Supernova

One URL. Complete Marketing.

Comment

🚀 Supernova — One URL. Complete Marketing.

Transform any product URL into a complete, production-ready marketing campaign in minutes. Genblaze orchestrates the AI reasoning pipeline while Backblaze B2 provides persistent asset storage, retrieval, and export.

Supernova turns one URL into a complete marketing campaign.

The average marketing team juggles 12+ disconnected tools:

  • Web scraper
  • AI writing assistant
  • Market research platform
  • Design software
  • Voiceover tool
  • Video editor
  • Cloud storage
  • Analytics dashboard
  • Scheduler
  • Email marketing platform

Average cost: $500-600/month Average time: 2-6 weeks per campaign Team required: 5-10 specialists

Supernova changes everything. One URL. One click. Complete campaign. Download ZIP.

⚡ The Solution: Autonomous AI Marketing

Supernova is not a simple prompt‑to‑video wrapper. It's a fully autonomous marketing department that transforms a single product URL into a complete, production‑ready campaign. Under the hood, specialized AI agents handle each task. Genblaze acts as the orchestration layer for Supernova, coordinating multi-model reasoning, fallback execution, structured outputs, and intelligent provider selection. Instead of relying on a single model, Genblaze manages resilient AI workflows across market intelligence, script generation, and campaign planning. The remaining stages use dedicated best-in-class APIs (Jina AI, Pexels, ElevenLabs, Remotion, Backblaze B2) to ensure each step is performed by the optimal tool.

The 7‑Stage Autonomous Pipeline

1. Intelligent Web Scraping
Jina AI extracts clean product data from any URL—Amazon, Shopify, or a custom store. Title, description, price, features, and high‑quality images are captured automatically.

2. Market Intelligence & Competitor Analysis
Genblaze orchestrates a multi‑model research phase. DeepSeek V3 and Claude analyze the product's market position, identify competitors, profile the target audience, and recommend high‑converting marketing angles. The output is a structured JSON report—not a vague paragraph.

3. Professional Script Generation
A multi‑model fallback cascade (Qwen → DeepSeek → Claude → Gemini) generates scripts optimized for video. Duration‑aware prompts ensure exactly the right word count for 15/30/45/60‑second ads, in both direct‑response and organic storytelling styles.

4. Cinematic B‑Roll Curation
Pexels API searches 20+ product categories to fetch relevant, license‑free video clips and images. Orientation filters and quality thresholds guarantee assets that match the brand palette and video format.

5. Natural Voiceover Production
ElevenLabs transforms the script into a natural, expressive voiceover with fine‑tuned stability, similarity, and speaker boost settings. The audio is perfectly synced to the video timeline.

6. Video Composition
Remotion assembles everything into a polished video: scenes, B‑roll, voiceover, and a CTA. Multiple customization options give marketers full creative control.

7. Enterprise Storage & Export
Every asset—scripts, reports, images, video—is uploaded to a structured Backblaze B2 folder. Presigned URLs allow secure sharing, and one‑click ZIP export bundles the entire campaign for offline use.

🎬 Video Generation Options

Supernova offers two powerful video styles, each with deep customization controls.

Content Styles

Mode Purpose Ideal For
Sales (High-Conversion Ad) Aggressive problem-solving hooks, direct‑response scripting, urgent CTAs Paid social ads, product launches
Viral (Organic B‑Roll) Aesthetic storytelling, educational narratives, natural product integration TikTok, Reels, YouTube Shorts

Video Configuration

Setting Options
Aspect Ratios 16:9 (Desktop/YouTube) or 9:16 (Mobile/Stories)
Duration 15s, 30s, 45s, or 60s
Brand Palettes 8 premium presets: Noir & Gold, Old Money Navy, Emerald & Pearl, Minimalist Monochrome, Rose Quartz & Wine, Cyber Neon, Sunset Ochre, Eucalyptus & Mint
Caption Styles 6 styles: Clean Minimal, Cinematic Outline, Luxury Accent, Bold Impact, Underline Focus, Feature Badge

Each palette applies a primary accent color, a secondary background, and matching text tones—so every ad feels professionally designed.

🖼️ AI Image Studio

Supernova includes a dedicated AI Image Studio for generating standalone ad creatives.

Creative Styles

Style Vibe
Bold Dynamic & eye‑catching
Luxury Premium & elegant
Minimal Clean & refined
Vibrant Colorful & playful
Elegant Sophisticated & chic
Modern Futuristic & sleek

Output Formats

Name Aspect Ratio Best For
Square 1:1 Instagram feed, Facebook
Landscape 16:9 YouTube thumbnails, banners
Portrait 9:16 Stories, Reels, TikTok
Instagram 4:5 Optimized for Instagram feed

Images are generated from the product photo using AI style transfer, then automatically stored in Backblaze B2 alongside the rest of the campaign assets.

