OCR AIInstall

Summary

Extract structured JSON data from a document using an AI provider according to a user‑defined JSON schema.

Inputs

• binary — When Input Type is 'binary', the binary property (named by Binary Property, default 'data') containing the document to process.
• url — When Input Type is 'url', the document URL to fetch and process.
• base64 — When Input Type is 'base64', the Base64 Content string and File Name (with extension) identifying the document.

Output shape

Two output connections: Success (main) and Error (main).

Each workflow item produces one output item on the matching connection.

Examples

Example 1: Extract invoice fields from a PDF using the default Gemini provider.

Provider: Gemini, Input Type: Binary Data, Binary Property: data, JSON Schema: (invoice_number, date, total, items[]), Options: defaults.