for building AI agents, emphasizing the importance of modules such as perception, interaction, memory, and reasoning. The article concludes with a discussion on the future of AI agents in biomedical research, highlighting their potential to revolutionize scientific discovery and the need for responsible development and implementation.