Turn recent customer reviews into clear insights with little manual work. The flow collects Trustpilot pages for one company, groups similar comments, and writes the key themes to a Google Sheet. It is a fast way for marketing and product teams to see what customers praise or complain about.
It starts fresh by clearing old entries in the vector store so you analyze clean data. The HTTP Request and HTML nodes pull and parse the latest Trustpilot reviews. Text is split, embedded with OpenAI, and stored in Qdrant with useful metadata like author, rating, and date. A subworkflow then loads reviews by date range, runs a K means clustering step, filters out small groups, fetches the full review payloads, and uses an OpenAI chat model to write short insights for each cluster. The results are formatted and appended to Google Sheets for easy sharing.
You will set the company domain, date window, and sheet details, then test the run to confirm points are saved and insights appear. Teams can track monthly themes, compare time periods, and cut analysis time from hours to minutes. This setup works well for brands that monitor voice of customer, campaign feedback, and common product issues at scale.