Flow Builder Documentation

The Flow Builder is a powerful visual workflow design system that enables administrators to create dynamic AI-driven conversations using an intuitive drag-and-drop interface. It allows teams to design conversational experiences, automate workflows, integrate external systems, and manage conversation logic without writing complex code.

The platform follows a response-driven architecture where every user response can dynamically change the next step of the conversation. This enables businesses to create both simple guided menus and highly advanced conversational AI workflows.

Flow Management

The Flow Builder dashboard serves as the central management interface for all conversational workflows available within the platform.

Administrators can create new flows, search existing flows, inspect saved configurations, and manage workflow lifecycle states directly from this interface.

Search flows using flow name, description, or ID.
Filter flows by Active, Draft, and Archived status.
Create new workflows from scratch.
Import existing flow configurations.
Manage workflow lifecycle states directly from this interface.

Figure 1 — Flow Builder Dashboard. Central management interface for all conversational workflows. Search flows by name, filter by status (All / Active / Draft / Archived), and click + Create Flow to begin.

1.1 Creating and Importing Flows

The platform supports both creating brand-new workflows and importing previously exported flow configurations.

Click + Create Flow to open a blank workflow canvas.
Use Import to upload a previously exported flow file.
Assign a flow name and version.
Start designing the conversation logic from the Node Library.
Connect nodes to define continuity while building.

Flow Canvas

The Flow Canvas is the primary workspace used to visually design conversations. It provides a scalable and interactive environment where nodes can be arranged, connected, grouped, and managed efficiently.

Drag-and-drop workflow design.
Visual connections between nodes.
Conversation branching based on user responses.
Integrated toolbar for flow management.
Versions and publication statuses.
AI-assisted workflow building support.

Figure 2 — Flow Canvas Interface. The visual workspace showing a connected node graph. Node Library is on the left; nodes are arranged and connected on the canvas.

The Flow Toolbar provides quick access to tools required for managing, validating, organizing, and publishing workflows.

Figure 3 — Flow Toolbar Controls. From left: status badge (PUBLISHED/DRAFT), Grid view, Variables editor, JSON editor, Flow groups, Validate, Export, and Save.

button	desc
Status Badge	Shows current flow state — DRAFT or PUBLISHED.
Grid View	Toggle grid overlay on the canvas.
Variables ({x})	Open Flow Variables panel to add/edit variables.
JSON Editor (</>)	Open the raw Flow JSON editor.
Flow Groups	Group related nodes visually for better organization.
Validate	Run validation checks on flow structure and connectivity.
Export	Download workflow configuration for backup and sharing.
Save	Save the workflow in Draft mode.

2.2 Flow Variables

Flow Variables allow dynamic data storage and reuse throughout conversations. Variables can store customer information, user responses, API outputs, or any reusable conversational data.

Variables improve personalization and allow workflows to maintain conversational context across multiple nodes.

Store user-transfer information.
Reuse values across nodes.
Create dynamic and personalized messages.
Support conditional logic and branching.

Figure 4 — Flow Variables Interface. Define variables using {VariableName} syntax. Add a Variable Name and Default Value, then save — variables appear in the Current Variables table.

2.3 Flow JSON Editor

The Flow JSON Editor provides advanced access to the underlying workflow configuration. This feature is useful for technical users who need fine-grained control over flow structure.

View node configurations and transitions.
Format and minify JSON.
Enable Auto-Apply updates.
Copy and edit raw workflow configuration.
Validate JSON structure in real time.

Figure 5 — Flow JSON Editor. Advanced editor showing the underlying JSON structure of the flow. Supports Reset, Minify, Format, Fold/Unfold, Copy, Wrap, and Auto-Apply.

The JSON editor also allows teams to quickly migrate, troubleshoot, or version-control workflow configurations.

