16-G | Devpost

Documents
Chat

Inspiration

In this project, we wanted to make new and AI technologies more accessible in research and scholar areas, making it easier to search and analyze information.

What it does

This platform allows users to make their own cloud data bases, where they can store different kinds of sources that are then analyzed and stored. This way, users can semantically search for information and interact with a chatbot to find the most relevant data. In addition, we implemented a text to speech functionality to allow more people to use this technologies.

How we built it

In order to take advantage of the most recent and advanced technologies, we implemented different services and APIs:

Challenges we ran into

The main challenge that we faced was to integrate the python services with the NEXT app. We began searching for cloud options, but eventually decided to implement a local solution for demonstration purposes.

Accomplishments that we're proud of

Implementation of different features using diverse technologies:

Authentication

Using Azure AD for user authentication.

Web and document scraping

Using python scripts and APIs to retrieve text from web links and documents.

Semantic search

Using embeddings to analyze texts and do a semantic search.

Context injection

Analyze large texts of information using vector and embedding technologies, also taking advantage of OpenAI chats.

Text to speech

Implemented AWS Polly to transform text into speech for more users to use the platform.

What we learned

We learned to use new services from Amazon, Google, Azure and the most relvant for this challenge: FRIDA from Softek.

What's next for Frida-Assistant

Even though the platform reached a production state that is actually working, there are still areas of improvement. Mainly regarding the scalability, since we would like to have a higher access of cloud storage.