-
Initial setup
-
Test case
Inspiration
While taking an English class last semester, I was told to use the library website, and I decided that with current advancements in AI, there should be a way to build a product that can be useful to professors, post-grads, and undergrads.
What it does
It is essentially a powerful research assistant that helps students and professors get to the resources they would spend weeks on in a matter of minutes. This tool is designed to leverage AI to alleviate the time-consuming part of research and learning.
How we built it
This tool is built on Python. Since it’s an MVP, it has a simple Flask web server for the look and feel. Under the hood, this system is powered by AWS Bedrock models with my custom agentic framework.
Challenges we ran into
Having multiple agents talk to each other on the backend and having them stick to a consistent format was quite challenging. I solved it with a lot of trial and error and prompt engineering.
Accomplishments that we're proud of
The fact that we were able to get this ready in time despite all the API issues and framework problems is honestly quite impressive!
What we learned
I learned that, especially for emerging technologies, the systems that are already in place are going to be inefficient. It is an incredible feeling that I, as a solo dev, and trillion-dollar companies like Google and Microsoft are at the same place in terms of applications of GenAI. That gives me hope and motivation to build products that will change the world.
What's next for digiLIB
My goal with this system is to get it into the hands of as many people as I can. I want them to experience it knowing that they can never go back to conventional research and to build this system such that it becomes useful to every person in the academic community.
Log in or sign up for Devpost to join the conversation.