Skip to content

Intelligent Document Processing

Overview

The Intelligent Document Processing (IDP) suite empowers users to automate document analysis, extraction, and transformation through AI-powered services. Whether you need to translate documents, extract key insights, summarize content, or parse structured data from invoices, these tools simplify the process with minimal manual intervention. The suite supports multiple formats, integrates directly into workflow automation, and provides instant testing for faster iteration.

Note: IDP functions may experience slower execution times depending on the size of the input text. Because Composite APIs, like all API calls, are restricted to a 29-second timeout, it is generally recommended not to use IDP capabilities within Composite APIs. Instead, the best practice is to run these functions inside processes, where longer execution times can be handled more reliably.

This recommendation is particularly important for the Text Summarization feature. Although the tester supports documents up to 5 MB, there is no guarantee that large files can be processed quickly enough to avoid an API timeout, especially when generating summaries for high-volume content.

Additionally, due to the typical execution times, timeouts may also occur in the tester dialog. The tester dialog is primarily intended to help you explore and understand the response data structure, rather than to guarantee full execution of these services.

Key Features

Text Translation

Overview:

This service automatically translates entire documents or plain text from one language to another while preserving the source file’s original format. The translated output is delivered in the same format as the source (e.g., .docx remains .docx). It’s ideal for global teams, multilingual content distribution, or quick document localization.

Input file supported formats: .txt, .docx, .html - Max file: 100KB | Max text: 10KB

How to use it:

  1. Create an AI Service activity – From your workflow builder, add a new activity of type AI Service.
  2. Select the service – Choose Text Translation from the available AI services.
  3. Set the source language – Specify the language of the original document or text.
  4. Set the target language – Choose the language you want the translation in.
  5. Select the input type – Choose between File (to translate an uploaded document) or Plain Text (to translate entered text).
  6. Provide the source – If using a file, enter the File ID of the uploaded document.
  7. Define the output file name – This will be the name of the translated file generated by the process.
  8. Save and run – Execute the workflow to produce the translated document.

Additional Options: You can instantly test the translation without running the full workflow by clicking Test Now in the editor. This opens a dialog where you can upload the file or type/paste text for immediate translation preview.

Analyze Sentiment

Overview:

This service evaluates the tone and emotional polarity of a document (positive, negative, neutral, or mixed). It can be used for analyzing customer feedback, social media posts, or survey responses stored in supported file formats.

Input file supported formats: .docx, .pdf, .txt - DOCX/PDF: max 5 pages; raw text: max 5KB

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Analyze Sentiment.
  3. Specify document language – Select the language the document is written in for more accurate analysis.
  4. Provide the file – Enter the File ID of the document you want analyzed.
  5. Run the service – The output will be returned as JSON data with sentiment scores and classification.

Additional Options: Use Test Now to quickly analyze sentiment from a file or directly pasted text without fully running the workflow.

Analyze US Government ID

Overview:

Specialized for US government-issued identification documents, this service can process US passports and US driver's licenses to extract relevant fields such as name, date of birth, and document number. This is especially useful for KYC (Know Your Customer) and onboarding processes.

Input file supported formats: .png, .jpg, .tif, .pdf - PDF: 1 page max; max 40in/2880pt; western character sets only; password-protected PDFs not supported

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Analyze US Government ID.
  3. Upload and provide file ID – Enter the File ID of the scanned ID image or PDF.
  4. Run the service – The extracted data is provided in JSON format, ready for further processing or storage.

Additional Options: Use Test Now to validate ID recognition on the spot before integrating into a workflow.

Analyze Key Phrases

Overview:

This service identifies and extracts the most important and frequently occurring phrases from a document. It’s perfect for summarizing themes, indexing content, or quickly grasping the main topics discussed.

Input file supported formats: .docx, .pdf - DOCX/PDF: max 5 pages; raw text: max 100KB

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Analyze Key Phrases.
  3. Select the document language – This improves the accuracy of phrase extraction.
  4. Provide the file – Enter the File ID of the source document.
  5. Run the service – Output is structured as JSON data containing extracted phrases and their occurrence counts.

Additional Options: Use Test Now to preview key phrases instantly in the editor.

Expense/Invoice Analysis

Overview:

Designed for finance and operations teams, this service extracts structured data such as invoice number, date, vendor name, line items, and totals from expense or invoice documents.

Input file supported formats: .png, .jpg, .tif, .pdf - PDF: 1 page max; max 40in/2880pt; western character sets only; password-protected PDFs not supported. Only supports English.

