Concepts
These pages explain the ideas behind AI Studio: what a flow is made of, how it starts, and how data moves from one step to the next. Once you've read them, the rest of the documentation is easier to follow.
How the pieces fit together
A flow is an automation you build on a canvas. Every flow has the same shape:
- A trigger starts it. That might be a form submission, a message in Slack, a schedule, or a request to a webhook URL.
- A series of steps runs after it. Each step is a node that does one job, such as sending an email, creating a CRM record, asking an AI model a question, or branching on a condition.
- Each step can use data from earlier steps. The trigger's data and the output of every earlier step are available to the steps after them.
When the trigger fires, AI Studio runs the flow once. That run is an execution, and you can open it later to see what each step received and returned.
Trigger Step Step Step
┌───────────┐ ┌───────────┐ ┌───────────┐ ┌───────────┐
│ Form │ ───▶ │ Create │ ───▶ │ Ask AI to │ ───▶ │ Send │
│ submitted │ │ CRM record│ │ draft a │ │ email │
└───────────┘ └───────────┘ │ reply │ └───────────┘
└───────────┘Words you'll see
| Term | Meaning |
|---|---|
| Flow | An automation: one or more triggers followed by steps. |
| Node | A box on the canvas. Triggers and steps are both nodes. |
| Trigger | The node that starts a flow when an event happens. |
| Event | Something a trigger listens for, such as New Message Posted in Slack. |
| Step | Any node after the trigger. |
| Action | What a step does. Most nodes offer several actions. The Slack node, for example, can Send Message or Get User. |
| Output | The data a trigger or step produces. Later steps can use it. |
| Execution | One run of a flow, recorded step by step. |
| Credential | A saved connection to another service, such as a Slack token, that nodes use to sign in. |
In this section
- Organizations and workspaces: who can see what.
- Flows and nodes: the canvas, steps, actions and settings.
- Triggers: the ways a flow can start.
- Using data from earlier steps: how to pass one step's output into the next.
- Branches, loops and waits: taking different routes, repeating steps, pausing.
- Storage: files and shared variables that last beyond one run.
- Executions: what happens when a flow runs.
- Versions and publishing: drafts, published versions and going live.