ReBridge — AI-Powered Circular Commerce Platform

Team lazyBot — Kavya Rai & Oindrila Singha

🔗 Live: https://d12xi8surv8so8.cloudfront.net
📂 GitHub: github.com/Vya234/ReBridge

One Line Pitch

ReBridge is an AI-powered circular commerce platform that gives every returned Amazon product a second life — graded, priced, routed, and resold in seconds.

From Return to Resale: The Circular Commerce Loop

The complete ReBridge lifecycle, from AI-powered product evaluation to trusted resale and sustainability impact.

Problem Statement

Problem	Impact
Amazon loses $28B annually to returns	Massive cost with zero value recovery
17–30% of e-commerce orders returned	1 in 5 purchases comes back
Most items go to landfill	Environmental waste at scale
Customers distrust refurbished products	Trust gap kills resale
Sellers inspect 200+ returns/month manually	Hours wasted on repetitive grading
P2P resale feels unsafe	No trust layer, no quality guarantee

Solution — The Intelligent Bridge

Return initiated
  → AI grades in <2 seconds (Nova Micro)
    → Routes: Resell / Refurbish / Donate / Recycle
      → Health Card + Suggested Price generated
        → Listed on certified refurb store with NLP search
          → Buyer negotiates via P2P real-time chat
            → Green Credits earned for sustainability
              → Pre-purchase risk check prevents future returns

Features (22)

Core AI

1. AI Grading & Value Recovery Routing Engine

Powered by Amazon Nova Lite (multimodal) via Bedrock converse() API
Acts as a Value Recovery Optimizer — calculates a dynamic recovery_score (0–100) to choose the optimal route based on value retention and market demand
Grades: A (Excellent), B (Good), C (Fair), D (Poor)
Routes intelligently: Resell / Refurbish / Donate / Recycle — driven by recovery potential, not rigid rules
Sub-2-second evaluation with confidence scores
Custom rules: warranty upgrades recovery score, missing accessories ≥ B, packaging doesn't affect grade
Calculates sustainability metrics per transaction: kg CO₂ prevented, liters water saved

2. Product Health Card

Trust breakdown: Appearance, Functional, Packaging scores (0–100%)
Optimal Value Recovery display with recovery_score percentage
Sustainability Impact block: 🌱 kg CO₂ Prevented + 💧 Liters Water Saved
Confidence score with visual progress bar
Condition summary (AI-generated one-sentence)
Route journey tracker (visual pipeline)
Animated rubber-stamp grade seal

3. ReBridge ID Auto-Generation

Format: RB-YYYY-XXXXXX (e.g. RB-2026-004521)
Auto-generates if item_id left blank
Prominently displayed on result page with copy button
Used for re-evaluation and QR deep-linking

4. Smart Condition Dropdowns

Category-specific condition options:
- Electronics: Fully functional, Minor scratches, Screen cracked, Battery issues, Water damage
- Clothing: Good condition, Faded, Missing buttons, Torn, Stains
- Books: Good condition, Highlighted, Torn pages, Cover damaged
- Home: Light wear, Missing pieces, Motor broken, Shattered/Destroyed
"Other" option with custom text input
Auto-fills image label based on condition

5. Extended Form Fields

Return Reason: Changed mind / Defective / Wrong item / Not as described / Better price found / Gift return
Warranty Remaining: No warranty → More than 1 year
Repair History: Never repaired → Repaired multiple times
All fed into AI prompt for smarter grading

6. AI Resell Price Suggestion

User enters original price (₹)
AI calculates: Grade A=80%, B=60%, C=30%, D=0%
Suggested price shown on health card and shop listings
Savings percentage displayed vs new price

7. Re-evaluate Feature

Toggle: "Re-evaluating a previously graded item?"
Enter existing ReBridge ID
Fetches old record, submits new evaluation
Shows comparison banner: "Grade changed from B → C" or "No change detected"
Accessible from Product Gallery and My Evaluations via "↻ Re-evaluate" button

Shopping & Discovery

8. Shop Certified Refurbished

Browse AI-graded inventory by category
Only surfaces Resell/Refurbish items
Product cards: grade stamp, suggested price, trust breakdown mini-bars
"Chat with Seller" button for Resell items

9. Natural Language Search

NLP-powered via Amazon Nova
Examples: "Grade A electronics under ₹20000", "refurbished books", "fully functional laptop"
Extracts: category, grade, route, max_price, min_functional, min_appearance, keywords
Search hint chips for quick exploration
Falls back to keyword matching if AI extraction fails

