Planning to use Chatbot? Think Dialogflow!

Recently, we got an opportunity to use Google Dialogflow for a chatbot. This article is an attempt to provide a brief overview of Google Dialogflow and how it compares to other players in the market.

Also, do read about the primary consideration one must think of while designing a chatbot, in the last section.

Chatbots…Why do we need them?

Chatbots, powered by Artificial Intelligence (AI), are playing a pivotal role these days in shaping up business interaction with users and potential customers.

More and more businesses are setting up chatbots to answer queries of visitors to their web/mobile applications, help existing customers with business processes, and also for automating internal processes.

What has AI got to do with it?

Well, AI helps to provide human-like intelligence to the chatbot.

Well-designed chatbots aid in driving users to contextually relevant areas in the application, thereby enriching customer experience and eventually driving up sales, all possible with a lean customer support staff.

We can use chatbots to automate customer support or assist human customer support for our web application.

What is Dialogflow?

Dialogflow, previously known as, is conversational AI from Google used by some of the largest brands in the world.

Dialogflow is a natural language understanding (NLU) platform used to build conversational applications for mobile apps, web applications, devices, bots, and interactive voice response systems.

Companies using Google Dialogflow include Verizon, CNBC, CNN, DPD, Giorgio Armani, Mercedes, Comcast, Ticketmaster, Wall Street Journal, Domino Pizza, Ubisoft, Best Buy, Easy Jet, KLM Royal Airlines, Malaysia Airlines, ING Bank, and many more.

Some key features of Dialogflow include:

  • It can analyze multiple types of input from customers, including text or audio inputs (like from a phone or voice recording)
  • It can also respond to customers in a couple of ways, either through text or with synthetic speech.
  • It can communicate in multiple languages like English, Hindi, Spanish, French etc.

Dialogflow comes in two flavors – ES and CX

Dialogflow provides 2 types of NLU modules to handle conversations for our system. It calls these virtual agents. These are, "Essentials Agent or ES agent" and "Customer Experience Agent or CX Agent".

The agent is responsible for translating end-user text or audio during the conversation to structured data that our apps can understand.

Similar to human agents, these virtual agents are also trained to handle expected conversation scenarios, and these samples of training need not be overly explicit.

ES (Essentials) Agent

Dialogflow initially came up with the ES agent and it still exists as an agent type, mostly for small and moderately complex conversation models.

In an ES agent, Intents are the building block of conversational design.

What are intents?

Intents are nothing, but a way to categorize the intent of the user in one conversation turn. To capture it, the agent allows us to provide some example phrases, known as training phrases that can come in a conversation. Also, there are entities that capture useful bits of information like name, age, salary, etc. in the conversation.

What is a Context?

Intents have a pretty flat structure. All intents exist adjacent to each other at one level, and the user’s conversation can match any of them in a positive use case scenario. In such a flat structure, to control the flow of a conversation, Dialogflow uses contexts, that is configured for an intent. When an intent matches user conversation, the corresponding output context becomes active. While any contexts are active, Dialogflow is more likely to match further intents that have input contexts matching the currently active context.

See how the console for ES looks with intents below:

CX (Customer Experience) Agent

CX Agent is provided in Dialogflow CX Edition and is provided with many improved features that make it possible to use it for large and complex conversation models.

It also has intents as its ES counterpart. However, CX has decoupled fulfillment and agent responses from intents to make them reusable, unlike ES. The biggest difference is that CX follows a state machine model with explicit conversation control using pages as nodes and state handlers to control paths in a graph-based model of flows and pages.

As compared to ES, CX has, an easy to navigate, graph-based look with flows and pages as nodes as shown below:

Following is a brief comparison between the 2 agent types:

ES Agent

CX Agent
EditionsDialogflow Trial Edition, Dialogflow Essentials EditionDialogflow CX Edition  
The purpose according to Size and ComplexityFor small to medium size conversation models with moderate complexityFor large and highly complex conversation models
Building blocks of conversational designThe flat structure of IntentsGraph structure of flows and pages
Conversation controlLinear conversation paths that simulate nonlinear paths using intents as nodes and contexts to control pathsState machine model with explicit conversation control using pages as nodes and state handlers to control paths
Console user experienceMostly text formsVisual graphs showing conversation paths and text forms for configurations
Intent reusabilityIntents are coupled with fulfillment, events, and responses; specific to a conversation state, so difficult to reuseIntents are simplified to remove this coupling and made highly reusable
Conditional response messagesRequires webhook calls where Dialogflow will hit our preconfigured API and we can handle sending conditional responses in our APICan be configured statically in fulfillment, with static conditions for a route, or with webhooks calls
Webhook error handlingErrors quietly ignored by agent, passed to API caller if presentExplicit error event handling built-in to our agent
Knowledge connectors (feature to parse knowledge docs like FAQs, articles)Present in betaYet to come
Agents per project1100
Pricing and quotasGranularSimplified

