Collect recent online reviews for a chosen company, group common themes, and send clear insights to a sheet your team can use. Marketing and product teams can spot patterns fast and decide what to fix or promote.
The run starts by clearing old records for that company in the vector database. It pulls a few review pages with an HTTP request and uses an HTML parser to grab author, rating, title, and text. Each review is turned into an embedding with OpenAI and stored in Qdrant with company tags. A linked subflow loads reviews in a date range, clusters similar items, keeps groups with at least three reviews, and fetches full details from Qdrant. A chat model then turns each cluster into short notes that explain what customers are saying. The flow formats the results and appends them to Google Sheets.
You need an OpenAI key, a Qdrant endpoint, and Google Sheets access. Set the company domain and the time window, then run a test. Expect to cut review analysis from hours to minutes and deliver consistent reports for voice of customer, churn clues, and product feedback.