PrePaper

main page 1
main page 2
example of an analyzed paper
example of an explained topic

Inspiration

In the age of AI reading a paper might be harder than you thought. Modern life nukes your attention span, AI tools throw information at you faster than you can process it, and opening a dense PDF after a long day feels wearing We wanted to make papers approachable again — not by simplifying them, but by giving people the conceptual “entry point” they’re missing.

What it does

PrePaper takes any scientific PDF and instantly builds a personalized “concept map” for you. It extracts the core topics, merges them into a clean list, and generates clear explanations for each — basically a mini‑course that teaches you the ideas behind the paper before you read it. Instead of diving blind into 20 pages of jargon, you start with a guided overview that makes the whole paper understandable.

How we built it

We built a full backend pipeline that:

extracts and cleans the paper’s text
splits it into meaningful chunks
analyzes each chunk with AI to find key concepts
merges overlapping topics into a coherent list
generates human‑friendly explanations for every topic

On the frontend, we designed a clean interface where users upload a PDF and instantly get a structured, readable breakdown of the paper’s conceptual landscape.

Challenges we ran into

Getting consistent topic extraction across wildly different papers
Merging similar concepts without losing nuance
Keeping the explanations accurate, not oversimplified
Handling large PDFs efficiently
Making the frontend feel smooth while the backend does heavy processing

Accomplishments that we're proud of

We built a complete end‑to‑end system in a very short time
The topic explanations feel genuinely helpful, not generic
The interface is simple enough for anyone to use
The pipeline handles real scientific papers surprisingly well
We created something that actually improves learning, not just automates tasks

What we learned

Throughout building PrePaper, we realized that scientific papers themselves aren’t the real obstacle — it’s the entry point that’s hard. Once you understand the core ideas, the rest becomes surprisingly manageable. We also learned how much good UX matters, even in AI‑heavy tools; the experience around the model is just as important as the model’s output. Finally, we learned the classic hackathon truth: everything is chaotic, nothing goes as planned, but building something genuinely useful makes the chaos worth it.

What's next for PrePaper

Looking ahead, we want PrePaper to evolve from a concept extractor into a full learning companion. One of the first steps is adding visual concept maps, so users can see how ideas connect. We also want to make the interface more interactive — for example, letting users click a topic and instantly jump to the exact section in the paper where it appears.

Another direction is generating prerequisite knowledge lists, giving readers a quick “what you should know before reading” overview. We plan to expand to multiple languages and eventually turn PrePaper into a tool that students, researchers, and curious learners can rely on whenever they face a complex paper and want a guided, friendly way in.