NabuAI

Have you ever found yourself drowning in a sea of research papers, desperately trying to organize 20+ citations while piecing together your thesis? Check our NabuAI to streamline the process!

Comment

Have you ever found yourself drowning in a sea of research papers, desperately trying to organize 20+ citations while piecing together your thesis?

Let me introduce NabuAI

Named after the Babylonian god of literacy, we are an AI-powered tool designed to revolutionize how students and researchers interact with academic material. Think of it as your personal organizer for writing papers.

Here’s how it works:

Know what you want to write about but don't know where to find sources?

You start by typing your thesis or even a rough idea into a chatbox. For example, “Public speaking and its impact on performance.” Instantly, our AI uses semantic decomposition through Gen AI to pick research papers or online articles that relate to your thesis. It breaks down the ideas into concepts - "public speaking," "performance," "stress response," etc.

If you want to search specifically for studies specifically about public speaking and performance, you can use our library to do so. We have a database of over 250 million scholarly works to search from, so you can select any article you’re interested in, and add it to your collection.

See a title that interests you? Rather than trying to read through the jargon in the paper's abstract or scanning 30+ pages of text, these papers will stay in Singlestore’s vector database to be automatically processed in the background by Aryn’s API to analyze the paper and generate a summarized report using GPT4o.

We have both a textual and graphical summary, allowing you to pick and choose the most relevant sources for you.

Example use cases

  • "I remember reading this statistic about GPA and public speaking ability, which paper was that in?"
  • "Find me a paper published within the last two months that is about public speaking"
  • "Find other works by the same author"
  • "How do these two studies compare in results? Which one is more relevant and credible?"
  • "Can you plot the data of these two papers together in one graph?"
  • "Generate MLA citation for these selected sources"

What we are most proud of

Our handling of large data, where we can process gigabytes of research paper information to be used as context for an AI chat model. By using singlestore’s vector database, we can process much more data than even GPT or Claude. In addition, our live scraping of academic databases and our future goals of scraping news articles and more sources will keep all our information up-to-date.

You can compare, contrast, summarize, and find statistics from each paper you upload. Finally, once you are happy with your sources, you can generate citations in the format of your preference, and copy them into whatever document you need.

To dive into technicalities

We used NextJS for the frontend and backend integration, with Firebase to store chatbot message data and collections. For GEN AI, we used a flask backend with GPT4o, and we processed pdfs using Aryn API to convert them to vector embeddings in Singlestore.

Our plans for the future:

  • Expand database to include multiple academic libraries and frequently updated sources.

  • User-generated visual aids and summaries.

  • Compatibility scores to find papers that are most applicable to your thesis.

  • Reliability scores to find the most trustworthy and recent studies.

Built With

Share this project:

Updates