Node Library

The Node Library contains all reusable building blocks required to create interactive AI workflows. Nodes can be dragged onto the canvas and connected together to define conversation flow and logic.

name	category	desc
Start Node	START & END	Entry point of every flow. A flow must contain exactly one Start Node to define the beginning of conversations.
End Node	START & END	Terminates the conversation. Multiple End Nodes can be used to terminate different conversation branches.
Ask Node	CORE NODES	Sends a message and waits for user response. Commonly used for collecting information.
Say Node	CORE NODES	Sends a message without waiting for a reply. Ideal for confirmations, greetings, or informational messages.
Action Node	FLOW CONTROL ACTIONS	Performs platform actions such as ending a call or sending an SMS.
API Call	FLOW CONTROL ACTIONS	Makes external API requests to integrate CRMs, databases, payment systems, and business applications.

3.1 Start and End Nodes

Every workflow must contain one Start Node and at least one End Node. These nodes define the beginning and completion of conversations. Multiple End Nodes can be used to terminate different conversation branches.

Figure 6 — Start Node. Every flow must have exactly one Start Node. The Node Library shows it as disabled once the flow already has a start node.

Figure 7 — End Node. Terminates the conversation branch. Multiple End Nodes can be used across the flow to handle different exit paths.

3.2 Core Conversation Nodes

Core nodes handle conversational messaging and user interaction.

Ask Nodes are commonly used for collecting information, while Say Nodes are ideal for confirmations, greetings, or informational messages.

Figure 8 — Ask Node. Sends a message to the user and waits for a response before moving to the next node.

Figure 9 — Say Node. Sends a message without waiting for a reply — ideal for confirmations, greetings, or informational messages.

3.3 Action and API Nodes

Action and API nodes extend workflow functionality beyond basic conversation handling.

Action Node — Triggers actions such as ending calls or sending SMS.
API Call Node — Connects with external APIs and backend systems.

API nodes make it possible to integrate CRMs, databases, payment systems, and external business applications directly into AI workflows.

Figure 10 — Action Node. Performs a platform action such as ending a call or sending an SMS without requiring user input.

Figure 11 — API Call Node. Makes an external API request to integrate CRMs, databases, payment systems, or other backend services directly into the workflow.

Flow Validation and Publishing

Before flows can be published, they must pass validation checks to ensure proper structure and connectivity.

Validation helps identify logical issues, missing connections, isolated nodes, and invalid configurations.

4.1 Flow Validation

Check Start Node configurations.
Detect isolated or disconnected nodes.
Validate outgoing and incoming connections.
Display detailed warnings and errors.
Refresh validation after corrections.

Figure 12 — Flow Validation Results. Shows Errors, Warnings, Total Nodes, and End Nodes count with detailed issue descriptions to guide fixes.

The validation panel provides detailed descriptions for each issue, making troubleshooting easier during workflow development. Common validation errors and their fixes:

error	fix
Start node is an end/action node — the flow will end immediately	Add an Ask or Say node as the start, then connect it to the end node.
ask node - start — no outgoing connections	Drag connections from this node to define where users go based on their responses.
ask node - start — completely isolated (no incoming or outgoing connections)	Connect this node to the flow or remove it.

4.2 Saving, Exporting, and Publishing

Flows can be saved as drafts during development and published once all validation checks are completed successfully.

action	desc
Save	Store workflow progress without activating it.
Export	Download workflow configuration for backup or reuse.
Publish	Make the workflow active for live conversations.

Figure 13 — Save and Publish Actions. Export downloads the flow config, Save stores it as a draft, and Publish makes the workflow active for live conversations.

Best Practices

Keep workflows modular and easy to understand.
Use descriptive node and variable names.
Validate workflows regularly during development.
Organize large flows using groups.
Avoid isolated or unreachable nodes.
Test API integrations before publishing.
Use variables for personalization and context retention.

Conclusion

The Flow Builder provides a scalable, low-code environment for designing advanced conversational AI experiences. By combining visual workflow design, node-based architecture, variables, API integrations, validation systems, and publishing controls, the platform enables organizations to create intelligent, automated, and highly dynamic conversational workflows.