Structured Data
Many automation steps consume or produce structured data — JSON-like objects with named fields — so you can branch, map, and pass values between steps.
Where structured data appears
- Trigger outputs — e.g. new email (subject, body, from, attachments), new contact (name, email, company).
- Extract Information — You define a schema (field names and optionally types); the step returns an object with those fields.
- Generate Structured Data — AI produces output that conforms to a JSON schema you specify.
- Assistant (structured output) — When using an Assistant step with structured output, the result is a defined shape.
- Integration actions — Create/update company, contact, deal, etc. return objects with IDs and properties.
Using structured data in steps
- Variables — Reference fields with
#StepName.fieldName#or nested#StepName.object.nestedField#. For lists, use#StepName[0].field#. See Variables. - Conditions — In PATHS or other conditional steps, use field values (e.g.
#ExtractInfo.dealStage# = 'Won'). - Inputs — Pass a field into the next step (e.g. email body into Generate Text, or contact ID into HubSpot Update).
Output schema
In the step configuration dialog, Output Schema shows the shape of the step’s output. Use it as the source of truth for field names when writing variable references.
Best practices
- Prefer referencing specific fields instead of whole objects when you only need one value.
- For optional fields, handle null/empty in conditions or later steps.
- Use Extract Information or Generate Structured Data when you need a fixed structure for branching or integrations.
Last updated on