10. P2P Real-Time Negotiation Chat

WebSocket API Gateway with 3 Lambda functions
"Negotiate with Buyer" button on Resell (Grade A) items
Slide-up chat panel with item summary
Real-time message broadcast via GSI query
Optimistic message display (zero latency UX)
Accessible from Shop page via "Chat with Seller"

Prevention & Intelligence

11. AI Purchase Advisor

Upgraded pre-purchase intelligence replacing binary risk warnings with deep compatibility analysis
Evaluates free-text purchase intent against product category via dedicated Bedrock prompt
Generates a 0–100 Compatibility Score (green ≥80, orange 50–79, red <50)
Outputs personalized reasoning (2–3 bullet points)
Strict compatibility checklist with Pass ✅ / Warn ⚠️ / Fail ❌ status per criterion
Explicitly identifies potential return risks
Delivers actionable final recommendation (e.g. "Verify software compatibility", "Shop Refurbished instead")
Premium UI matching the Product Health Card layout — form hides on result, full-screen focus
"Shop Refurbished Instead" CTA when score is below 50

Gallery & History

12. Product Gallery

Public view of ALL evaluated items
Smart client-side search with structured query parsing
Supports: "Electronics Grade A", "Refurbish", combined filters
Search hint chips: Electronics, Grade A, Refurbish, Clothing Grade B, Donate
Expandable rows with trust breakdown + QR code
Re-evaluate button in expanded view

13. My Evaluations

Personalised dashboard filtered by current user_id
Same table design with expandable rows
QR code + Re-evaluate in expanded view
Empty state: "You haven't evaluated any items yet"

Incentives & Identity

14. Green Credits Wallet

Per-user persistent credits: Donate=50, Refurbish=30, Resell=20, Recycle=10
Green pill in navbar: 🌱 {credits} credits
Celebration animation: "+X Green Credits Earned!" on result page
Accumulates across sessions via DynamoDB

15. Per-User Identity with Duplicate Detection

Welcome modal on first visit — no login required
Enter name/email → normalized as user_id
Checks GreenWallet for existing account
"Is this you?" confirmation if account found with credits
"No, I'm someone else" → appends random 4 digits
"Switch User" in account dropdown menu

16. QR Code + Share Health Card

QR on every health card: links to https://d12xi8surv8so8.cloudfront.net?item={id}
Deep-link: scanning QR or visiting URL loads that item's health card directly
"Share Health Card" button copies URL to clipboard
QR also shown in Product Gallery and My Evaluations expanded rows

17. Seller Bulk Upload Portal

CSV drag-and-drop upload
Format: item_id, category, condition_notes, simulated_image_label
Preview table before processing
Batch AI grading (concurrent, batches of 3)
Live progress bar
Results table with grade + route as they complete
Summary: total items, route breakdown, green credits earned
Download results CSV button

18. Location-Based Seller Discovery

Sellers optionally provide City (required) and Locality (optional) when submitting a return
Buyers can filter Shop and Product Gallery by city (case-insensitive partial match)
Matched products display a 📍 Locality, City tag on their card
Connects buyers to nearby sellers who can then negotiate via the existing P2P chat
No geo libraries needed — lightweight text-based matching

19. Real Image Upload + AI Vision Analysis

Upload product photos directly from the evaluation form (drag-and-drop or click)
Accepted formats: JPG, JPEG, PNG, WebP
Max file size: 5MB — shows error for larger files
Auto-compression: Images resized to max 1920px and compressed to JPEG quality 0.8 (most photos under 1MB)
Shows "Compressing image..." loading state and "Photo ready (245 KB)" after processing
Images analyzed by Amazon Nova Lite (multimodal vision model) via Bedrock converse() API
AI uses the photo as primary evidence for appearance scoring
Photos stored permanently in S3 (rebridge-product-images bucket)
Image URL saved to DynamoDB and displayed across all pages
Product photos visible in: Result page, Product Gallery, My Evaluations, Shop cards, QR deep-links

20. Similar Item Price Intelligence

After grading, scans DynamoDB for items with same category + same grade + suggested_price > 0
Calculates price range from up to 10 similar items
Displays: "📊 Similar Grade B Electronics: ₹2,900–₹3,300 (based on 5 items)"
Helps sellers price competitively based on market data

21. S3 Persistent Image Storage

Dedicated S3 bucket with public read access for product photos
Images uploaded as {item_id}.jpg — permanent, shareable URLs
Photos render in Shop cards, Product Gallery expanded rows, and QR-linked health cards
Supports the full product lifecycle from grading to resale

