Dialogflow CX Node

The Dialogflow CX node sends user input to a Google Dialogflow CX agent for natural language understanding (NLU). It returns the detected intent, confidence score, extracted entities, and an optional fulfillment response. Use it when you need structured intent classification alongside your flow-based logic.

When to Use

• You have an existing Dialogflow CX agent and want to integrate its NLU capabilities into your OmniBots flow.
• You need precise intent detection with entity extraction (e.g., extracting dates, amounts, product names from user messages).
• You want to combine Dialogflow CX's NLU with OmniBots' flow builder, knowledge bases, and LLM capabilities.
• You are migrating from a Dialogflow-powered bot and want to reuse your trained intents.

Configuration

Property	Description	Default
Integration	The Dialogflow CX integration to use. Selected via the integration selector.	None (required)
Agent ID	Override the Dialogflow CX agent ID from the integration (optional). Use this if you have multiple agents under one integration.	From integration
Query Input Type	The type of input to send: text, audio, or event.	text
Text Variable	The session variable containing the user's text input (for text input type).	last_message
Event Name	The event name to trigger in Dialogflow CX (for event input type).	Empty
Session ID Variable	Variable to store the Dialogflow session ID for multi-turn context.	session_id
Parameters	Key-value pairs to pass as session parameters to the Dialogflow agent.	Empty

imageDialogflow CX node config panel showing integration selector dropdown, agent ID override field, query input type radio buttons, text variable input, and session parameters key-value editor

Dialogflow CX node configuration panel

Per-Node Integration

Each Dialogflow CX node can use a different integration, allowing you to route to different Dialogflow agents within the same flow. If no integration is selected, the node uses the tenant's default NLU integration configured in Settings > Integrations.

Output Variables

The node stores the Dialogflow CX response in configurable output variables:

Output Variable	Default Name	Contains
Intent	detected_intent	The name of the matched intent (e.g., "order_status", "cancel_order")
Confidence	intent_confidence	Confidence score from 0.0 to 1.0
Fulfillment Text	fulfillment_response	The fulfillment text configured in Dialogflow CX for the matched intent
Parameters	intent_parameters	Extracted entities as key-value pairs (e.g., {"order_id": "12345", "date": "2026-03-01"})
End Interaction	end_interaction	Boolean flag indicating if Dialogflow marked this as an end-of-conversation turn

Using the Results

After the Dialogflow CX node, use a Condition node to branch based on the detected intent:

[Dialogflow CX] --> [Condition: detected_intent]
                      ├── "order_status"   --> [API Call: Get Order]
                      ├── "cancel_order"   --> [Go to Flow: Cancellation]
                      ├── "talk_to_agent"  --> [Handoff]
                      └── Default          --> [LLM Response: General Answer]

You can also use the extracted parameters directly:

Variable: order_id = {{intent_parameters.order_id}}

TIP

Combine Dialogflow CX for intent classification with OmniBots' LLM Response node for generating natural answers. Use Dialogflow to understand what the user wants, then use the LLM to craft the response.

imageDialogflow CX node output showing detected intent name, confidence score bar, extracted entity parameters as key-value pairs, and fulfillment text in the debug panel

Intent detection output with confidence scores

Multi-Turn Conversations

The Dialogflow CX session ID is stored in a variable (default: session_id) and reused across turns. This preserves Dialogflow's conversation context, including slot filling, follow-up intents, and session parameters.

Handles

Handle	Direction	Description
Input	In	Receives execution from the previous node
Success	Out	Intent was detected successfully
Fallback	Out	No intent matched or Dialogflow returned a fallback response

WARNING

Dialogflow CX requires a properly configured integration with valid GCP credentials and a deployed Dialogflow CX agent. Test your integration in Settings > Integrations before using it in a flow.