Limitations in Now Assist in Document Intelligence

  • Release version: Australia
  • Updated March 12, 2026
  • 2 minutes to read

    • There are several important limitations to be aware of when you’re using Now Assist in Document Intelligence.

    Now Assist in Document Intelligence skill limits

    The following table lists the important limitations for the skills in Now Assist in Document Intelligence.
    Table 1. Skill limits for Now Assist in Document Intelligence
    Limit Description
    File formats
    The supported file formats are JPEG, PNG, PDF, and DOCX.
    Note:
    Encrypted files are not supported.
    File size limit

    The file size limit is 20 MB.
    Page count limit
    Image mode supports 10 to 50 pages per file, although this isn’t a hard limit. The 50-page limit serves as a suggested upper boundary, and actual performance may vary based on the following factors:
    • Model constraints: Some models have specific limits on the number of images per request. For instance, some models may limit requests to a maximum of 20 images.
    • File complexity and size: Documents with dense text or large images consume more tokens and memory, which can impact processing time and efficiency.
    • Platform payload limits: There is a 32MB upper limit for the data passed to the language model. Since image sizes can vary considerably, the exact number of pages that can be processed within this limit cannot be determined in advance.

    Since these constraints vary based on your configuration, document characteristics, and the model used, it is essential to test your documents by gradually increasing the page count to identify the effective limits for your specific use case.

    If image mode is turned off, the page count limit is:
    • 200 pages per file if no tables are defined for the use case.
    • 20 pages per file if a table is defined for the use case.

    Image mode is selected during use case setup. For more information, see Set up a use case for Now Assist in Document Intelligence.
    Maximum number of fields per use case

    The maximum number of fields per use case is 50.
    Supported languages

    For image files that need optical character recognition (OCR) to detect the text in them, there are OCR models to support different language groups. For more information, see Languages supported by Now Assist in Document Intelligence.

    For text-based files, the skill recognizes any language supported by the selected or default model, as described in the model card for the LLM. For more information on LLMs, see Large language models used by Now Assist in Document Intelligence.

    Limits for the document and visual insights AI agent

    The following table lists the important limitations for the document and visual insights AI agent.
    Table 2. Document and visual insights AI agent limits
    Limit Description
    File formats

    The supported file formats are JPEG, PNG, PDF, DOCX, XLSX, CSV, TXT, and PPTX.
    File size limit

    The file size limit is 20 MB.
    Page count limit

    For each document task, the page count limit is 200 pages per file.

    If the AI agent performs the extraction based on a use case that includes a table, the limit decreases to 20 pages per file.
    Supported languages

    For image and text-based files, the AI agent recognizes any language supported by the selected or default model, as described in the model card for the LLM. For more information on LLMs, see Large language models used by Now Assist in Document Intelligence.