On April 7, Tsinghua University's Institute for Intelligent Industry (AIR) and Shuimu Molecular jointly released OpenBioMed Skills. This initiative involves encoding the entire decision-making process of biomedical experts into executable code as Agent Skills, with the related code made open-source on GitHub.

The initial release of OpenBioMed Skills includes 45 meticulously developed professional capabilities. These enable researchers, even without deep engineering backgrounds, to build and run end-to-end workflows covering the entire biomedical research and development process.

The first set of 45 selected skills is divided into five major categories, encompassing biochemistry/drug R&D, protein analysis and design, and single-cell omics analysis. It also provides auxiliary skills such as data retrieval and tool invocation.

To allow R&D personnel and industry scholars to use this skill set more easily, OpenBioMed Skills has been deployed on the OpenBioMed platform. This enables direct calls to multimodal models like PharMolixFM, BioMedGPT-R1, and MutaPLM. Compared to using general models like ChatGPT, these provide superior understanding of professional knowledge. Furthermore, experts can remotely direct their biomedical Agents through platforms such as Feishu, DingTalk, and WeChat.

Nie Zaiqing, Professor at Tsinghua University, Chief Researcher at the Institute for Intelligent Industry (AIR), and Chief Scientist at Shuimu Molecular, stated: "The deep integration of biomedicine and artificial intelligence is fundamentally reshaping the underlying logic and industrial paradigms of drug R&D. From proposing the fourth-generation drug R&D paradigm, to creating the OpenBioMed open-source development platform, and now to launching the BioMed Skills professional AI skill set for pharmaceutical research, our core mission remains to lower the development barriers in life sciences and biomedicine and to empower scientific innovation with large models. We are systematically transforming expert knowledge and decision models into reusable intelligent workflows, making it a reality to invoke drug R&D processes using natural language. This not only allows biomedical researchers to build their own specialized agents more efficiently but also promotes a shift in the pharmaceutical R&D field from 'reliance on expert experience' to 'intelligent collaborative innovation.'"