Vision-Based AI Agent Scraper - with Google Sheets, ScrapingBee, and Gemini

For the Vision-Based AI Agent Scraper, automate data extraction from webpages using screenshots and HTML. This workflow integrates Google Sheets for managing URLs and storing results, ScrapingBee for capturing full-page screenshots, and the Gemini-1.5-Pro AI model for accurate data parsing. It efficiently converts HTML to Markdown, optimizing processing costs, and is designed for e-commerce scraping, ensuring structured data is easily accessible and customizable for various needs.

7/4/2025
29 nodes
Complex
manualcomplexlangchainsplitoutgooglesheetssticky noteexecuteworkflowtriggermarkdownadvancedapiintegration
Categories:
Complex WorkflowManual TriggeredData Processing & AnalysisBusiness Process Automation
Integrations:
LangChainSplitOutGoogleSheetsSticky NoteExecuteWorkflowTriggerMarkdown

Target Audience

This workflow is ideal for:
- E-commerce Businesses: Companies looking to gather product data from competitor websites for pricing analysis, inventory management, or market research.
- Data Analysts: Professionals who need to extract structured data from various online sources for reporting and analysis.
- Web Developers: Developers who want to automate the process of data collection from web pages for their applications.
- Digital Marketers: Marketers aiming to track promotional offers and product details across different platforms for campaign optimization.

Problem Solved

This workflow addresses the challenge of manually extracting data from web pages, which is often time-consuming and prone to errors. By leveraging a vision-based AI Agent alongside ScrapingBee, it automates the process of capturing screenshots and retrieving HTML data, ensuring accurate and structured information extraction. This is particularly beneficial for users who need to gather data quickly and efficiently, without the need for extensive coding or technical expertise.

Workflow Steps

  • Manual Trigger: The workflow begins when the user manually triggers it by clicking ‘Test workflow’.
    2. Google Sheets Integration: It retrieves a list of URLs from a specified Google Sheet, which contains the pages to be scraped.
    3. Set Fields: The workflow sets the necessary parameters, particularly the URL, to be sent to the ScrapingBee API for data extraction.
    4. ScrapingBee - Get Page Screenshot: It captures a full-page screenshot of the specified URL using ScrapingBee, which is crucial for the AI Agent to analyze the content visually.
    5. Vision-Based Scraping Agent: The AI Agent analyzes the screenshot to extract relevant product information, such as titles, prices, and promotional details. If it encounters difficulties, it falls back on an HTML-based scraping tool.
    6. HTML-Based Scraping Tool: If needed, the agent retrieves the HTML content of the page for further analysis, ensuring no data is missed.
    7. Structured Output Parser: Extracted data is formatted into a structured JSON format suitable for easy integration into Google Sheets.
    8. Split Out: The structured data is split into individual rows for better organization.
    9. Google Sheets - Create Rows: Finally, the workflow appends the extracted data as new rows in the designated Google Sheets results sheet.

    • Customization Guide

    Users can customize this workflow by:
    - Modifying the Google Sheets Document ID: Change the document ID in the Google Sheets nodes to point to a different sheet that contains URLs or results.
    - Adjusting the Structured Output Parser: Tailor the JSON schema in the Structured Output Parser node to fit the specific data fields required for their use case.
    - Adding Additional Fields: Users can enhance the Set Fields node to include more parameters to be sent to the ScrapingBee API, depending on their data needs.
    - Choosing Different AI Models: Users can experiment with different AI models or configurations within the Google Gemini Chat Model node to optimize performance based on their specific tasks.
    - Customizing Prompts: The prompts used in the Vision-Based Scraping Agent can be adjusted to refine the extraction process and improve accuracy.