Workflows

Workflows let you build automated pipelines that chain together AI calls, data transformations, API requests, and human review steps, all without writing code. Design once, run on a schedule, a webhook, a form submission, or at the click of a button.

Core Concepts

Concept	Description
Trigger	What starts the workflow: a schedule, a form submission, a webhook call, or an event from an integration
Node	A single step in the workflow: an AI agent call, a code block, an API request, a condition branch, a loop, or an integration action
Edge	A connection between two nodes. Data produced by one node flows along edges to the inputs of the next
Run	One execution of the workflow from trigger to completion. Each run has its own log, status, and cost
Data Table	Persistent tabular storage that any workflow can read from and write to across multiple runs
Subworkflow	A reusable workflow that can be called from other workflows, just like a function

When to Use Workflows

Workflows are the right tool when you need to:

• Run the same AI task repeatedly on different inputs (new emails, form submissions, scheduled data)
• Chain multiple AI calls where the output of one step feeds the next
• Combine AI with external systems: pull data from Jira, process it with an AI agent, post results to Slack
• Build forms that trigger backend AI processing and return structured results
• Implement human-in-the-loop review before an automated action takes effect

For one-off or conversational AI tasks, use Chat instead.

How a Workflow Executes

What's in This Section