World News Reader

Inspiration

We were talking about how same news can be covered in different ways in different countries. We wanted to build this tool so users can get an insight into what the people in other countries are seeing as their news.

What it does

It gets the trending news articles and then passes them to gpt 4 to generate a concise newsletter. Users can go on the website, select the country they are interested in and they'll get a newsletter with title, description and source giving them an idea of most recent happenings in these countries.

How we built it

We used Vercel's AI package to render the frontend and openai's APIs to interface with GPT.

Challenges we ran into

We are all backend devs, making a usable frontend with streaming the output from GPT was the biggest challenge.

Accomplishments that we're proud of

Figuring out the frontend work with the help of chatGPT :)

What we learned

We learnt the limitations of applications of GPT-4. We started with the idea of optmizing databses, and then moved on to try detecting malware. Both of them required more work then using GPT-4 out of the box so were out of the scope for this hackathon. We learnt frontend dev work and used GPT in things it is good at, like generating newsletter.

What's next for World News Reader

This tool can be further improved by giving users options to generate newsletter based on specific terms like "USA", "Bitcoin", "War" etc. We can also give options for fifields like Science, Sports, Politics etc. We can also expand the country coverage to cover more geographies.

Built With

• gpt
• nextjs