Macrodata

Real-time trending narratives about your brand. Personalized audio reports.

Comment 2

Inspiration

Igor worked on a similar challenge before and can see its potential to become valuable for businesses. Also, Lumon Industries (obviously). However, we did this app in retro style just for fun.

What it does

Macrodata helps you analyze trending narratives about your brand, helping you make data-backed decisions. When you enter a brand (or any other term) into the app, it fetches the latest tweets about this topic, categorizes them by trends, and provides a short text and audio briefing about the brand. It allows you to see current prevailing narratives that you should pay attention to.

How we built it

Coding

Front-end: React + TypeScript Backend: Python

APIs:

Mistral AI Agent: Generatic topic names based on top tweets and daily briefing text Perplexity AI Agent: Generating tweet-looking data based on the latest news (X API is not free) EleventLabs Agent: Audio briefing with custom voice

Categorization:

  • SentenceTransformer ('all-MiniLM-L6-v2') for semantic embeddings
  • TF-IDF with n-grams for word pattern analysis
  • K-means clustering for grouping similar posts

Challenges we ran into

  1. Accessing real-time data. X API is very limited, so we needed to improvise and generate fresh data using an AI agent
  2. Topic categorization is far from perfect
  3. Categorization takes too much time

Accomplishments that we're proud of

  1. It works and shows real data
  2. Design

What we learned

  1. Learned a little Python
  2. Learned that AI is not good at summarising large amounts of data
  3. Working with friends is super fun

What's next for Macrodata

  1. Igor wants to work on the challenges we encountered
  2. Add sentiment analysis and more insights
  3. Igor wants to build a company around it

Notes

Deployment is not really working. Didn't have time to figure out

Built With

Share this project:

Updates