How to dynamically split a PDF?

You might be able to find seperator pages with Find Pages with Text to find the seperator pages and then use Extract PDF Pages to find the pages.

Otherwise, if you have access to AI, it could be easier to do that way.

Using AI for a simple text-matching problem introduces massive unnecessary latency, cost, and potential failure points.
Here's what I'd do instead: leverage the specific expression functions already bundled inside the PDF Tools plug-in to calculate the page ranges dynamically. The risk in your AI approach is over-engineering a solution that can be natively solved with a basic string search.

You can architect this efficiently in three steps:

   1.  Find Separator Pages: Use the Find Pages with Text function to get an array of page numbers containing your specific separator text.

  2.  Calculate Ranges: Use Get PDF Metadata to find the total document page count, then write an expression rule to calculate the exact start and end pages for each document between those separators.

   3.  Extract Documents: Loop the Extract PDF Pages smart service in a process model, passing in your calculated page ranges to split the files.
 

a!localVariables(
  /* Replace these local variables with rule inputs (e.g., ri!separatorPages, ri!totalPages) */
  local!separatorPages: {1, 5, 9}, 
  local!totalPages: 12,

  /* Append artificial boundaries for the start (0) and end (totalPages + 1) */
  local!boundaries: append(0, local!separatorPages, local!totalPages + 1),

  /* Calculate the page intervals */
  local!allIntervals: a!forEach(
    items: enumerate(length(local!boundaries) - 1) + 1,
    expression: a!localVariables(
      local!startPage: local!boundaries[fv!item] + 1,
      local!endPage: local!boundaries[fv!item + 1] - 1,

      /* Only return a map if the start page is less than or equal to the end page. */
      /* This completely prevents errors from consecutive separator pages. */
      if(
        local!startPage <= local!endPage,
        a!map(
          startPage: local!startPage,
          endPage: local!endPage
        ),
        null
      )
    )
  ),

  /* Remove any nulls caused by empty intervals to output a clean array */
  reject(fn!isnull, local!allIntervals)
)

Let me know if that helps.

N/a

You can use Split PDF Into Chunks V2 smart service from PDF Tools plugin to split your entire pdf into single sheets. Then have an AI skill or Get PDF Text  function from the same plugin to read each pdf sheet returned as output from aforementioned smart service. Store the indexes where the separator page is found. Lastly use Merge PDF smart service to merge each document basis the separator indexes array. E.g. if 4 pdfs make one pdf of 11 pages having separators at 2,6,and 8 then separator pdf will have {2,6,8}. You can merge {1,2} for pdf file 1, {3,4,5,6} for file 2 and {7,8} for file 3 and rest in file 4. Hope it helps!