NYUSH Digital Innovation Challenge 2025
INTRODUCTION
New York University Shanghai Digital Innovation Challenge (NYUSHDIC) is a tech-based innovation competition hosted by a student organization NYUSH Digital Innovation Community. Students from NYUSH and other universities are welcome to form cross-disciplinary teams, utilize cutting-edge digital technology, and pitch innovative ideas to address real-world problems. This year, the theme of the competition is AI Agent Unlock Business Innovation.
AI Agents, as one of the latest advances in artificial intelligence, are redefining how people interact with technology. They are intelligent systems that can perceive, reason, and act autonomously. Beyond executing human instructions, AI Agents are able to learn, adapt, and optimize, making them valuable tools for solving complex problems.
For this year’s challenge, under the theme AI Agent Unlock Business Innovation, participants will apply AI Agents to address practical business and social issues.
The competition will include several tracks composed of different fields, tentatively in the areas of Education, Mental Wellbeing, and Finance. Following the model from last year, each track will be supported by industry partners, faculty members or schools, who will set the problems, provide guidance, and conduct evaluations.
THEME
AI Agent Unlock Business Innovation
Track information
01 Education Track — Kiwi
Mentor: Prof. Hongyi Wen from NYUSH
Background:
Kiwi is a research platform developed at NYU Shanghai to support students and faculty in creating a seamless, intelligent learning environment. The Education Track of the 2025 Digital Innovation Challenge invites participants to design AI-enhanced educational tools inspired by the Kiwi Platform. While participants are not expected to modify Kiwi directly, they may connect their applications through the Model Context Protocol (MCP) or align with Kiwi’s operational framework for technical compatibility.
Future applications developed under this track can integrate with Kiwi via MCP-based interfaces, enabling seamless interoperability while keeping systems independent and modular. Illustrative examples include automated grading systems that provide formative feedback or AI-generated flashcards that extract key learning points. However, these are merely examples—participants are encouraged to explore innovative, AI-driven solutions that enhance teaching efficiency and empower self-directed learning.
The Key Goals:
- Develop Intelligent and Modular Tools: Create educational applications inspired by or connected to the Kiwi platform through MCP to ensure flexible integration and scalability.
- Advance AI-Driven Innovation: Design systems that leverage AI to enhance teaching efficiency and provide personalized learning experiences.
- Promote Student-Centered Learning: Empower learners to take greater control of their educational journey through adaptive and self-directed learning tools.
- Ensure Technical Interoperability: Maintain independence between systems while ensuring compatibility and seamless communication with the Kiwi platform.
02 Education Track
Building AI Agents for Student Motivation and Resilience
Cooperative Enterprise: ChillPrep
Background:
In modern education, psychological well-being is increasingly recognized as a key factor influencing academic performance and long-term success. Non-cognitive traits such as resilience, motivation, and emotional regulation often determine whether students can sustain progress in their learning journeys. Meanwhile, advances in generative AI and agent technologies have made it possible to deliver personalized, real-time mental health support at scale. This track explores how AI agents can create integrated mental health support systems in educational contexts—helping learners achieve holistic, sustainable growth through technology.
The Key Goals:
- Personalized and Scalable Support: Develop AI agents that build psycho-educational profiles and deliver tailored mental health interventions to approximate one-on-one professional support.
- Proactive Monitoring and Early Intervention: Create systems that analyze behavioral data to detect early signs of stress or motivational decline and provide preventive interventions.
- Context-Aware Coaching: Integrate emotional and cognitive support directly into learning platforms to provide just-in-time guidance during academic challenges.
- 24/7 Accessibility: Design agents that provide always-available, empathetic, and responsible support, with clear escalation to human professionals when needed.
- Address Ethical and Adoption Challenges: Ensure privacy protection, ethical boundaries, and user-friendly, stigma-free engagement through thoughtful design.
03 Healthcare Track
Designing AI Agents Across the Drug Development Life Cycle
Cooperative Enterprise: Henlius
Background:
The track challenges participants to build AI Agents to empower healthcare teams by efficiently gathering, synthesizing, and applying this distributed knowledge throughout the entire drug life cycle.
Developing a new drug is a lengthy, complex, and information-intensive process, from initial research to commercialization. Scientific breakthroughs, clinical trial progress, manufacturing standards, market intelligence, and internal operations are all critical information. However, they are scattered across various sources, including research papers, clinical databases, regulatory bodies, news sites, and internal knowledge bases. This fragmentation makes it difficult and time-consuming for healthcare professionals to access and act upon the necessary, timely, and accurate information, which is essential for success.
The Key Goals:
- Design an Integrated AI Agent System: Design an AI Agent (or a system of agents) that integrates at least two of the five core domains of the drug development life cycle: Drug Literature & Pipeline Monitoring; Clinical Trial Finder & Comparator; Manufacturing & Quality Assurance Assistant; Drug Licensing & Commercial Intelligence; Pharma Operations Assistant
- Showcase Multi-Domain Value: Demonstrate how combining multiple knowledge areas creates a more powerful and versatile digital assistant for healthcare teams.
- Incorporate Agent Capabilities: The AI Agent is expected to incorporate critical functions, including monitoring, summarizing, comparing, Q&A support, and proactive alerting, to help researchers, clinicians, quality/manufacturing teams, and business functions.
- Enable Faster, Better-Informed Decisions: The ultimate goal is to provide timely, accurate, and actionable information to support Henlius teams in making faster and better-informed decisions across various functions.
04 Finance Track
AI Agents Unlock Finance for Sustainability
Cooperator: Duke Kunshan University
Background:
The Finance Track of the 2025 Digital Innovation Challenge focuses on the intersection of AI agents, FinTech, and sustainability. AI agents, as adaptive decision-making systems, are transforming financial services, making them more efficient, transparent, and accessible. When combined with FinTech, these technologies can promote global sustainability by enabling green investment allocation, supporting carbon accounting, and aligning financial behavior with environmental goals. The track aims to inspire solutions that advance both financial services and sustainable development.
The Key Goals:
- Design Innovative Solutions: Task participants with providing innovative, ethical, and sustainable financial solutions. Harness AI for Sustainability: Demonstrate how AI agents can be used to unlock finance as a force for long-term sustainability and equitable progress.
- Focus on Key Technologies: Encourage exploration of technologies like blockchain, carbon accounting, and credit scoring systems for sustainable finance development.
- Prioritize Foundational Values: Value solutions based on Frontier Agents (autonomy, adaptability, real-world integration), Openness, Ethics & Inclusion, and clear Impact on Sustainable Development Goals.
- Produce Tangible Projects: Guide teams to complete tangible projects, supported by data and feasibility analysis, that have the potential for real-world implementation.