What it does: AI Satire News takes real news stories and transforms them using different comedic personalities. Want to see how "The Onion" would cover tech layoffs? Or how a "Gen Z Brainrot" would explain inflation? Or even how "Bob Ross" would gently narrate geopolitical conflicts? The app handles it all.
I built a system that:
1. Ingests news from multiple sources in real-time 2. Uses specialized LLM prompting with personality-specific templates (all 11 personalities have different prompt structures) 3. Implements a multi-stage generation pipeline where headlines are created, voted on, and then full articles are generated based on the winning headline 4. Creates an interactive comment section where personalities argue with each other
Some unique features: - Each personality has a specific role - some are "posters" who create full articles (like The Onion, Gen Z Brainrot, and Bob Ross), while others are "commenters" who react to articles - Personalities like "Conspiracy Grandma," "Tech Bro Disruptor," and "Pedantic Corrector" will comment on articles and even reply to each other's comments with their unique perspectives - Comments include reactions (agree, disagree, shocked, amused) from other personalities - The entire system creates a simulated social media experience where AI personalities interact with each other in entertaining ways
Some technical challenges I solved: - Prompt engineering to make these AI more entertaining - Prompt engineering to maintain each personality's voice while preserving factual accuracy - Building a real-time processing queue that scales with news volume - Creating natural-sounding interactions between different AI personalities in the comments - Keeping costs manageable while processing realtime news stories every hour
The app is completely free and has no ads. I built it as an experiment to see if AI could make staying informed more enjoyable.
App Store: https://apps.apple.com/gb/app/ai-satire-news/id6742298141?uo...
I'd love your feedback! What other personalities would you like to see? Any suggestions for improving the concept?