Understanding individual blocks is important, but the real power of Jsonify comes from combining them into effective workflow patterns. This section highlights several common patterns you can adapt for your specific needs.

​
1. List Processing (Batch URL Opening & Extraction)

This is one of the most fundamental patterns for processing multiple similar pages when you have direct URLs.

  • Blocks:
    1. Open Websites (configured with a list of URLs, or a numbered range, or imported CSV)
    2. Extract Data (configured to extract desired information from each page)
  • Use Case: Extracting product details from a list of product URLs, scraping contact information from multiple company “About Us” pages.
  • How it Works: The Open Websites block feeds one URL at a time to the subsequent Extract Data block. The Extract Data block then processes that single page according to its schema. This repeats for every URL in the input list.
  • Diagram: [Open Websites (List of URLs)] ➙ [Extract Data]

​
2. Deep Dive Extraction (Navigate & Scrape)

This pattern is used when you start on a general page and need to navigate to more specific sub-pages to find the data.

  • Blocks:
    1. Open Websites (opens an initial overview or listing page)
    2. Find Links (identifies links to detail pages, e.g., individual product links)
    3. Follow Links (navigates to each found link, typically using “Follow each link”)
    4. Extract Data (extracts detailed information from each detail page)
  • Use Case: Scraping all movies and their details from a movie listing site; collecting information about all speakers from a conference attendees page.
  • How it Works: The agent opens the main page, finds all relevant detail links, then visits each detail page one by one to extract the required information.
  • Diagram: [Open Websites] ➙ [Find Links] ➙ [Follow Links] ➙ [Extract Data]

​
3. Paginated Data Handling (Iterating Through Pages/Content)

This pattern addresses extracting data that spans multiple “pages.” This can include traditional numbered pages, content loaded by “Next” or “Load More” buttons, or even dynamically loaded infinite scroll feeds.

  • Blocks (Primary Method - Using Pagination Block):
    1. Open Websites (opens the first page of the list/feed)
    2. Paginate a list (is used to handle “Next” buttons, numbered page buttons, or infinite scroll for a set number of iterations/pages. For infinite scroll, each scroll action that loads new content can be considered a “page” or viewport change.)
    3. Extract Data (extracts data from the content visible on each “page” or after each pagination action)
  • Alternative (Direct URLs - if pagination structure is known):
    1. Open Websites (configured with “Add a numbered range of URLs” to generate direct links for page 1, page 2, page 3, etc., if the URL structure supports this)
    2. Extract Data (extracts data from each numbered page)
  • Use Case: Collecting all search results for a query; scraping all products from a category that has multiple pages or uses infinite scroll; extracting all posts from a social media feed.
  • Note on Pagination Block: The pagination block intelligently interacts with “Next” buttons, numbered page links, or simulates scrolling for dynamically loading content. For infinite scroll, you define how many “scrolls” (viewports) it should process.
  • Why Direct URLs (Alternative) Can Be Good: As mentioned in “Workflow Design Principles,” providing direct links for each page (if possible) can sometimes be more robust or less likely to trigger anti-bot measures, as each direct URL might initiate a new browser session, potentially with a unique IP address.
  • Diagram (Primary): [Open Websites (Initial Page)] ➙ [Paginate through Listings] ➙ [Extract Data (runs for each paginated segment)]

​
4. Dynamic Search & Extract

This pattern is used when your starting point is a set of search terms or other variables, and you need to find and extract information related to each.

  • Blocks:
    1. Open datasets (provides a list of search terms, company names, etc.)
    2. Search on Google (or other search block, using the variable in the query, e.g., Search for "{{company_name}} official website")
    3. Find Links (identifies the most relevant link from the search results, e.g., Find the official website link)
    4. Follow Links (navigates to the found relevant link, typically “Follow the first link”)
    5. Extract Data (extracts information from the target page)
  • Use Case: Finding official websites and contact emails for a list of companies; looking up product specifications for a list of product names.
  • How it Works: For each variable set, the agent performs a search, identifies the best link, navigates to it, and then extracts data.
  • Diagram: [Open datasets] ➙ [Search on Google] ➙ [Find Links] ➙ [Follow Links] ➙ [Extract Data]

​
5. Site Exploration for Targeted Data Extraction or Summarization

This pattern leverages the AI’s ability to understand and navigate to find specific pages, after which data is extracted or summarized.

  • Blocks:
    1. Open Websites (opens a starting page or a list of pages)
    2. Explore Content (instructs the agent to find specific target pages, e.g., “Find all pages related to pricing policies” or “Locate the ‘About Us’ and ‘Contact’ pages”)
    3. Extract Data (configured to extract specific information from the found pages, or to generate new content like a summary or FAQ based on the content of these targeted pages. The summarization/generation logic is defined within this Extract Data block’s instructions.)
  • Use Case: Creating an FAQ by first finding the website’s help/FAQ pages and then extracting Q&As; summarizing key information by first locating relevant blog posts or reports on a topic, then instructing the Extract Data block to summarize their content.
  • How it Works: The agent first explores the site based on your goal to find relevant pages. Once on a target page (or after collecting content from several target pages if Explore Content is set up to do so), the subsequent Extract Data block processes this information from the now current page(s), potentially using generative AI capabilities for summarization or Q&A generation.
  • Diagram: [Open Websites] ➙ [Explore Content (Finds Target Pages)] ➙ [Extract Data (Extracts/Summarizes from Found Pages)]

These patterns are starting points. Feel free to combine and adapt them to create sophisticated workflows tailored to your unique automation challenges!