22. Seller Return Analytics Dashboard

Intelligence tool integrated into the Seller Portal
Tracks top return reasons with visual percentage breakdown (Changed Mind 42%, Wrong Size 23%, Defective 18%)
Identifies the most returned category with distribution insights
Provides actionable AI suggestions to help sellers prevent future returns (e.g. "Improve sizing charts to reduce apparel returns by 15%")
Enables data-driven decision making for inventory and listing optimization

Future Development / Roadmap

Geo-Intelligent Local Exchange — Extend location discovery with coordinates, radius-based filtering (5km/10km/25km), haversine distance ranking, AI routing that factors local demand, carbon emission savings estimates, and community pickup hubs
Image Upload — Real visual grading via Amazon Rekognition
Seller Analytics Dashboard — Return patterns, grade distribution, revenue insights
Green Credits Redemption — Redeem credits for discounts or donate to environmental causes
Multi-language Support — Expand trust cards to global marketplaces
Enhanced Bulk Upload Vision — Future iterations of the Seller Bulk Upload Portal will support processing real product images at scale, allowing sellers to provide image URLs directly within the upload CSV or upload a bundled ZIP file of product photos for batch AI vision analysis

Amazon Problem Statement Coverage

Amazon Requirement	ReBridge Features
AI deciding resell/refurbish/donate/recycle	AI Routing Engine + Extended Fields + Warranty Rules
Smart quality grading	Product Health Card + Appearance/Functional/Packaging Scores
Personalized refurb recommendations	Shop Certified Refurbished + NLP Search + Price Suggestion
Sustainable incentives	Green Credits Wallet + Celebration Animations
Easy P2P resale	Real-time WebSocket Chat + AI Price Suggestion
Predictive return prevention	AI Purchase Advisor

Tech Stack

Layer	Technology	Purpose
Frontend	React 18.3 + Vite 6 + Tailwind CSS 3.4	SPA with editorial design system
Hosting	AWS S3 + CloudFront (HTTPS)	Global CDN with custom domain
Backend	AWS Lambda (Python 3.12) × 9 functions	Serverless compute
HTTP API	Amazon API Gateway (HTTP API)	REST endpoints + CORS
WebSocket API	Amazon API Gateway (WebSocket)	Real-time P2P chat
AI/ML	Amazon Bedrock — Nova Lite (multimodal vision) + Nova Micro (text)	Grading with image analysis, NLP search, Purchase Advisor
Database	DynamoDB (3 tables, on-demand)	Zero-config storage
Storage	Amazon S3 (`rebridge-product-images`)	Persistent product photo storage
QR	qrcode.react + api.qrserver.com	Scannable health card verification
Dev Tooling	Kiro IDE (spec-driven)	AI-assisted implementation

Architecture

┌───────────────────────────────────────────────────────────────────────────┐
│                  React Frontend (S3 + CloudFront HTTPS)                   │
│  Home • Evaluate • Result • Shop • Gallery • My Evals • Seller • Chat     │
└──────────────────────┬─────────────────────────────────┬──────────────────┘
                       │ HTTPS                           │ WSS
                       ▼                                 ▼
┌──────────────────────────────────────────┐  ┌────────────────────────────┐
│        HTTP API Gateway (REST)           │  │  WebSocket API Gateway     │
│  CORS: * │ Auto-deploy │ ap-south-1      │  │Route: $request.body.action │
└──┬─────┬──────┬──────┬────┬──────┬───────┘  └┬─────┬───────┬─────────────┘
   │     │      │      │    │      │           |     │       │
   ▼     ▼      ▼      ▼    ▼      ▼           ▼     ▼       ▼
┌─────┐┌────┐┌─────┐┌────┐┌────┐┌──────┐   ┌─────┐┌──────┐┌─────┐
│grade││get_││get_ ││get_││get ││natur-│   │ws_  ││ws_   ││ws_  │
│_and_││hlth││items││all_││wall││al_   │   │conn ││disco ││msg  │
│route││card││_cat ││item││et  ││search│   │ect  ││nnect ││     │
└──┬──┘└──┬─┘└──┬──┘└──┬─┘└──┬─┘└──┬───┘   └──┬──┘└──┬───┘└──┬──┘
   │      │     │      │     │     │          │      │       │
   ▼      │     │      │     │     ▼          │      │       │
┌───────┐ │     │      │     │  ┌───────┐     │      │       │
│Bedrock│ │     │      │     │  │Bedrock│     │      │       │
│(Nova) │ │     │      │     │  │(Nova) │     │      │       │
└──┬────┘ │     │      │     │  └──┬────┘     │      │       │
   │      ▼     ▼      ▼     ▼     │          ▼      ▼       ▼
   │  ┌─────────────────────────────────────────────────────────────┐
   └─►│                     Amazon DynamoDB                         │
      │  ┌──────────────────┐ ┌─────────────┐ ┌─────────────────┐   │
      │  │ProductHealthCards│ │GreenWallet  │ │ChatConnections  │   │
      │  │PK: item_id       │ │PK: user_id  │ │PK: connectionId │   │ 
      │  │Fields: grade,    │ │total_credits│ │GSI: itemId-index│   │
      │  │route, price,     │ │             │ │sender, itemId   │   │
      │  │trust, image_url  │ │             │ │                 │   │
      │  └──────────────────┘ └─────────────┘ └─────────────────┘   │
      └─────────────────────────────────────────────────────────────┘