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Expense/Invoice Analysis.
  3. Provide the file – Enter the File ID of the expense or invoice document.
  4. Run the service – Extracted details are returned as JSON data for integration into accounting systems.

Additional Options: Use Test Now to verify extraction accuracy before deploying the workflow.

Analyze PII

Overview:

This service scans documents for Personally Identifiable Information (PII) such as names, addresses, phone numbers, and identification numbers. It’s essential for compliance with data protection regulations like GDPR or CCPA.

Input file supported formats: .docx, .pdf, .txt - DOCX/PDF: max 5 pages; raw text: max 100KB

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Analyze PII.
  3. Select document language – Ensures accurate detection.
  4. Provide the file – Enter the File ID of the document.
  5. Run the service – Output is JSON data with detected PII elements and their locations in the text.

Additional Options: Use Test Now to quickly detect PII in a sample document.

Query Document

Overview:

This service enables you to ask specific questions or define prompts to extract targeted data from a document. The results can be stored in custom workflow variables for use in later steps.

⚠️ Important: Query Document is designed to extract values directly tied to a specific keyword or label present in the document (e.g., “What is the quantity of Widget A?”). Generic or table-wide queries may return incomplete or inaccurate results.

Input file supported formats: .png, .jpg, .tif, .pdf - PDF: 1 page max; max 40in/2880pt; western character sets only; password-protected PDFs not supported. Only supports English. Max 15 queries.

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Query Document.
  3. Provide the file – Enter the File ID of the document.
  4. Add instructions – Enter your prompt/question and specify the variable name where the answer should be stored.
  5. Use the format: “What is the {description/value} of {keyword}?”
  6. Example: “What is the payment method for Invoice #12345?”
  7. Run the service – The output variables can be used in subsequent workflow actions.

Best Practices & Recommendations: - Always tie your question to a specific keyword or field label visible in the document.
- Avoid broad queries like “What is the highest amount?” or “List all expenses”.
- For table-like data, target specific cells or labels rather than entire rows or columns.
- In the tester UI, label the field as: Prompt (e.g., “What is the {description/value} of {keyword}?”).

Additional Options: Use Test Now to trial prompts and check accuracy before workflow integration.

Analyze Tables

Overview

Automatically detects and extracts data from tables present in images or scanned PDFs, returning results in a structured JSON format. This is useful for processing forms, reports, or tabular datasets.

Input file supported formats: .png, .jpg, .tif, .pdf

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Analyze Tables.
  3. Provide the file – Enter the File ID of the source document.
  4. Run the service – Output is JSON data containing the detected table structure and cell values.

Additional Options: Use Test Now to ensure table recognition works correctly for your document layout.

Text Summarization

Overview

This service condenses a document into a shorter version while retaining the main ideas. You can control the length of the summary, choose a tone/style (e.g., formal, casual, bullet points), and specify the language of the output.

Supported output languages:

English, French, German, Italian, Portuguese, Spanish, Vietnamese, Thai, Hindi, Indonesian, Tagalog

  • Input file supported formats: .docx, .pdf, .txt

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Text Summarization.
  3. Set the word count – Choose from the available length options.
  4. Choose the style – Select the tone or summary style you prefer.
  5. Set the language – Select one of the supported output languages.
  6. Provide the file – Enter the File ID of the source document.
  7. Run the service – Output is stored in a variable containing the summarized text

Additional Options: Use Test Now to generate summaries instantly in the editor.

Extract Text

Overview

Extracts plain text from images or PDFs (one page only), converting scanned documents into editable content. Ideal for digitizing printed material or processing single-page forms.

Input file supported formats: .jpg, .png, .pdf (1 page only), .tif/.tiff

How to use it:

  1. Create an AI Service activity in your workflow.
  2. Select the service – Choose Extract Text.
  3. Provide the file – Enter the File ID of the image or PDF.
  4. Run the service – Output is stored in a variable containing the extracted text.

Additional Options: Use Test Now to generate summaries instantly in the editor.

Data Privacy & Security

Our AI services handle user data solely for the purpose of executing the requested operation. We have explicitly disabled any option that would allow third-party AI providers to use this information for training or improving their own models. This means:

  • No data retention for AI learning – Your content is not stored beyond the processing task.

  • No reuse for model training – Providers cannot use your documents to enhance their AI capabilities.

  • Task-specific processing only – Data is processed in real time to produce results and then discarded.

This ensures that every document you process remains secure, private, and used only for the intended purpose.