Summarize Customer Emails Using AI and RAG

Overview

This n8n workflow is designed to automatically summarize incoming customer emails. It utilizes a Retrieval Augmented Generation (RAG) approach, which combines the power of a language model with a knowledge retrieval system.

1. Webhook Trigger: Receives incoming customer emails.
2. Text Splitter: Breaks down the email content into manageable chunks.
3. Embeddings: Converts these text chunks into vector embeddings using Cohere.
4. Weaviate Insert: Stores these embeddings in a Weaviate vector database for efficient retrieval.
5. Weaviate Query: When a new email arrives, it queries Weaviate to find similar past email embeddings, providing context.
6. Vector Tool: Processes the retrieved vector information.
7. Window Memory: Maintains a history of interactions for context.
8. Chat Model: An Anthropic chat model processes the query and retrieved context.
9. RAG Agent: The core AI component that orchestrates the retrieval and generation of the summary.
10. Append Sheet: Logs the summarization status to a Google Sheet.
11. Slack Alert: Sends an alert to Slack in case of any errors.

This workflow is ideal for quickly digesting large volumes of customer communication, identifying key issues, and improving response times.