      ┌─────────────────────────────────────────────────────────────┐
      │                     Amazon S3                               │
      │  Bucket: rebridge-product-images                            │
      │  Pattern: {item_id}.jpg │ Public read │ ap-south-1          │
      └─────────────────────────────────────────────────────────────┘

API Endpoints

HTTP API

Method	Endpoint	Description
`POST`	`/evaluate-return`	AI grading + routing + pricing + wallet update
`GET`	`/health-card/{item_id}`	Retrieve stored Product Health Card
`GET`	`/items/{category}`	Browse by category (Resell/Refurbish only)
`GET`	`/items`	All items sorted by timestamp (newest first)
`GET`	`/wallet/{user_id}`	Green Credits balance
`POST`	`/search`	Natural language search with AI filter extraction

WebSocket API

Route	Handler	Description
`$connect`	`ws_connect`	Save connectionId + itemId via query params
`$disconnect`	`ws_disconnect`	Remove connection
`sendMessage`	`ws_message`	GSI query → broadcast to all item connections

POST /evaluate-return

Request:

{
  "item_id": "",
  "category": "Electronics",
  "condition_notes": "Minor scratches",
  "simulated_image_label": "light_scratches",
  "original_price": 2999,
  "return_reason": "Changed mind",
  "warranty_left": "6-12 months",
  "repair_history": "Never repaired",
  "user_id": "kavya"
}

Response:

{
  "item_id": "RB-2026-004521",
  "user_id": "kavya",
  "category": "Electronics",
  "grade": "A",
  "condition_summary": "Device in excellent condition with minor surface wear",
  "assigned_route": "Resell",
  "confidence_score": 0.92,
  "trust_breakdown": { "appearance": 0.85, "functional": 0.98, "packaging": 0.80 },
  "green_credits": 20,
  "original_price": 2999,
  "suggested_price": 2399,
  "return_reason": "Changed mind",
  "warranty_left": "6-12 months",
  "repair_history": "Never repaired",
  "timestamp": "2026-06-14T10:30:00+00:00"
}

POST /search

Request:

{ "query": "Grade A electronics under 5000" }

Response:

{
  "items": [...],
  "count": 3,
  "filters_applied": { "category": "Electronics", "grade": "A", "max_price": 5000, "keywords": [] }
}

Project Structure

ReBridge/
├── specs/
│   ├── 1-trust-database.md
│   ├── 2-ai-routing-engine.md
│   └── 3-api-gateway.md
├── infrastructure/
│   ├── create_table.py                  # ProductHealthCards table
│   ├── deploy_items_endpoint.py         # GET /items/{category}
│   ├── deploy_wallet.py                 # GreenWallet + GET /wallet
│   └── deploy_websocket.py             # WebSocket API + Chat Lambdas
├── lambdas/
│   ├── shared/
│   │   └── db_client.py                 # DynamoDB helper (put/get)
│   ├── grade_and_route/
│   │   └── handler.py                   # AI grading + routing + pricing + credits
│   ├── get_health_card/
│   │   └── handler.py                   # Health card retrieval
│   ├── get_items_by_category/
│   │   └── handler.py                   # Category browse
│   ├── get_all_items/
│   │   └── handler.py                   # All items (Gallery)
│   ├── get_wallet/
│   │   └── handler.py                   # Green Credits balance
│   ├── natural_search/
│   │   └── handler.py                   # NLP search via Nova
│   └── websocket/
│       ├── connect.py                   # $connect
│       ├── disconnect.py                # $disconnect
│       └── message.py                   # sendMessage broadcast
├── frontend/
│   ├── src/
│   │   ├── components/
│   │   │   ├── HomePage.jsx             # Hero + Check Before You Buy
│   │   │   ├── EvaluatePage.jsx         # Smart form + extended fields
│   │   │   ├── ResultPage.jsx           # Health Card + QR + Price + Chat
│   │   │   ├── ShopPage.jsx             # Refurb store + NLP search + chips
│   │   │   ├── HistoryPage.jsx          # Product Gallery + smart search
│   │   │   ├── MyEvaluations.jsx        # Personal dashboard
│   │   │   ├── SellerPortal.jsx         # Bulk CSV upload
│   │   │   ├── ChatPortal.jsx           # WebSocket chat panel
│   │   │   └── WelcomeModal.jsx         # Identity + duplicate detection
│   │   ├── App.jsx                      # Router + state + account dropdown
│   │   ├── main.jsx
│   │   └── index.css                    # Tailwind + editorial system
│   ├── index.html
│   ├── package.json
│   ├── vite.config.js
│   ├── tailwind.config.js
│   └── postcss.config.js
├── deploy.py                            # One-click backend deployment
├── sample_bulk_upload.csv               # Sample CSV for testing
├── requirements.txt
├── PITCH.md
└── README.md

