Chatbot with Dialogflow CX: key concepts demystified #1

Demystifying AI with Engaging Tutos and Comprehensive How-To Series

To build your own conversational AI that serves your purpose, whatever it might be, you will probably brainstorm just to pick the right tool. Today, we will present you with Dialogflow CX from Google. However, before using Dialogflow CX you need to understand its components. Below is a deep dive into the platform’s concepts with everyday analogies to help you understand.

What is Dialogflow CX?

Dialogflow CX is Google’s enterprise level conversational AI platform for building chatbots and virtual assistants with advanced natural language understanding (NLU). It supports text, voice and multi-channel interactions and can handle complex multi-turn conversations.

Dialogflow CX excels at managing sophisticated conversations, providing businesses with the ability to create reliable, adaptive virtual assistants that enhance user experiences.

Dialogflow CX features:
  • Modular Flows and Pages: Manage different parts of a conversation for scalability.
  • Multi-language Support: Talk to users in multiple languages.
  • Rich NLU: Extract parameters and intent recognition.
  • Webhooks and Fulfillment: Integrate with external systems for real-time data exchange.
  • Multi-channel Deployment: Deploy across web, apps, Google Assistant or phone systems.
How to Use Dialogflow CX:
  • Manage Flows for Scalability: Use modular flows for different scenarios (like support tickets or account inquiries).
  • Use Entities and Parameters: Extract data from user input for personalized conversations.
  • Use Webhooks and Fulfillment: Get real-time data from APIs for dynamic responses (like order status).
  • Test thoroughly in the Test Console: Before you deploy.
  • Deploy across Channels: Make your chatbot available on web, mobile apps and voice platforms.

With the description, features and how to use Dialogflow CX out the way, let’s dive into its Key Concepts. 

shape-down-purple

Dialogflow CX Concepts

Agent

Definition

An Agent is the brain of your chatbot. It holds all the conversational logic, flows and intents. It takes requests from your users, connect to the backend, get the response and reply to users

Analogies
  • Hotel Receptionist: The receptionist knows who to call or what to tell based on the guest’s request.
  • Smartphone Hub: Just like your phone manages contacts, apps and messages, an agent manages all the components for conversations.
Location
  • Go to Dialogflow CX
  • Select a project in the project Dropdown at the top of the page
  • Next click “Create agent” and choose your option between Auto-generate and Build your own

Flows

Definition

A Flow is a part of a conversation, focused on a specific topic or task. Multiple flows allow the agent to manage different types of conversations separately.

Analogies
  • Different Store Departments: In a store, one section handles groceries while another handles electronics and the customer is directed accordingly.
  • Phone Menus: When calling customer service, pressing 1 takes you to billing and pressing 2 takes you to tech support.
Location
  • From the agent list, click on the newly created Agent.
  • The Flow page opens as a canvas with starting point the “Start page”. This is the entry point between your users and your agent.

Pages

Definition

A Page is a single state in a conversation. Each page has prompts, user responses and transitions to move the conversation forward.

Analogies
  • Board Game Spaces: Each space represents a step in the game and players move from one to another based on rules or dice rolls.
  • Recipe Steps: You move through each step in a recipe, gathering ingredients and following instructions to get to the final dish.
Location
  • In the bottom left part of the Flow page, you can have the section Pages
  • Click the plus button to add a new page to your flow.
  • Click the list button to list the existing pages of your flow.

Intents

Definition

Intents help the agent understand what the user wants to do. They contain examples of user phrases and associated actions.

Analogies
  • Ordering at a Coffee Shop: When you say “I’d like a cappuccino” the barista knows you want a specific drink without needing the exact words.
  • GPS Destinations: No matter how you say “take me home” (e.g. “I need to get home”, “Navigate to my place”) the GPS knows you want to go to a destination.
Location
  • Click on the Manage tab, then if Intents is not the default selection, click on it.
  • You can see the “Default Welcome Intent” with 16 Training phrases.
  • Click on “Default Welcome Intent” to review the its details page and its Training phrases.

Entities

Definition

They are responsible for the transformation from human language sentences to structured data by extracting key pieces of information like names, numbers, colors, date. There are system entities for most, however you can create your custom entity to match your custom data.

