​

How Glood.AI Works

​

System Architecture Overview

​

Data Flow Process

​

1. Data Collection from Shopify

• Storefront Interactions: Every click, view, and action is tracked through our Web Pixel
• Purchase History: Complete order data including products, quantities, and customer segments
• Product Catalog: Real-time sync of your entire product database including variants, collections, and metadata
• Customer Profiles: Behavioral patterns, preferences, and purchase history

​

2. Intelligent Processing

• Real-time Analysis: Events are processed within milliseconds for immediate personalization
• Pattern Recognition: Identifying shopping behaviors, seasonal trends, and product affinities
• Segmentation: Automatic customer grouping based on behavior and preferences
• Inventory Awareness: Recommendations consider stock levels and availability

​

3. AI Model Selection

• OpenAI GPT-4: For complex natural language understanding and semantic product matching
• Google Gemini: For multi-modal analysis combining text and image data
• Glood Fine-tuned Models: Specialized models trained on e-commerce data for specific recommendation types

​

4. Recommendation Generation

• Context-Aware: Considers current page, time of day, device type, and session history
• Multi-Algorithm Approach: Combines collaborative filtering, content-based filtering, and deep learning
• Business Rules: Respects your merchandising rules, exclusions, and promotional priorities
• Performance Optimization: A/B testing and continuous learning improve recommendations over time

​

5. Real-time Ranking & Delivery

• User Context Integration: Current session behavior, cart contents, and browsing patterns instantly affect ranking
• Dynamic Re-ranking: Products are re-ordered based on individual user preferences and likelihood to convert
• Personalized Scoring: Each product gets a unique relevance score for each user
• Instant Adaptation: Rankings update in real-time as users interact with your store

​

Key Technologies

​

The AI Advantage

​

Why Multiple AI Models?

1. OpenAI GPT-4
   • Understands complex product descriptions and relationships
   • Excels at semantic matching and context understanding
   • Powers natural language search and filtering
2. Google Gemini
   • Analyzes product images alongside text descriptions
   • Identifies visual similarities and style matching
   • Enhances fashion and design-focused recommendations
3. Glood Fine-tuned Models
   • Optimized specifically for e-commerce patterns
   • Trained on millions of successful purchase combinations
   • Provides the fastest inference for real-time recommendations

​

Manual Recommendations & Merchandising Control

​

Dashboard Configuration

​

Manual Configuration Methods

​

1. 1:1 Product Relationships

• Visual Dashboard Interface: Drag-and-drop product associations
• Specific Mappings: Define exact products to recommend together
• Override AI: Force specific recommendations regardless of AI scoring
• Test Mode: Preview recommendations before going live

​

2. Bulk CSV Upload

• Large-Scale Configuration: Upload thousands of product relationships at once
• Template Support: Pre-formatted CSV templates for easy setup
• Validation: Automatic error checking and product ID verification
• Scheduled Updates: Set up recurring imports for seasonal changes

​

3. Attribute Prioritization

• Brand Weighting: Prioritize same-brand or cross-brand recommendations
• Price Range Control: Keep recommendations within specific price bands
• Category Rules: Define cross-category recommendation strategies
• Inventory Priority: Boost products with high stock levels
• Margin Optimization: Favor higher-margin products

​

Real-Time Personalization Layer

final_recommendation = (
    manual_weight * manual_rules +
    ai_weight * ai_predictions +
    personalization_weight * user_preferences
)

​

Dynamic Ranking Features

• User Behavior Adaptation: Manual rules adjust based on individual user patterns
• Session Context: Real-time ranking considers current browsing session
• A/B Testing: Test manual vs AI recommendations performance
• Fallback Strategies: Automatic fallback to AI when manual rules don’t apply

​

Common Use Cases

​

Inventory Management

• Push overstocked items through manual boosting
• Hide out-of-stock products from recommendations
• Promote end-of-season clearance items

​

Brand Partnerships

• Feature partner brands in specific recommendation slots
• Create co-marketing recommendation campaigns
• Implement exclusive brand showcases

​

New Product Launches

• Manually insert new products into recommendation streams
• Override lack of historical data for new items
• Create launch-specific recommendation strategies

​

Seasonal Campaigns

• Configure holiday-specific product bundles
• Adjust recommendations for seasonal events
• Schedule automatic rule changes for campaigns

​

Integration with AI

1. AI Baseline: Machine learning provides the foundation
2. Manual Override: Specific rules take precedence when defined
3. Hybrid Scoring: Combine manual and AI signals for optimal results
4. Performance Tracking: Monitor manual vs AI recommendation performance

​

Data Privacy & Security

• GDPR & CCPA Compliant: Full compliance with data protection regulations
• Encrypted Data Transfer: All data transmitted using industry-standard encryption
• Shopify Standards: Follows all Shopify partner requirements and best practices

​

Performance Metrics

• 40% Average AOV Increase: Through intelligent upselling and cross-selling
• 25% Higher Conversion Rates: By showing the right products at the right time
• 60% Click-through Rate: On personalized recommendation sections
• < 100ms Response Time: For instant recommendation loading

​

Integration Points

1. Theme Integration: Native Liquid templates for seamless design matching
2. Checkout Extensions: Upsell opportunities at critical conversion points
3. API Access: Headless commerce and custom implementations
4. Admin Dashboard: Real-time analytics and configuration

​

Next Steps