Keynote Speaker: Jingyi

Founder and CEO of Huabai Jian

To do a good job, one must first sharpen his tools; to sharpen the tools of AI, one must first improve its "prompts". The practical camp deeply reveals the core methodology for efficient collaboration with large models such as DeepSeek and Doubao: role setting, task decomposition, format constraints, and iterative feedback.

Through the analysis of real-life cases and the extraction of core principles, students quickly grasp the skills of efficient dialogue with AI. With tools such as DeepSeek and Kimi intelligent assistants for instant practice, they witness the instant response and implementation of AI effects, ranging from simple instructions to complex requirements.

Throughout the day, immersive practical operations were carried out centered around four core scenarios: "research and development, decision-making, vision, and knowledge management". Students brought their own real data, and multiple intelligent tools worked together to make "AI writing, calculating, and managing on behalf of humans" a reality.

▪ AI R&D Express Workshop: Utilizing proprietary research topics, we operate tools such as Kimi and academic search tool AMiner to instantly filter out key papers from a vast array of literature, automatically generating abstracts and opinion summaries. We also employ BioGDP for biological visualization, and even perform rapid information association and visual analysis for specific targets.

▪ AI Pharmaceutical Decision-Making Workshop: Based on market and project requirements, we will demonstrate on-site how to instruct AI to quickly generate a framework for competitive product analysis reports, use Table Answers or ChatExcel to automate data processing, and generate project valuation models and sensitivity analyses in seconds after entering key data. The preliminary work for decision support can be significantly compressed.

▪ AI Vision and Design Workshop: Utilize Draw.io to create flowcharts, leverage Mitai AI or Tongyi to transform complex data into publication-quality charts with a single click, and quickly draft academic-standard presentation slides with the help of Madangxiu PPT. The efficiency of visual presentation is significantly enhanced, addressing the "pain of expression" in scientific research achievements.

▪ AI Enterprise Knowledge Base Workshop: Utilizing platforms such as Tencent IMA Knowledge Base or Dify, build lightweight AI knowledge bases to achieve intelligent aggregation, classification, and instant retrieval of R&D materials, competitive product information, policies, and regulations. Furthermore, AI-assisted functions such as planning meetings, generating agendas, and inviting scripts demonstrate its potential in enhancing organizational collaboration efficiency.

In this sharing session, multiple teams and individuals engaged in in-depth discussions on the application practices of AI in the pharmaceutical field, team collaboration, and professional exploration, generally emphasizing the core values of teamwork, human-machine collaboration, and innovative thinking.

Through specific cases, the participants demonstrated how to efficiently complete professional tasks with the help of AI tools. For example, some teams successfully interpreted and visually presented the literature on the mechanism of action of ADCs and CAR-T cell therapy by collaboratively reading literature, dividing tasks reasonably, and utilizing AI to generate core summaries, mind maps, 3D animations, and videos.

In addition, in the process of solving practical problems, some teams draw on industry cases to train AI to generate program outlines and content, and combine manual verification to ensure the accuracy of information. In science popularization work, the presenter points out that AI can effectively assist in understanding complex interdisciplinary knowledge, and also raises the challenge of balancing professionalism and popularity in content presentation.

It is noteworthy that, despite AI's remarkable efficiency in handling large-scale data and information, attention must still be paid to data security and the assessment of tool accuracy in practice, and excessive reliance on it should be avoided.

Further discussions did not shy away from a sober reflection on the technology itself: current AI still has limitations in terms of professional depth and logical coherence, and may pose ethical risks. Therefore, in promoting the application of technology, it is essential to emphasize the coordination between it and ethical norms as well as the long-term development needs of society.