Analogies
  • Filling Out a Form: A form extracts your name, email and phone number from the information you provide.
  • Calendar App: When you say “Schedule a meeting on Friday” the app knows that “Friday” is a date.
Location
  • Click on the Manage tab, then click on Entity types.
  • You have two options, Custom entities and System entities.
  • Click on “Create new” to create your custom Entity.

Parameters

Definition

Parameters store the values captured by entities so you can reuse them throughout the conversation. You can pass these parameters throughout pages when interacting with the user.

Analogies
  • Sticky Notes: You write down important details (like a phone number) on sticky notes to refer to later.
  • Shopping List: Once you note down items, you use the list at different stores without asking again what to buy.
Location
  • To access Parameters, you first need to create a page, then from the page details, click on Parameters.
  • In the parameters list, click on Add new to create one.
  • The type of parameter is very important and should match your intended use.

Fulfillment

Definition

Fulfillment how and where your Agent get the response to the end-user.

It executes backend logic to get preset or generate dynamic responses. It can extract the information requested from the back end data. It often involves calling external services to get up-to-date information.

Analogies
  • Waiter Checking the Kitchen: When you ask if a dish is available the waiter checks the kitchen before confirming.
  • Weather App: The app fetches real-time weather data from an external server when you open it.
Location
  • Create an intent/Route, fill out other option like Training phrases, then expand Fulfillment
  • From the list of option(Parameters presets, Generators, Webhook, etc), select how to generate the response to the end-user.

Webhook

Definition

A Webhook is the connection that allows Dialogflow CX to send data to or receive data from an external service. It powers the fulfillment process.

Think of it as a way for the back end service to collect data through external APIs.

Analogies
  • Pizza Delivery Tracker: The app calls the restaurant’s system to check the status of your order and displays the result.
  • Bank ATMs: The ATM connects to your bank’s server to get your balance.
Location
  • In the Fulfillment section expanded, expand Webhook settings.
  • Click “Enable webhook” to activate the dropdown, then select or create a webhook.
  • In the next page, you will provide all required informations about the webhook like URL, Authentication, Request headers, etc.

Transitions

Definition

Transitions determine how the conversation moves from one page to another based on conditions or user input. When you start the chat and tell the chatbot your intention(intent), then your are forwarded(Transition) to the next page.

Analogies
  • Traffic Lights: Based on the light’s color (input) you know whether to stop or go.
  • Choose-Your-Own-Adventure Book: Your choice on one page takes you to a different outcome in the story.
Location
  • The Transition is mostly available as the last option of an intent.
  • Create an intent/Route, fill out other option like Training phrases, Fulfillment, Webhook.
  • Then only at the end, decide where to go based on the previous data.

Test Console

Definition

The Test Console lets you simulate conversations to see how the agent behaves with different inputs. It’s where you test before you go live. From the user input to the Agent response, you can pretty much test everything here: Training phrases, Intents, Entities, Parameters, Fulfillment, Webhooks, Transitions.

Analogies
  • Practice Interview: You simulate the real interview with a friend to practice your answers.
  • Test Drive: Before buying a car you test how it feels on the road to see if it meets your needs.
Location
  • At anytime, in any window of your project, click on “Test Agent” at top right conner of your screen
  • A dialog opens and invites you to start interacting with the Agent.
  • Type something like “Hi there”, the agent should reply. Then you have it
Conclusion

Dialogflow CX gives you a framework to build natural, dynamic conversations by breaking down key concepts into flows, pages, intents and parameters. With webhooks and fulfillment you can create highly interactive virtual agents. Understanding these with everyday analogies makes it easier to design a smooth user experience.

With Dialogflow CX, conversational agents can be as smart as a well-trained concierge ready to guide, assist, and impress users.

Stay Connected and Informed

We believe in continuous learning and keeping our audience informed about the latest AI trends and technologies. By subscribing to our YouTube channel and following our blog, you’ll stay ahead of the curve with the most up-to-date AI insights and practical applications.

Join us on this exciting journey as we explore the limitless possibilities of AI together. Don’t forget to subscribe to our YouTube channel, follow our blog, and engage with our content to learn, practice, and stay informed.

Join the AI Revolution

Don't get left behind, Subscribe Now and be a part of the AI revolution!
Subscribe Now