Extract Data From Document
AI-generatedSummary
Extract data from a document by submitting it to a selected Cradl AI agent, which processes the document using AI and optionally human review, returning extracted structured data.
Inputs
- Agent Name or ID (required) — Select the Cradl AI agent to process the document. This choice controls how data is extracted from the document.
- Use Previously Uploaded Document for Testing — Toggle to use a previously uploaded document instead of uploading a new file, for testing or debugging purposes.
- Recent Document Name or ID (required) — If using a previously uploaded document, select the document from earlier executions by name or ID.
- Document Binary Data (required) — The binary data property name of the document to process, such as a PDF or image, typically from prior node output.
- Wait for Results — Decide whether the node waits for processing to complete and returns the extracted data, or continues immediately for asynchronous processing.
- Variables — JSON object of variables sent along with the document, accessible within the Cradl AI workflow and included in the output.
- Resume URL Variable Name — When waiting for results, specifies the variable name storing the resume URL for workflow continuation.
- HMAC Secret — Shared secret for generating and verifying the HMAC signature for webhook validation to ensure secure callbacks.
Output shape
a list of items containing the initial run data from Cradl AI when starting document processing; if waiting for results, the node pauses execution and expects data returned later via webhook; otherwise, returns run data immediately.
If 'Wait for Results' is enabled, the node execution pauses and resumes once the asynchronous document processing completes and the webhook callback is received; if disabled, the node returns immediately with the run initiation data, and workflow must handle retrieving results separately.