24 Track Satisfaction Scores

## 🤖 **Section 2: Scrape Reviews Using AI Agent**

### 🧩 Nodes:

* `🧠 Scrape Reviews with Agent (MCP)` *(AI Agent)*

  * 🔌 `🎯 Prompt & Guide Agent` *(OpenAI Chat Model)*
  * 🌐 `Execute Web Scrape (Bright Data)` *(MCP Client)*
  * 📦 `Parse Agent Response` *(Structured Output Parser)*

### 📝 What This Does:

This is where the **magic** happens — using AI + Bright Data to fetch live reviews.

* The **AI Agent** launches a **headless browser** via Bright Data's Mobile Carrier Proxy (MCP), mimicking a real user to visit the review site.
* It’s guided by an OpenAI model that understands natural prompts, like:
  *“Extract customer ratings, comments, and dates from Trustpilot”.*
* After scraping, the **Structured Output Parser** turns unstructured HTML into clean JSON data — a list of reviews like:

  ```json
  { "rating": 2, "comment": "..." }
  ```

### 🧠 Why It Matters:

* Many review sites use JavaScript — Bright Data MCP bypasses anti-bot detection.
* No need to write complex scraping code — just describe what you want in plain English.
* This is your **data foundation** — reliable, structured, fresh.

---