This n8n workflow automates the process of categorizing support tickets using AI and evaluating the accuracy of the classifications. It starts with a webhook trigger that receives support tickets via HTTP request. The workflow then reads a dataset of support tickets from a Google Spreadsheet and uses an AI agent powered by a large language model (GPT-4) to classify each ticket based on its subject and body content.
The AI agent is configured with system instructions to identify the category and priority level of the support query, returning the results in JSON format. The workflow compares the AI’s classifications with the expected categories in the dataset, checking for matches in both category and priority.
The results of the comparison are evaluated and metrics are set based on the matching accuracy. These metrics can be used to assess the performance of the AI classification over time. The process includes steps for parsing the output, checking evaluation status, and recording key metrics, enabling continuous monitoring and improvement.
This workflow is practical for teams wanting to automate support ticket triage, improve classification accuracy, and quantify AI performance through metrics, all integrated within their support systems or customer service platforms.
Reviews
There are no reviews yet.