What Makes Supernova Different

Typical AI Tool Supernova
Single AI call Multi‑agent pipeline orchestrated by Genblaze
Manual tool switching One URL → complete campaign, no human intervention
No brand customization 8 palettes, 6 caption styles, multiple formats
Local or no storage Enterprise B2 storage with CRUD, search, and ZIP export
Fragile, no fallback Multi‑model fallback cascade with circuit breaker logic
Demo‑only Production‑grade: auth, health checks, request tracing, graceful degradation

Real‑World Impact

A marketing team that used to spend 2–6 weeks and $500–600/month on a dozen disconnected tools can now generate a complete, publish‑ready campaign in under 5 minutes at a fraction of the cost. Supernova replaces the entire fragmented stack, delivering campaigns in minutes instead of weeks.

🏗️ Architecture

┌─────────────────────────────────────────────────────────────────┐
│                      SUPERNOVA ARCHITECTURE                     │
└─────────────────────────────────────────────────────────────────┘
                             ┌─────────┐                           
                             │  USERS  │                           
                             └────┬────┘                           
                                  │                                
                                  ▼                                
┌─────────────────────────────────────────────────────────────────┐
│          FRONTEND — Next.js 15 + React 19 + TypeScript          │
│ • 68 Components with Framer Motion  • 11 Serverless API Routes  │
│ • 8 Brand Palettes, 6 Caption Styles, 2 Video Ratios            │
│ • Vercel Edge Runtime                                           │
└────────────────────────────────┬────────────────────────────────┘
                                 │                                
                                 ▼ HTTPS                          
┌─────────────────────────────────────────────────────────────────┐
│                BACKEND — FastAPI + Python 3.10+                 │
│ • 3 API Routes: /script, /market-intelligence, /storage         │
│ • GenBlaze AI Orchestration Layer                               │
│ • 828-line B2 Storage Module                                    │
│ • Railway Cloud                                                 │
└────────────────────────────────┬────────────────────────────────┘
                                 │                                
       ┌─────────────────────────┼─────────────────────────┐      
       │                         │                         │      
       ▼                         ▼                         ▼      
┌──────────────┐        ┌──────────────────┐      ┌──────────────────┐
│   GENBLAZE   │        │   BACKBLAZE B2   │      │  EXTERNAL APIs   │
│ ──────────── │        │ ──────────────── │      │ ──────────────── │
│ Multi-Model  │        │  S3-Compatible   │      │ • Supabase       │
│ Fallback:    │        │  Object Storage  │      │ • Pexels         │
│ • Qwen 3.6   │        │                  │      │ • ElevenLabs     │
│ • DeepSeek   │        │ • Full CRUD      │      │ • Jina AI        │
│ • Claude     │        │ • Presigned      │      │                  │
│ • Gemini     │        │ • ZIP Export     │      │                  │
│              │        │ • Search         │      │                  │
│  OpenRouter  │        │                  │      │                  │
└──────────────┘        └──────────────────┘      └──────────────────┘

Genblaze Orchestration Layer

Genblaze is the intelligence backbone of Supernova. Rather than calling models directly, every reasoning task passes through a centralized orchestration layer that handles:

  • Multi-model routing
  • Structured JSON outputs
  • Automatic fallback logic
  • Error handling
  • Provider abstraction
  • Request tracing

This architecture allows Supernova to remain reliable even when individual AI providers fail, rate-limit, or return invalid responses.

🎯 How We Built Supernova

✅ Real-World Utility

The Problem: Marketing teams waste $500-600/month on 12+ disconnected tools and 2-6 weeks per campaign.

Our Solution: Supernova transforms any product URL into a complete marketing campaign in under 5 minutes for a fraction of the cost. Real users—small businesses, marketers, content creators—can actually use this today.

Proof of Real-World Usage:

  • ✅ Users paste a real product URL (Amazon, Shopify, any website)
  • ✅ AI scrapes product details automatically via Jina AI
  • ✅ Market intelligence analyzes competitors, audience, and positioning
  • ✅ AI generates professional scripts for 15/30/45/60 second videos
  • ✅ Pexels provides cinematic B-roll assets
  • ✅ ElevenLabs generates natural voiceovers
  • ✅ Remotion composes complete video with captions
  • ✅ Backblaze B2 stores everything persistently
  • ✅ One-click ZIP download for offline use

✅ Production Readiness

The Problem: Most hackathon projects are demos that break in production.

Our Solution: Supernova is built like a funded startup:

Feature Implementation
24,848 lines 21,999 frontend + 2,849 backend
14 API endpoints 11 Next.js serverless + 3 FastAPI
68 React components Full UI library with animations
Supabase Auth Row Level Security policies
Health endpoints Provider status, B2 availability
Request ID tracing Every API call has unique ID
Circuit breakers Error classification, graceful fallbacks
Graceful degradation API starts even without B2
TypeScript strict mode Enforced throughout
5 deployment guides Complete documentation

Production Deployment:

  • Frontend: Vercel Edge Runtime
  • Backend: Railway Cloud
  • Database: Supabase PostgreSQL
  • Storage: Backblaze B2

✅ B2 Storage & Data Orchestration

The Problem: Most projects upload a file and forget it.

