> ## Documentation Index
> Fetch the complete documentation index at: https://support.humanizing.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Building Conversation Flows in Plural
> Learn how to design and navigate conversation flows in Plural using frames, elements, and flow arrows for robots and avatars.
The Flow Builder is the heart of Plural — a visual canvas where you connect elements into conversation flows that your robot or avatar executes when it detects a human nearby. Understanding how flows work helps you create engaging, natural interactions from the first moment a user appears.
## How Flows Work
Every Plural project starts with two special system nodes at the top of your canvas: **Human detected** and **No human detected**.
* When a robot or avatar detects a human (via face detection), it begins executing the flow connected to the **Human detected** node, following arrows from element to element.
* When no human is in the robot's field of view, the device loops on the **No human detected** node — your **idle screen**.
* If a robot or avatar loses sight of a face while a flow is running, the flow immediately jumps back to the **No human detected** node.
NAO behaves differently from other devices. While a flow is running, NAO does **not** cancel it when no human is detected. You can use open-ended flows (elements with no outgoing arrow) to give NAO natural pause points where it can restart or wait for interaction. See the section below for more details.
## Awareness Feature
The **Awareness Feature** governs the transition between idle and active states:
* The "No human detected" node acts as a repeating idle state — it loops as long as no human is seen.
* As soon as a face is detected, the flow following "Human detected" triggers and runs until no human has been seen for **3 seconds**.
* Use the **"No human interaction for 20 seconds"** trigger on elements to prompt the robot to repeat a question or offer a hint if the user goes quiet.
## Frames
Frames are containers on the canvas that group a set of connected elements into a logical flow. Each frame is associated with a specific device type (Avatar, Pepper, Temi, or NAO).
Click the **+** button on the canvas or use the toolbar to add a new frame and select the device type.
Click the frame's border to highlight it, then drag any side or corner to resize. Frames also expand automatically when you drag elements to the edge.
Hover over the **top bar** of the frame until your cursor changes, then click and drag to reposition it on the canvas.
Right-click the frame and choose **Delete** from the context menu, or click the robot-type icon in the upper-left corner, open the sidebar panel, and select **Delete** followed by confirmation.
Changing the device type of a frame after you have already added device-specific elements will fail. For example, if your Temi frame contains navigation elements, you cannot switch it to an Avatar frame because avatars cannot move.
## Flow Navigation
Use these controls to move around large canvases comfortably:
| Action | How to do it |
| ------------------------ | ----------------------------------------------------- |
| Pan the canvas | Hold the left mouse button (or middle-click) and drag |
| Zoom in / out | Scroll the mouse wheel |
| Jump to a specific frame | Use the frame selector in the top toolbar |
| Move a frame | Drag its top bar |
## Connecting Elements with Flow Arrows
Draw connections between elements by dragging the **output circle** on one element to another element or to an empty area (which opens the element picker). The arrow defines the sequence the robot follows.
* Add **back buttons** by drawing an arrow from a later element back to an earlier one.
* For long flows, return users to the **second** element at the end rather than the first to skip repeated introductions.
* Leave **no dead ends** — an element with no outgoing arrow leaves the robot frozen on the last screen. Connect open ends back to a menu or the flow start.
Check out the Plural example project, the Wine Sommelier at [https://go.plural.io/s/9c453bf582](https://go.plural.io/s/9c453bf582), to see a well-structured long-form flow in action.
## General Design Suggestions
Good flows feel like natural conversations, not monologues. Keep these principles in mind:
Avoid long introductory speeches, especially ones that repeat on every visit (like main menus). Guide users by suggesting options they can say or tap.
Always provide back buttons or verbal escape routes so users can revisit content or exit gracefully.
When displaying images or video, give users enough time to view them. Consider hiding buttons and letting users say "next" or "back".
Use the "No human interaction for 20 seconds" feature to prompt users who have gone quiet — repeat the question or offer a gentle hint.
## NAO-Specific Differences
NAO is a screenless robot with unique flow behavior. Keep these differences in mind when designing for NAO.
Unlike Pepper, Temi, and Avatar — which stop a flow when face detection is lost — NAO continues running the active flow regardless of human presence. This means:
* **Open ends are intentional** — leaving an element without an outgoing arrow gives NAO a natural breakpoint where it can return to its idle state.
* **No screen** — NAO cannot show visual buttons or UI panels. Every instruction must be delivered verbally. Always tell the listener what words or phrases NAO expects to hear.
* **No "No human interaction for 20 seconds"** fallback is needed in the same way — use open-ended flows instead to create return-to-idle points.
## Context Menu
Right-clicking anywhere on the canvas or on a frame's border opens the **context menu**, which gives you quick access to options such as adding a new frame, deleting the current frame, and other canvas-level actions.