# Structured Outputs

> Structured outputs are model responses constrained to a defined format, usually a JSON schema with named fields and data types, rather than free-form text. Many model APIs can enforce a schema directly, and applications typically validate the response before using it. Structured outputs let software reliably read what a model produced, for example the vendor name, date and total from an invoice.

## Why it matters for your business

Most business automations need data that another system can accept, such as a CRM record, an accounting entry or a ticket field. Free-form text has to be parsed and often breaks on edge cases. Requesting structured output, then validating it in code, makes AI steps predictable enough to sit inside a workflow. A schema also documents exactly what the AI step is responsible for, which makes testing easier. Validation still matters, because a response can match the schema and still contain a wrong value.

## Example

A freight broker extracts data from bills of lading. The workflow sends each scanned document to a model with a schema requiring shipper, consignee, BOL number, piece count, weight and line items. Responses that fail validation, such as a weight that is not a number, go to a person instead of being written to the transportation management system.

Related terms: https://www.lumeor.studio/glossary/tool-calling/, https://www.lumeor.studio/glossary/intelligent-document-processing/, https://www.lumeor.studio/glossary/llm-large-language-model/, https://www.lumeor.studio/glossary/evals-ai-evaluation/, https://www.lumeor.studio/glossary/hallucination/

---

Source: https://www.lumeor.studio/glossary/structured-outputs/