Elevator Pitch

Turn long-form videos into viral-ready TikTok clips with one click!


Inspiration

Every day, creators spend hours scrubbing through long-form videos trying to find one moment that might go viral. In today’s algorithm-driven world, consistently posting short-form content isn’t optional — it's the primary way creators grow, stay relevant, and reach new audiences.

Podcasters, streamers, and video creators sit on hours of footage and need multiple short clips by the next morning. One of our team members is a creator with 10K+ followers and 4.5M+ total views, and behind every post is hours of scrubbing, guessing, and hoping the right moment finds its audience.

The content is there — but finding it is a grind.

We built Lets Go Viral to eliminate exactly that.

Instead of hunting for moments, creators upload once — and data-driven AI does the rest.


What It Does

Lets Go Viral is an AI platform that transforms long-form content into viral-ready short clips.

Upload your podcast, livestream, or YouTube video, and our pipeline takes it from there:

Analyzes Everything

Our engine:

  • Transcribes your video
  • Detects shot changes and natural breakpoints
  • Runs multimodal Gemini analysis to identify:
    • Emotional peaks
    • Humor
    • Controversy
    • High-engagement moments

Every second of your footage gets evaluated.

Scores Virality Potential

Segments are scored and ranked based on predicted engagement, so you always know which clips are worth your time.

Generates Optimized Short Clips

We extract 15–90 second clips formatted for:

  • TikTok
  • Instagram Reels
  • YouTube Shorts

Boundary snapping ensures every cut lands on a natural pause or scene transition — no awkward mid-sentence chops.

Intelligent Sound Design (Powered by ElevenLabs)

Our engine:

  • Detects where sound effects increase impact
  • Generates custom, content-aligned SFX
  • Places them at optimal timestamps

Background Music

Generates music matched to the emotional tone and energy of each clip.

The result? Professionally edited content ready to upload.


Human-in-the-Loop Control

Because the best tool is one you can trust, creators stay in full control.

  • Adjust timestamps
  • Move or swap sound effects
  • Change background music
  • Prompt our AI assistant to refine the clip

When you’re happy, download and post.

We handle the heavy lifting. You take the credit.

What used to take hours now takes minutes.


How We Built It

We designed Lets Go Viral as a scalable, cost-efficient, AI-native pipeline.


Frontend (React + Vite)

  • React 18 + TypeScript
  • Vite for fast local development and optimized builds
  • TailwindCSS + Emotion for styling
  • Lucide for UI iconography
  • Real-time progress updates using Server-Sent Events

Backend & Infrastructure

  • Node.js + Express API
  • Google Cloud Run (scales to zero when idle)
  • Cloud Run Jobs for asynchronous video processing
  • Terraform-managed infrastructure
  • Cloud SQL, Pub/Sub, Cloud Storage, CDN

The entire pipeline runs asynchronously and streams live progress updates back to the app — so users never stare at a frozen screen.

Data & AI Pipeline

Data Extraction

  • Transcript + word-level timestamps
  • Shot boundaries
  • Silence intervals

AI Scoring

Gemini scores segments for:

  • Engagement likelihood
  • Emotional resonance
  • Viral potential

What’s Next for Lets Go Viral

We are incredibly passionate about this idea and will continue to work on this as a startup! We also deeply appreciate Hacklytics for giving us the opportunity and tools to get it off the ground!

Future Steps

  • Direct publishing to TikTok, Reels, and YouTube Shorts
  • Trend matching — suggest edits based on trending sounds and formats
  • AI-generated thumbnails with engagement prediction

Built With

  • and-speech-to-text-ai-analysis:-vertex-ai-(gemini)-for-clip-scoring-and-viral-potential-prediction-voiceovers:-elevenlabs-api-for-ai-generated-intros-and-hooks-infrastructure:-terraform-managed-gcp-(cloud-sql
  • gcp
  • pub/sub
  • video-intelligence-api
Share this project:

Updates