Design a workflow with nodes and connections

Start with the outcome

Before adding nodes, write down the input, required decisions, and possible results. A small explicit diagram is easier to test than a process with premature branches.

Minimum structure

Every workflow needs exactly one Start and at least one End node. Connect steps in execution order and make sure every relevant path can terminate.

Use action-oriented names such as “Check request type” rather than “Step 2.” These labels appear in execution traces.

Configure conditions

A condition evaluates a field with an operator and expected value. Label outgoing connections so each result is unambiguous.

Include a fallback for unexpected or missing values. Without one, a valid but unforeseen input may have no next step.

Add responses

A Response node sends a message to the user. Use it for confirmations or path-specific outcomes. Dynamic data should come from the input or earlier steps rather than invented static text.

Connect branches

Use a node's connection control and select its destination. Review connections after moving or deleting steps. A visible but disconnected node is not part of the process.

Keep the design readable

Lay out the primary route consistently, separate exception branches, avoid cycles, split large processes into published subworkflows, and validate after structural changes.