Definition
Structured extraction asks an agent to transform raw text into fields that people and systems can use. It tests schema discipline, missing-field handling, date formats, and resistance to hallucinated data.
AI Agent term
Structured Extraction
Turning unstructured text into reliable fields such as JSON, dates, amounts, and labels.
Why it matters
Automation often breaks when outputs are fluent but structurally invalid. Extraction tasks reveal whether an agent can connect to downstream systems.
Example
Extract order ID, delivery date, issue type, and requested action from a customer email.