Automated Monthly Expense Report Processing Workflow

This n8n workflow automates the entire process of managing monthly expense reports, from receiving raw data to updating Google Sheets and handling potential errors. The workflow begins with a webhook trigger that captures expense data submissions. It then splits the text into manageable chunks using a LangChain Text Splitter, making it suitable for embedding generation.

Next, the workflow uses OpenAI’s embedding model to convert text chunks into vector representations, which are stored in a Weaviate vector database. This setup facilitates efficient retrieval of relevant data. When a query is made, the workflow retrieves similar vectors from Weaviate and feeds this context into a LangChain Chat Model (using Anthropic API) to generate responses or process data.

A Retrieval-Augmented Generation (RAG) agent manages the interaction, handling data input, context, and response generation. The final step involves appending the processed report status to a Google Sheet for logging purposes. If any errors occur during processing, a Slack notification is sent to alert the team.

This workflow is ideal for organizations looking to automate monthly expense report analysis, ensure centralized data logging, and get real-time error notifications, significantly reducing manual effort and enhancing data accuracy.