Structured data extraction
Turns the free text of a request or an email into fields — so they can be stored, counted or passed on.
Graph preview
How it works
Instead of an agent this graph uses a Structured output node: it carries a JSON schema
with name, email and summary, and the model must return an object of that shape rather
than free text. The result can then go into a database, a CRM, or a field-by-field
comparison.
What to configure after import
- Describe your own schema: field names and descriptions are the instruction to the model, so “date as YYYY-MM-DD” in a description works better than the same sentence in a prompt.
- Choose required fields deliberately: the more there are, the more often the model invents something just to fill them.
- db_insert is a natural next node — store what was extracted.