Rottify Home Page

🚀 Rottify: Where AI Meets Chaos Theory

Turn Text into Viral "Brain Rot" That Secretly Teaches

📖 Project Story: From Mindless Scrolling to Microlearning

The Inspiration

We noticed something:

📱 Our friends were spending hours making "StudyTok" summaries, trying to make learning engaging.
🎥 Existing tools? Clunky. Professional video editors? Expensive. AI voiceovers? Uncanny valley nightmares.
⚡ Meanwhile, TikTok’s algorithm was optimizing for brain rot—pumping out snappy, high-energy, chaotic content.

So, during a 10 AM to 6 PM hackathon, fueled by water and existential dread, we had a thought:

What if we could automate the entire "brain rot" video creation process—turning any text into a viral-ready, high-energy microlearning video?

And so, Rottify was born.

🔥 The Breakthrough: Rottify’s Unholy Trinity

We combined state-of-the-art AI models to make text unreasonably engaging:

🧠 Perplexity Sonar – Shreds documents into viral hooks ("Shakespeare’s rizz vs. Newton’s trauma")
🎤 ElevenLabs – AI voiceovers trained on 69,420 hours of “slay queen” ASMR compilations
🎥 Remotion + FFmpeg – Generates TikTok-style videos with hypnotic captions & meme timing

The result?
Rottify makes 60-second videos faster than an over-caffeinated YouTuber.

🛠️ How We Built It

Tech Stack

How It Works

Text Processing:
- Perplexity Sonar extracts key insights & hooks from dense documents.
- It rewrites text into snappy, viral-friendly scripts (e.g., “Why your Econ textbook is gaslighting you”).
Voiceover Generation:
- ElevenLabs AI adds charismatic, Gen-Z-style narration with custom voice presets ("Girlboss Gaslighting", "NPC Streamer #57").
Video Synthesis:
- Remotion + FFmpeg dynamically syncs captions, visuals, and effects.
- Uses AI meme detection to auto-insert relevant sound effects & text animations.

Bonus: We built an Edge-Cached Pipeline in Next.js, so Rottify runs 3x faster than industry tools.

🌟 What We Learned

💡 Remotion is underrated. We had zero prior AI/video experience, but in 8 hours, we built a fully automated content pipeline.
⚡ AI storytelling is hard. Fine-tuning text-to-video pipelines to not sound robotic took experimentation & memes.
🔄 Syncing AI-generated voices to visuals is pain. We learned how to adjust timing algorithms using FFmpeg.

🚧 Challenges We Faced

AI Syncing Woes:
- Making AI voiceovers match caption timing required black magic (and subbae lib).
- We hacked FFmpeg scripts to auto-align subtitles to voice pitch changes.
Campus Wi-Fi = Pain:
- Training AI models on Wi-Fi slower than a sloth on melatonin was... frustrating.
- We offloaded voice generation to Edge Functions for faster processing.
TikTok Algorithm is a Mystery:
- We tested which video styles got the most engagement.
- Turns out, flashing captions + absurd sound effects = content goldmine.