GDF vs Other Options

With the rising interest of businesses in AI chatbots, many conversation platform providers have come into the picture. Here is a brief comparison among the big players:


Google Dialogflow

Amazon Lex

IBM Watson Assistant

Azure Bot
Communication ChannelsVoice, textVoice, textVoice, textVoice, text
User Interface for bot creationProvides web interfaceProvides web interfaceProvides a good and easy way to navigate user interfaceProvides web interface, open source web chat widget available on Github
Integrations with existing conversation platformsES:
Slack, Viber, FB Messenger, Twitter, Twilio, Skype, Workplace by Facebook, etc.

FB Messenger,
SMS, Slack, Kik, FB Messenger, TwilioSlack, WordPress, FB Messenger, Voice Agent, etc.Skype, Slack, Kik, FB Messenger, Telegram, Twilio, etc.
LanguagesSupports 20+ languages like Hindi, English, Spanish, French, etc.Only US EnglishSupports 10+ languages (mostly in beta) like English, Spanish, Japanese, Italian, etc.Supports many languages like English, German, Spanish, French, etc.
Cost (as in Jan 2022)ES Agent:

Free (Trial Edition)

Essentials Edition
– Text: $0.002 per request  
– Audio Input: $0.0065 per 15 seconds of audio
– Audio output:      – Standard voices: $4 per 1 million characters,  
   – WaveNet voices: $16 per 1 million characters

CX Edition    – Text: $0.007 per request    – Audio Input/Output: $0.06 per minute
10,000 text requests and 5,000 speech requests or speech intervals per month for free for the first year  

Request-Response interaction:
speech: $0.004 per request text: $0.00075 per request  

Streaming conversation:   speech: $0.0065 per unit
text: $0.0020 per unit

*Every 15 seconds of input (including silence) is counted as one interval; any input is rounded up to the nearest 15-second interval.  
Lite (free)

Up to 1,000 unique monthly active users, up to 10,000 messages per month, and other limitations.  


Starts at $140 and covers from 0 to 1,000 monthly active users per service instance per billing period.

Additional MAUs are billed at USD 14 per 100 MAUs. When a user connects on voice, there is an additional cost of USD 9 per 100 voice MAUs  


More than 50,000 MAUs with the availability of other high limits.
Standard Channels (Microsoft 1st party services like Skype, Microsoft Teams, Cortana) – free, unlimited messages

Premium Channels (that allow a bot to communicate with users in our own applications)
10,000 messages/month, $0.50 per 1,000 messages,   Additional pricing for other features .  

The list of players keeps increasing and we have some other good options as well like BotPress, which has a free open source edition and an enterprise edition on Cloud. Other names include Chatfuel (no-code platform for Facebook), Hubot, and many more.

How best to design a Chatbot to maximize the benefits it provides

Chatbots are still a relatively new business technology. Therefore, businesses are trying to figure out what is the best way to use them in order to make it effective.

One way is to have Quick Reply or Menu Based Chatbots, that present a fixed set of options to the user, and the user has to select any of those as answers to the questions. These types of chatbots are highly accurate and are good for questionnaire-based questions that deal with a fixed business process, like filling up a profile form, etc.

However, for other scenarios, we may need to have a more open-ended interaction with the user to capture product interests and the likes and dislikes of the visitor/customer. For that, we need Keyword recognition-based and Contextual chatbots that respond according to important keywords from customer responses.

Sometimes a hybrid approach is what works the best. In this approach, the bot can provide fixed format quick reply questions to users when keyword recognition is ineffective.

To conclude, AI-powered chatbots can prove to be a very powerful tool to enhance business communication, if used, the right way.

Finding this helpful?

If you need any additional details in this regard then feel free to chat with our live agent at

Thank you !!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s