OCR AIInstall

Summary

Extract plain text from a document using an AI provider (Gemini, OpenAI, Claude, Grok, or Vertex AI).

Inputs

• binaryPropertyName (required) — Name of the binary property containing the document (used when Input Type is Binary Data). Default: 'data'.
• url (required) — URL of the document to extract from (used when Input Type is URL).
• base64Content (required) — Base64 encoded document content (used when Input Type is Base64).
• fileName (required) — File name with extension, needed to determine file type (used when Input Type is Base64). Default: 'document.pdf'.

Output shape

Two output streams: Success (first) and Error (second).

Each item on either output includes an 'itemIndex' field indicating its position in the input array.

Examples

Example 1: Extract text from a PDF received via HTTP Request node.

Set Input Type to Binary Data, Binary Property to 'data' (or the name used by HTTP Request), leave other fields at defaults, optionally set Provider, Language, Temperature, etc.

Example 2: Extract text from a publicly accessible image URL.

Set Input Type to URL, provide the URL, keep other fields at defaults.

Example 3: Extract text from a Base64‑encoded document stored in a previous node.

Set Input Type to Base64, map the Base64 string to Base64 Content, provide a suitable File Name (e.g., 'scan.png'), keep other fields at defaults.