Actions
How teams define reusable automation in TrialStack and decide when an action should be created, updated, or invoked.
Reuse the right workflow
flowchart LR
Need[Repeatable task] --> Action[Action]
Action --> Run[Execution]
Actions are the reusable automation layer in TrialStack. They define what can be executed, what inputs are expected, and what kind of operational outcome the workflow should produce.
Define the contract once
An action owns:
- the execution purpose
- the expected input shape
- the reusable behavior the team wants to run again
- the handoff into runs, jobs, and artifacts
Choose models deliberately
Action operation forms use the organization’s AI model defaults as placeholders. Leave a model blank when the action should follow the workspace default for that operation, or choose an explicit approved model when the action needs a different execution profile.
Use actions for repeatable work
Create or maintain an action when:
- the same workflow should run consistently across multiple records
- a team needs a named automation surface that can be reviewed and rerun
- the logic is important enough that it should not stay buried in one-off manual steps
Do not confuse it with the result
An action defines what can happen. It is not the execution history itself and it is not proof that the resulting output is ready for use. That review boundary lives in the run and the wider workflow around it.