Output

The Output section defines which results are expected to be returned by the AI agent at runtime. Outputs are configured through parameters, which act as structured placeholders and can be used as event received parameters in triggers, dataflows, or external integrations.

Correct output configuration is essential for predictable and reliable agent behavior.


Output parameters

Parameters define the outputs an agent is expected to produce based on the input, instructions, and sources.

  • Parameters are available within the system instruction and the runtime context

  • Each parameter consists of:

    • A name

    • A type

    • An optional description

Supported parameter types

  • Text

  • Boolean

  • Integer

  • Fraction

  • Date

  • Datetime


Important limitation: file generation

An agent cannot generate one or more files as direct output.

However, a PDF file can be generated from text output produced by an agent by invoking an available dataflow designed for this purpose.

As a result:

  • Output parameters of type File or MultipleFiles are not supported

  • Using these types as event parameters to start a dataflow will not work

  • In such cases, the agent or reports will display an error message


Parameter configuration

Field
Description

Name

Unique identifier of the parameter. Used to reference the output within instructions and runtime context.

Type

Defines the data type of the output (e.g. Text). Determines how the output is processed.

Description

Optional explanation of the purpose of the parameter.


Notes

  • Outputs should always be designed to be machine-readable and reusable in downstream flows.

  • For document generation use cases, return structured text output and trigger a dedicated dataflow to create files such as PDFs.

Last updated

Was this helpful?