> ## Documentation Index
> Fetch the complete documentation index at: https://docs.glood.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Glood.AI's Recommendation Engine
> Inside Glood.AI: How our proprietary algorithms and dashboard controls create perfect recommendations
# Glood.AI's Recommendation Engine: AI Meets Merchant Control
Glood.AI's recommendation platform combines cutting-edge artificial intelligence with intuitive merchant controls, giving you the best of both worlds—automated intelligence and business control. Our engine processes millions of data points in real-time while respecting your merchandising strategies configured in the Glood Dashboard.
## Our AI Model Stack
### Primary Models
1. **OpenAI GPT-4**
* Natural language understanding for product descriptions
* Semantic similarity matching
* Context-aware recommendations
* Cross-lingual product matching
2. **Google Gemini Pro**
* Visual similarity analysis
* Multi-modal product understanding
* Style and aesthetic matching
* Image-based recommendations
3. **Glood Proprietary Models**
* Fine-tuned on 100M+ e-commerce transactions
* Optimized for conversion prediction
* Real-time collaborative filtering
* Sequential pattern mining
## Recommendation Types Overview
Individual-level personalization using deep learning
Association rule mining and purchase pattern analysis
Content-based and visual similarity matching
Real-time trend detection and momentum scoring
## Glood Dashboard Configuration
Before diving into the AI, let's understand how you control recommendations through the Glood Dashboard:
### Section Configuration
* **Create Sections**: Build recommendation sections with point-and-click interface
* **Choose Algorithm**: Select from 13+ recommendation types
* **Set Parameters**: Configure number of products, filtering rules, and display options
* **Preview Mode**: Test recommendations before deploying to your store
### Business Rules Setup
* **Product Exclusions**: Block specific products or collections
* **Price Boundaries**: Set min/max price ranges for recommendations
* **Inventory Thresholds**: Hide products below certain stock levels
* **Brand Controls**: Prioritize or exclude specific brands
## How Glood Selects the Right Model
The Glood engine automatically selects the optimal AI model based on your configuration:
```mermaid theme={null}
graph LR
A[Recommendation Request] --> B{Context Analysis}
B --> C[User History]
B --> D[Product Type]
B --> E[Page Context]
B --> F[Business Goals]
C --> G{Model Selection}
D --> G
E --> G
F --> G
G --> H[GPT-4
Semantic Understanding]
G --> I[Gemini
Visual Analysis]
G --> J[Glood Models
Behavioral Patterns]
H --> K[Hybrid Recommendation]
I --> K
J --> K
K --> L[Final Results]
```
## Data Signals We Analyze
### User Signals
* Browsing history and session data
* Purchase history and frequency
* Cart additions and abandonments
* Time spent on products
* Device and location context
### Product Signals
* Category and subcategory relationships
* Price points and discount patterns
* Inventory levels and availability
* Product attributes and specifications
* Visual features and style elements
### Behavioral Signals
* Click-through rates
* Conversion rates by segment
* Seasonal patterns
* Cross-category affinities
* Return and review data
## Model Training & Optimization
### Continuous Learning Pipeline
1. **Data Collection**
* Real-time event streaming from Shopify stores
* Aggregated behavioral patterns
* Purchase outcome tracking
2. **Feature Engineering**
* Automated feature extraction
* Temporal pattern encoding
* Cross-feature interactions
3. **Model Training**
* Daily retraining cycles
* A/B testing new algorithms
* Performance validation
4. **Deployment**
* Gradual rollout of improvements
* Real-time monitoring
* Automatic fallback mechanisms
## Performance Optimization
### Speed & Scalability
* **Caching Strategy**: Multi-level caching for instant responses
* **Edge Computing**: Recommendations served from global CDN
* **Batch Processing**: Pre-computed recommendations for common scenarios
* **Real-time Inference**: On-demand calculation for personalized results
### Quality Metrics
We continuously monitor and optimize for:
* **Relevance Score**: How well recommendations match user intent
* **Diversity Index**: Variety in recommended products
* **Novelty Factor**: Balance between familiar and new discoveries
* **Business Impact**: Revenue lift and conversion improvement
## Advanced Techniques
### Ensemble Methods
Combining multiple models for superior results:
* Weighted voting from different algorithms
* Contextual bandits for exploration vs exploitation
* Meta-learning for model selection
### Deep Learning Architectures
* Transformer models for sequence prediction
* Graph neural networks for relationship modeling
* Attention mechanisms for feature importance
### Reinforcement Learning
* Multi-armed bandits for recommendation optimization
* Contextual bandits for personalization
* Deep Q-learning for long-term value optimization
## Glood Platform Features
### Analytics Dashboard
Monitor your recommendation performance in real-time:
* **Conversion Tracking**: See which recommendations drive sales
* **CTR Analysis**: Click-through rates by section and algorithm
* **Revenue Attribution**: Direct revenue impact of each recommendation type
* **A/B Testing**: Built-in testing framework for optimization
### API Access
Integrate Glood recommendations anywhere:
```json theme={null}
GET https://api.glood.ai/v1/recommendations
{
"shop": "your-store.myshopify.com",
"type": "similar_products",
"product_id": "7234567890",
"limit": 4,
"filters": {
"in_stock": true,
"price_range": [10, 100]
}
}
```
### Customization Options
Every aspect of Glood recommendations can be customized:
* **Visual Templates**: Match your store's design perfectly
* **Responsive Layouts**: Optimized for mobile and desktop
* **Language Support**: Multi-language recommendations
* **Currency Handling**: Automatic currency conversion
## Explore Glood's Recommendation Algorithms
Dive deeper into how each Glood recommendation type works:
* [Glood Personalized For You →](/how-it-works/recommendations/personalized-for-you)
* [Glood Frequently Bought Together →](/how-it-works/recommendations/bought-together)
* [Glood Similar Products →](/how-it-works/recommendations/similar-products)
* [Glood Trending Products →](/how-it-works/recommendations/trending-products)