What This Workflow Does
This advanced n8n workflow creates an intelligent AI personal assistant powered by GPT-4o that integrates seamlessly with WhatsApp. The assistant combines conversational AI with persistent memory, document intelligence, and scheduling capabilities to provide a comprehensive automated assistant experience. It leverages vector embeddings, document processing, and structured data storage to maintain context across conversations and deliver personalized responses.
How It Works
The workflow operates through a sophisticated multi-stage process that combines real-time messaging with intelligent data processing:
- WhatsApp messages trigger the workflow via webhook integration
- Incoming messages are processed through the GPT-4o language model for intelligent interpretation
- PostgreSQL database stores conversation history and user information for persistent memory
- Vector embeddings convert documents and messages into searchable semantic representations using OpenAI embeddings
- Supabase vector store enables intelligent document retrieval and context matching
- The agent node orchestrates complex multi-step tasks combining multiple tools
- Scheduling and wait nodes handle time-based actions and follow-ups
- Redis manages temporary data and session state for improved performance
- Responses are formatted and sent back to WhatsApp with contextual accuracy
Use Cases
- Customer Support Assistant: Automatically respond to customer inquiries on WhatsApp with contextual knowledge from stored documentation and conversation history
- Personal Productivity Assistant: Schedule reminders, manage tasks, and retrieve information from uploaded documents through natural conversation
- Knowledge Base Query System: Allow users to ask questions about uploaded documents and receive accurate answers powered by vector embeddings and semantic search
- Appointment Scheduling Bot: Handle meeting requests, check availability, and send confirmation reminders via WhatsApp
- Research Assistant: Process multiple documents, extract relevant information, and provide synthesized summaries based on user queries
Nodes Used
- @n8n/lmChatOpenAi: Primary language model for conversational responses
- @n8n/openAi: Additional OpenAI capabilities for text processing
- @n8n/embeddingsOpenAi: Converts text into vector embeddings for semantic search
- @n8n/textSplitterRecursiveCharacterTextSplitter: Chunks documents for optimal processing
- @n8n/documentDefaultDataLoader: Loads and processes uploaded documents
- @n8n/vectorStoreSupabase: Manages vector storage and semantic search
- @n8n/memoryPostgresChat: Maintains conversation memory in PostgreSQL
- @n8n/agent: Orchestrates complex multi-tool workflows
- @n8n/toolWorkflow: Enables workflow chaining and modular actions
- @n8n/toolVectorStore: Provides vector search capabilities to the agent
- @n8n/chainLlm: Chains multiple language model operations
- postgres: Primary database for storing user data and conversation history
- supabase: Vector database for semantic search and document retrieval
- redis: In-memory data store for session management and caching
- webhook: Receives WhatsApp messages and triggers workflow execution
- convertToFile: Transforms data into file format for document processing
- extractFromFile: Pulls structured data from uploaded documents
- set: Initializes and manages workflow variables
- if: Implements conditional logic for branching operations
- switch: Routes data based on multiple conditions
- wait: Introduces delays for scheduled actions
- aggregate: Combines multiple data streams into unified output
- merge: Combines workflow branches and data sources
- executeWorkflow: Triggers child workflows for modular task execution
- stickyNote: Provides workflow documentation and notes
Prerequisites
- Active n8n instance (self-hosted or cloud)
- OpenAI API key with access to GPT-4o and embedding models
- WhatsApp Business Account with API access or WhatsApp integration configured
- PostgreSQL database for storing conversation history and user data
- Supabase account with vector database capabilities enabled
- Redis instance for session and cache management
- Webhook URL configured to receive WhatsApp messages
- Document files prepared for knowledge base ingestion
Difficulty Level
Advanced. This workflow requires comprehensive understanding of AI/ML concepts including vector embeddings, semantic search, and conversation memory management. Setup involves configuring multiple external services, database schemas, and API integrations. Users should be comfortable with database administration, API authentication, and prompt engineering to fully customize the assistant behavior.
This workflow template is shared under the n8n fair-code license. Free to use and modify.
Leave a Reply