Our Solution: We built a complete enterprise storage system:

campaigns/{campaign_id}/
├── metadata.json         # Campaign metadata
├── data/                 # Scripts, market intelligence reports
└── images/               # Product visuals, generated ad posters

What We Built (828 lines in b2_storage.py):

Operation Description
Upload Asset Binary upload with metadata
Upload JSON Structured data with metadata
Download Object Raw bytes retrieval
Generate Presigned URL 3600s temporary links
List Campaigns All campaign IDs
List Objects Files in campaign
Search Campaigns Filter by term, status, date, provider
Get Campaign Full metadata + presigned URLs
Delete Object Single file removal
Delete Campaign Batch cleanup
ZIP Export One-click archive download

11 dedicated storage API endpoints. Zero hardcoded paths. Graceful degradation when B2 is unavailable.

✅ Use of Genblaze

The Problem: Most projects make one API call and hope it works.

Our Solution: We built a production-grade AI orchestration layer:

# genblaze_openai.chat() is the foundation
response = chat(
    model="qwen/qwen3.6-flash",
    prompt=prompt,
    api_key=api_key,
    base_url="https://openrouter.ai/api/v1",
    temperature=0.7,
    max_tokens=800
)

Multi-Model Fallback Cascade:

Priority Script Generation Market Intelligence
1 (Primary) qwen/qwen3.6-flash deepseek/deepseek-v3.2
2 (Fallback) qwen/qwen3.7-plus anthropic/claude-sonnet-4-20250514
3 (Fallback) qwen/qwen3.6-plus:free google/gemini-2.5-flash
4 — qwen/qwen3.6-plus
5 — mistral/mistral-nemo

Intelligent Circuit Breaker:

Error Action
404 / Model not found Try next model
Auth failure Fail fast
Rate limit (429) Fail fast
Network timeout Try next model
Empty response Try next model

Provider Catalog + Structured Logging:

PROVIDERS = {
    "openai": ProviderConfig(model=_get_script_model(), temperature=0.7),
    "market_intel": ProviderConfig(model=_get_market_intel_model(), temperature=0.3)
}

Every call logs: model used, timing metrics, script length, models tried before success.

🔧 APIs Used

API Purpose Integration
Genblaze/OpenRouter AI orchestration, multi-model fallback genblaze_openai.chat()
Backblaze B2 Campaign storage, ZIP export, presigned URLs boto3 S3-compatible
Supabase Auth, PostgreSQL database Row Level Security
Pexels B-roll videos and images Video/image search
ElevenLabs Text-to-speech voiceover eleven_flash_v2_5 model
Jina AI Web scraping, markdown extraction r.jina.ai reader
Vercel Frontend deployment Edge Runtime
Railway Backend deployment Persistent service

📦 What We Built: The Supernova Ecosystem

Live Platform Landing Website Engineering Journal Pitch Deck
Main App Landing Journal Pitch Deck
GitHub GitHub GitHub GitHub

Quick Start:

git clone https://github.com/robloxsagax-web/Supernova.git
cd Supernova && npm install
cp .env.example .env.local  # Add your API keys
npm run dev

🏆 Why Supernova Stands Out

Criterion Capability Supernova
Real-World Utility Demo script or single feature Complete end-to-end pipeline from URL to downloadable ZIP
Production Readiness Hardcoded values, no error handling 24,848 lines, circuit breakers, health endpoints, graceful degradation
B2 Storage & Data Orchestration Single file upload or no storage Full CRUD, presigned URLs, ZIP export, campaign search, 828-line module
Use of Genblaze Single API call with no fallback Multi-model cascade, error classification, provider catalog, structured logging

Product feedback submitted to Genblaze SDK: Feature Request: Native Chat/Text Provider

🎬 Watch Supernova in Action

🎬 Demo Video: Video

🚀 Try the App: Landing Page | Direct App Login – click "Explore Demo" for instant access.

🗂️ GitHub: GitHub

Built for the Backblaze Generative Media Hackathon.

For any Hackathon information/inquiry about the project, please contact: [email protected]

Supernova: The future of autonomous marketing, built today.

Built With

  • ai
  • automation
  • b2
  • backblaze
  • css
  • elevenlabs
  • fastapi
  • framer
  • genblaze
  • generative
  • jina
  • learning
  • machine
  • marketing
  • motion
  • next.js
  • openrouter
  • pexels
  • python
  • railway
  • react
  • supabase
  • tailwind
  • typescript
  • vercel
Share this project:

Updates

Submission history