Too much news, too little time?
I recently built a small AI project to experiment with summarizing what actually matters for any topic and turning it into audio.
From topic → insight → audio
Here's what happens under the hood:
- Scrapes relevant articles from Google News using Playwright
- Pulls real-time discussions from Hacker News via Playwright MCP
- Uses an AI model to generate a concise, unified summary
- Converts the summary into speech using ElevenLabs
The idea was to explore how browser automation + MCP tools + AI models can work together to reduce information overload and deliver content in a more consumable way.
What I learned
- Using Playwright beyond testing
- Combining multiple data sources for better context
- Understanding summarization trade-offs
- Making AI outputs more accessible through audio
Still evolving, but a fun and insightful experiment.