Setup & Deployment

Prerequisites

Python 3.12+, Node.js 18+
AWS CLI configured (default profile)
Bedrock model access: Nova Lite + Nova Micro in ap-south-1

Backend

pip install -r requirements.txt
python deploy.py                              # Core: DynamoDB + Lambdas + HTTP API
python infrastructure/deploy_items_endpoint.py # GET /items/{category}
python infrastructure/deploy_wallet.py        # GreenWallet + GET /wallet
python infrastructure/deploy_websocket.py     # WebSocket chat system

Frontend

cd frontend && npm install && npm run build
aws s3 sync dist/ s3://rebridge-app-2026 --delete
aws cloudfront create-invalidation --distribution-id YOUR_ID --paths "/*"

Environment

Config	Value
Region	ap-south-1 (Mumbai)
Bedrock Model	apac.amazon.nova-lite-v1:0
HTTP API	s3r8aqjg75.execute-api.ap-south-1.amazonaws.com
WebSocket	kvhtc0th50.execute-api.ap-south-1.amazonaws.com/production
CloudFront	d12xi8surv8so8.cloudfront.net

ROI for Amazon

Metric	Impact
20% return cost recovery	$5.6B saved annually
New refurb GMV stream	Revenue at lower COGS
P2P commission revenue	New transaction fee stream
Green Credits → loyalty	Higher customer LTV
Sustainability story	Brand value with conscious consumers
Seller efficiency	Bulk processing eliminates manual inspection

Hackathon Context


Event	Amazon HackOn Season 6
Theme	Second Life Commerce
Team	lazyBot (Kavya Rai, Oindrila Singha)
Timeline	48 hours
Dev	Kiro spec-driven IDE
Live	d12xi8surv8so8.cloudfront.net
GitHub	github.com/Vya234/ReBridge

Name		Name	Last commit message	Last commit date
Latest commit History 43 Commits
.vscode		.vscode
api_gateway		api_gateway
assets		assets
frontend		frontend
infrastructure		infrastructure
lambdas		lambdas
specs		specs
PITCH.md		PITCH.md
README.md		README.md
deploy.py		deploy.py
requirements.txt		requirements.txt
response.json		response.json
sample_bulk_upload.csv		sample_bulk_upload.csv
test_nova.py		test_nova.py
ws_policy.json		ws_policy.json

Folders and files

Latest commit

History

Repository files navigation

ReBridge — AI-Powered Circular Commerce Platform

One Line Pitch

From Return to Resale: The Circular Commerce Loop

Problem Statement

Solution — The Intelligent Bridge

Features (22)

Core AI

1. AI Grading & Value Recovery Routing Engine

2. Product Health Card

3. ReBridge ID Auto-Generation

4. Smart Condition Dropdowns

5. Extended Form Fields

6. AI Resell Price Suggestion

7. Re-evaluate Feature

Shopping & Discovery

8. Shop Certified Refurbished

9. Natural Language Search

10. P2P Real-Time Negotiation Chat

Prevention & Intelligence

11. AI Purchase Advisor

Gallery & History

12. Product Gallery

13. My Evaluations

Incentives & Identity

14. Green Credits Wallet

15. Per-User Identity with Duplicate Detection

16. QR Code + Share Health Card

17. Seller Bulk Upload Portal

18. Location-Based Seller Discovery

19. Real Image Upload + AI Vision Analysis

20. Similar Item Price Intelligence

21. S3 Persistent Image Storage

22. Seller Return Analytics Dashboard

Future Development / Roadmap

Amazon Problem Statement Coverage

Tech Stack

Architecture

API Endpoints

HTTP API

WebSocket API

POST /evaluate-return

POST /search

Project Structure

Setup & Deployment

Prerequisites

Backend

Frontend

Environment

ROI for Amazon

Hackathon Context

License

About

Topics

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages