This workflow automates the process of extracting, analyzing, and summarizing Trustpilot reviews for a given company. It scrapes reviews from Trustpilot, stores them in a Qdrant vector database, applies a clustering algorithm to group similar feedback, and uses an LLM (OpenAI) to generate actionable insights. Finally, it exports the summarized insights, sentiments, and suggested improvements into a Google Sheet.
Key Features:
- Automated Review Collection – Scrapes recent Trustpilot reviews for a specified company.
- Vector Database Integration – Stores review embeddings in Qdrant for powerful similarity searches.
- Clustering Analysis – Uses K-means clustering to group similar reviews, identifying common themes and issues.
- LLM-Powered Insights – Generates summaries, sentiments, and improvement suggestions from grouped reviews using OpenAI.
- Date-Scoped Analysis – Can limit insights to specific time ranges (e.g., last month).
- Google Sheets Export – Outputs insights, sentiment analysis, and raw clustered review data into a Google Sheet for reporting and sharing.
- Reusable Subworkflow – Includes a triggerable subworkflow for targeted analysis on demand.
