Inspiration
Research is cumulative, but the tools we use to read it often aren't. Scientific papers contain interconnected concepts, yet understanding them means navigating dense PDFs, scattered notes and endless AI conversations.
As students and researchers, we wanted to make this process more intuitive, visual and connected.
What it does
ResView turns research papers into interactive networks of knowledge.
Upload a PDF, and AI generates a Paper Graph connecting its key concepts, explanations and source references. Users can explore the graph alongside the original paper and rate their understanding.
Concepts they understand contribute to a personalised Knowledge Graph, connecting ideas across multiple papers. Through integrations with ChatGPT and Claude, users can discuss concepts with their familiar AI assistants and update their graphs.
How we built it
We used OpenAI GPT-6.1 Sol for paper analysis and graph generation, React Flow for interactive visualisation, and FastAPI, Supabase and Clerk for the backend, storage and authentication. We also built an MCP integration to connect ResView with existing AI assistants.
Challenges and what we learned
Our idea evolved significantly throughout development. What began as a way to visualise research papers gradually became a broader vision for connecting scientific knowledge across papers and AI conversations.
Our biggest challenges were extracting meaningful relationships from complex papers, creating an intuitive graph interface, and integrating everyone's work into one functioning application.
It was a great experience working together, experimenting with ideas and seeing how much the product changed as we built and tested it.
What's next?
We want ResView to become increasingly personalised, adapting explanations and connections to each user's existing knowledge and learning style.
Our vision is to make scientific understanding something you can see, connect and continuously build upon.
Built With
- clerk
- dagre
- fastapi
- gpt-6.1
- javascript
- katex
- mcp
- model-context-protocol
- openai
- postgresql
- pymupdf
- python
- react
- react-flow
- react-pdf
- sqlalchemy
- supabase
- vercel
- vite