1. Main project screenshot(A dashboard showing the research question input, AI-generated summary, and source references.)
2. Research results screen(Show relevant studies, simplified summaries, and clickable citations.)
3. System architecture(Illustrate how the user, Gemini, research retrieval, and cited response connect.)

Inspiration

Medical research contains valuable information, but understanding complex research papers and finding reliable evidence can be time-consuming. We wanted to explore how AI could make medical research easier to access and understand. This inspired us to create MedResearch AI, an intelligent research assistant that helps users explore medical studies while prioritizing transparency, reliable sources, and responsible AI.

What it does

MedResearch AI is a Gemini-powered medical research assistant that allows users to ask questions about medical research in natural language. It helps users discover relevant studies, understand complex findings through simplified summaries, compare evidence across sources, and explore supporting references. Users can also ask follow-up questions to investigate a topic further. The system is designed for research and educational purposes, not for medical diagnosis or treatment recommendations.

How we built it

We designed MedResearch AI around Google's Gemini models, AI agent workflows, and retrieval-augmented generation (RAG). The planned workflow retrieves relevant information from trusted medical sources, uses Gemini to generate clear, context-aware summaries, and provides references to help users verify the information. Tool calling enables the agent to perform research tasks, while a user-friendly web interface makes the experience accessible. Our focus is on combining useful AI capabilities with source transparency and responsible information handling.

Challenges we ran into

One of our biggest challenges was considering how to reduce unsupported AI-generated claims when dealing with medical information. Medical research also contains technical terminology that can be difficult to simplify without losing important context. We identified source reliability, accurate summarization, relevant information retrieval, and clear communication of AI limitations as key challenges to address.

Accomplishments that we're proud of

We are proud of developing a practical AI project idea that applies agentic AI to a meaningful real-world problem. MedResearch AI brings together conversational AI, medical research exploration, evidence comparison, and source transparency in one proposed workflow. We are especially focused on helping users understand complex information without presenting AI-generated content as a substitute for professional medical advice.

What we learned

Through this project, we explored how large language models, retrieval-augmented generation, and tool calling can support more useful AI applications. We also learned that a trustworthy AI assistant needs more than fluent responses: it must prioritize reliable information, provide verifiable references, communicate uncertainty, and recognize its limitations.

What's next for MedResearch AI

Our next steps are to integrate and test medical literature retrieval, improve source ranking, implement evidence comparison, and strengthen citations so users can trace answers back to original studies. We also plan to evaluate the quality and factual consistency of generated summaries, improve the user interface, and introduce safeguards against unsupported medical claims. Our long-term goal is to make medical research easier to explore, understand, and verify through responsible AI.