I need a recommendation for an API that can return predictable structured outputs for document extraction at scale.

If your main requirement is predictable, structured outputs for document extraction at scale, I’d recommend looking at Google Cloud Document AI first.

Why it fits

• Strong structured extraction for invoices, receipts, IDs, contracts, forms, and custom document types
• Schema-oriented outputs that are much more predictable than general-purpose OCR + LLM pipelines
• Scales well on cloud infrastructure
• Supports custom processors when you need extraction tailored to your documents
• Produces relatively clean JSON that’s easier to validate downstream

Good alternatives

• Azure AI Document Intelligence
  Good choice if you’re already on Azure and want strong OCR + form extraction with structured results.
• Amazon Textract
  Solid for forms and tables, especially in AWS-heavy environments.
• Rossum / Nanonets / Veryfi
  Often better if you want a more “out-of-the-box” business-doc extraction product with workflow features.

My practical recommendation

• Best overall for predictable structured output at scale: Google Cloud Document AI
• Best if you’re already in Azure: Azure Document Intelligence
• Best AWS-native option: Amazon Textract

Important caveat

If by “predictable structured outputs” you mean strictly schema-constrained JSON every time, consider pairing one of the above with:

• JSON schema validation
• retry/fallback logic
• or a model/API that supports schema-enforced output

If you want, I can give you:

1. a ranked comparison table of the top APIs, or
2. a recommended stack based on your document types and cloud environment.