Inspiration
Time is a resource that ought to be well utilized. As a developer sometimes large developer docs are intimidating and overwhelming. Using generative AI sometimes seems like a fix but responses are outdated and at times wrong.
Konverse aims to reduce time used in scrolling long documents and instead combine TIDB Vector with Gemini AI for faster comprehension and information retrieval.
What it does
Konverse uses TIDB Vector Search to extract related information and use it as context for Generative AI for updated responses that uses recent data. With Konverse your pdf document becomes source of truth for generative AI.
Konverse uses Gemini AI to create text embeddings and TIDB Vector for storage and retrieval to get close matches, the use returned documents as context for Gemini AI prompt.
Konverse aims to reduce time it takes to read documents and instead use Generative AI and Vector Databases for quick and concise information retrieval.
How we built it
We built Konverse front-end using Angular and back-end uses python flask framework. Our Live site will be hosted on Vercel. Live python code we used Python everywhere.
Challenges we ran into
Accomplishments that we're proud of
Working MVP. Lots room for improvements though.
What we learned
What's next for Konverse
We are still developing Konverse to add support for more documents, add authentication, improve UI/UX.
Built With
- angular.js
- flask
- python
- python-everywhere
- tidb
- vercel
Log in or sign up for Devpost to join the conversation.