Workflow Overview

This is an n8n automation workflow named "Weekly Shopify Sales Summary," designed to process and analyze Shopify sales data. The workflow integrates RAG (Retrieval-Augmented Generation) technology, vector database storage, and an AI chat model to deliver intelligent sales data processing and summarization capabilities.

Workflow Architecture

Trigger Mechanism

• Webhook Trigger: Activated via a POST request to the endpoint /weekly-shopify-sales-summary.
• Enables external systems or scheduled tasks to initiate the data processing pipeline through HTTP requests.

Core Processing Pipeline

1. Data Preprocessing Stage

Text Splitter

• Splits incoming data into smaller text chunks.
• Chunk Size: 400 characters
• Chunk Overlap: 40 characters
• Purpose: Optimizes subsequent vectorization and retrieval efficiency.

2. Vectorization and Storage

Embeddings Model

• Uses Cohere’s embed-english-v3.0 model.
• Converts text data into vector representations.
• Connects to the Cohere API for embedding generation.

Pinecone Vector Database

• Insert Mode (Pinecone Insert): Stores vectorized data in an index named weekly_shopify_sales_summary.
• Query Mode (Pinecone Query): Retrieves relevant vector data from the same index.
• Enables long-term data storage and semantic search functionality.

3. AI Intelligence Layer

RAG Agent

• System Prompt: "You are an assistant for Weekly Shopify Sales Summary."
• Orchestrates multiple AI components collaboratively:

Chat Model

• Utilizes Anthropic’s Claude AI model.
• Responsible for generating intelligent responses and performing data analysis.

Vector Tool

• Name: "Pinecone"
• Description: "Vector context"
• Provides the RAG agent with vector-based contextual retrieval capabilities.

Window Memory

• Maintains conversation history.
• Ensures contextual coherence across interactions.

4. Output and Monitoring

Google Sheets Integration (Append Sheet)

• Operation Type: Append data
• Target Document: Google Sheet with ID SHEET_ID
• Target Worksheet: Sheet named "Log"
• Logs processing status and results.

Error Handling (Slack Alert)

• Monitoring Channel: #alerts
• Automatically sends error notifications to Slack when workflow failures occur.
• Message Format: Weekly Shopify Sales Summary error: {error details}

Technology Stack Integration

Required API Credentials

1. Cohere API: For text embeddings
2. Pinecone API: For vector database storage
3. Anthropic API: For the Claude AI model
4. Google Sheets OAuth2: For data logging
5. Slack API: For error alerts

Connection Topology

Webhook → Text Splitter → Embeddings → Pinecone Insert
                                    ↓
                               Pinecone Query → Vector Tool
                                                      ↓
Window Memory ←────────────────────────────→ RAG Agent → Append Sheet
                                                      ↓
Chat Model ──────────────────────────────────┘   (Error) → Slack Alert

Workflow Features

Advantages

1. Intelligent Processing: Combines RAG and large language models for smart data analysis.
2. Scalability: Vector database supports large-scale data storage and fast retrieval.
3. Fault Tolerance: Comprehensive error handling and alerting system.
4. Data Persistence: Google Sheets logging ensures traceability.
5. Context Awareness: Window memory maintains conversational continuity.

Use Cases

• Automated sales data summarization
• Periodic sales trend analysis
• Intelligent Q&A system (for sales-related queries)
• Automatic sales report generation

Node Statistics

• Total Nodes: 12
• Trigger Nodes: 1 (Webhook)
• AI/ML Nodes: 7 (Text Splitter, Embeddings, 2 Pinecone nodes, Vector Tool, Memory, Chat Model)
• Integration Nodes: 2 (Google Sheets, Slack)
• Other Nodes: 2 (Sticky Note, RAG Agent)

Execution Settings

• Execution Order Version: v1
• Trigger Count: 1