DocsAlyzer
DocsAlyer is a platform that enables users to pose questions in everyday language, leveraging internal documents. The responses provided are dependable, supported by quotes and references from the original source material, ensuring utmost trustworthiness.
Project Description
This project is powered Danswer and OpenAI. Danswer is an open source Python project, this was a learning project to understand the capability of Danswer and OpenAi. I thoroughly enjoyed using it and I cant wait to help improve this project. I have spoken to many judges and they all have great insight as to where to improve this project. I am looking forward to collaborating with everyone :).
Danswer adn OpenAI allows the user to use natural language to against documents such as Documentation for your protocol, tutorials, how to guides, explanation, and references. All of your resources are scraped and indexed. The AI is now able to filter the information and find the right resource via direct quote matching or use Deep Learning to asset the intent and find the closest source.
It can connect to websites and/or github to find the data to store.
How it's Made
It is a python based project, that uses web scraping and parsing. It stores it into a PostgreSQL DB for fast retrieval. It used Generative AI models, Intelligent Document Retrieval using the latest LLMs, and it connects with Github API.
We used these sources for documentations: Ethereum Foundation, Wallet Connect, Push Protocol, The Graph, UMA
We also followed https://diataxis.fr/ educaiton resources and will continue build upon it.
