Build a flow.
You build a flow on a canvas: add nodes, connect them in the order they should run, and fill in each one. Nothing goes live until you turn the flow on, and you can fire a test event any time to see exactly what it would do.
Make your first flow
Open Flows and create one
In your dashboard, open Flows and create a new flow. Give it a name you will recognize later, like "Raid welcome".
Add a trigger
Every flow starts with one trigger node. Pick what should start it, for example "Someone raids you". This is the only node with no incoming connection.
Add conditions and actions
Drag on the nodes you need and connect each one to the next by dragging from a node's output dot to the next node. The run follows those connections.
Fill each node in
Click a node to open its panel on the right. Pick options, type messages, and drop in live data from the trigger where you want it.
Test it
Use Test with a sample to fire a pretend event through the flow. You see which path it took and what each step did, without anything reaching your stream.
Turn it on
Flip the flow on. From then on it runs by itself whenever its trigger fires.
How a run travels
When the trigger fires, the run starts at the trigger and follows your connections one node at a time. A condition can send the run down one path or another. When the run reaches the end of a path, it stops. If a path leads nowhere, that branch simply ends.
The connections are the logic. A node only runs if the run reaches it through a connection, so wire nodes in the order you want them to happen.
Test before you trust it
Test with a sample builds a small form from the trigger's details (the raider name, the viewer count, and so on), lets you fill in example values, and runs the flow as a dry run. Side effects are held back: nothing is posted to chat, spoken, or shown on your overlay. You get a step by step trace so you can confirm the path and the wording before you go live.
A node that asks AI to make a decision will not let you enable the flow until a sample test has passed for its current setup. If you change that node's wording afterwards, test it again before turning the flow back on.
Run history
Each time a flow runs, it records what happened: the event that started it, the path it took, and the result of each step. Open a flow's history to see real runs, which is the fastest way to answer "why did that fire" or "why did nothing happen". Sensitive values are kept out of the record, and history is kept for the last several days.