Revolutionizing web search: The winning solution of AI21 Labs' Hackathon

The winner of AI21 Labs’ first ever hackathon is Rahel Gunaratne, who nabbed the first place prize with his WebIndexer project - An AI-based question-answering tool designed to help users find specific information using questions instead of keyword search within a webpage. WebIndexer can be integrated into any website, offering users a much-needed solution to the slow and tedious process of keyword search, and can serve as a valuable alternative to both ChatGPT and traditional search bars like Google.

Check out WebIndexer on Github | Try the demo on Streamlit

The AI21 Labs Hackathon, which was produced in conjunction with Lablab AI, kicked off on January 13th and concluded eight days later on the 21st. Over 1,900 participants took part in the event, with 24 different projects competing for the first prize.

The winning projects were chosen by a panel of judges which includes AI21’s Yuval Belfer (who also initiated and produced the event), Yotam Shinan, and Jack Yulzari, alongside representatives from NewNative - the company behind Lablab AI.

The projects were selected according to the following criteria:

1. It must be an innovative idea
2. It has to solve a real-world problem or pain point
3. It needs to be built on beautiful code
4. It must have great documentation

The Finalists

Coming in second place were three notable projects:

1. Kitchen Genie by Team Last Minute - A smart cooking solution that aims to reduce household food waste worldwide by using AI to help users determine what meals to prepare based only on the ingredients that they have left in their refrigerator.
   
   Check out Kitchen Genie on Github | Watch the demo on Youtube
   ‍
2. T.E.A (Test English AI) by Team Think Tank - A web application designed to help English students excel in language proficiency tests with the help of personalized, AI-generated reading exercises. Trained on IELTS Academic as well as general reading tasks, T.E.A brings users one step closer to a real-world exam experience.
   
   Check out T.E.A on Github | Try the demo on Heroku
   ‍
3. MedBrief by Team MedBrief - Designed with the aim of significantly reducing medical review time, MedBrief provides users with summaries of lengthy patient reports and highlights significant medical terminology. A big standout feature is Medbrief’s ability to answer questions based on an individual’s medical records by incorporating AI21’s tech with their own well-written code.
   
   Check out MedBrief on Github | Try the demo on Vercel

‍

‍" The hackathon's submissions left me, as a mentor, astounded. The number of production-ready projects, coupled with the broad range of innovative solutions they provided to real-world issues, serves as proof of the participants’ talent. Additionally, it's a testament to the infinite possibilities of generative AI. "

- Jack Yulzari