Overview

Conversational Agents provides a simple, visual bot building approach to conversational agent design. Bot designers now have a much clearer picture of the overall bot building process and multiple designers are able to easily collaborate on the same agent build. Conversational Agents supports many languages for your agent to use and understand, but this lab will be using only English.

In this lab you will build a conversational agent using Conversational Agents console.

Prerequisites

You should be generally familiar with the basic concepts of conversational AI. Read through these Sample Transcripts to get an idea of what client transcripts might look like. Often the first step in creating an agent is to read through client transcripts and/or other contextual data to understand the use case and specific business requirements.

Setup

Before you click the Start Lab button

Read these instructions. Labs are timed and you cannot pause them. The timer, which starts when you click Start Lab, shows how long Google Cloud resources will be made available to you.

This Qwiklabs hands-on lab lets you do the lab activities yourself in a real cloud environment, not in a simulation or demo environment. It does so by giving you new, temporary credentials that you use to sign in and access Google Cloud for the duration of the lab.

What you need

To complete this lab, you need:

• Access to a standard internet browser (Chrome browser recommended).
• Time to complete the lab.

Note: If you already have your own personal Google Cloud account or project, do not use it for this lab.

Note: If you are using a Pixelbook, open an Incognito window to run this lab.

How to start your lab and sign in to the Google Cloud console

1. Click the Start Lab button. If you need to pay for the lab, a dialog opens for you to select your payment method. On the left is the Lab Details pane with the following:

   • The Open Google Cloud console button
   • Time remaining
   • The temporary credentials that you must use for this lab
   • Other information, if needed, to step through this lab

2. Click Open Google Cloud console (or right-click and select Open Link in Incognito Window if you are running the Chrome browser).

   The lab spins up resources, and then opens another tab that shows the Sign in page.

   Tip: Arrange the tabs in separate windows, side-by-side.

   Note: If you see the Choose an account dialog, click Use Another Account.

3. If necessary, copy the Username below and paste it into the Sign in dialog.

   {{{user_0.username | "Username"}}}

   You can also find the Username in the Lab Details pane.

4. Click Next.

5. Copy the Password below and paste it into the Welcome dialog.

   {{{user_0.password | "Password"}}}

   You can also find the Password in the Lab Details pane.

6. Click Next.

   Important: You must use the credentials the lab provides you. Do not use your Google Cloud account credentials. Note: Using your own Google Cloud account for this lab may incur extra charges.

7. Click through the subsequent pages:

   • Accept the terms and conditions.
   • Do not add recovery options or two-factor authentication (because this is a temporary account).
   • Do not sign up for free trials.

After a few moments, the Google Cloud console opens in this tab.

Note: To access Google Cloud products and services, click the Navigation menu or type the service or product name in the Search field.

Enable the Dialogflow and AI Applications APIs

1. In the Google Cloud Console search field, type in "Dialogflow API".
2. Select the Dialogflow API tile.
3. Click the Enable button.
4. To activate the AI Applications api, navigate to AI Applications Console and click on Continue and activate the API.

Click Check my progress to verify the objective. Enable the Dialogflow and AI Applications APIs

Task 1. Create your agent

1. To create a new conversational agent, in AI Applications Apps Console choose Conversational agent as the app type, and click Create.

2. A new page for Conversational Agents opens. If a project selection pop-up appears, choose your Project ID .

   Note: If you don't see your Project ID listed, look at the user on the right side to confirm that you are using Conversational Agents as student.

   

3. Select Build your own on the Get started with Conversational Agents popup.

4. On the Create agent page, for the agent display name enter Flight booker.

5. Select location as global.

6. Set the Conversation start to Playbook, and then click on Create button.

7. In the Conversational Agents console, navigate to the Agent Overview. Click Default Generative Playbook, then click the Settings tab, and finally click the (Agent) Settings button.

8. To enable logging for your agent, on the Agent Settings page, click the General tab. Then, scroll down to the Logging settings section and check the boxes next to Enable Cloud Logging and Enable Conversation History. It will generate logs for this agent.

   

9. Click Save.

Click Check my progress to verify the objective.

Create an Agent

Knowledge check

Task 2. Intents

Intents are the reasons an end-user has for interacting with the agent, for example, ordering something. You can create an intent for every topic they may want to navigate.

Intents can be reused across Pages and Flows. Each intent is defined by training phrases end-users typically ask. These can be annotated or "labeled" to collect specific parameters, such as arrival city or departure date.

Conversational Agents will suggest annotations as you include training phrases for the intent; they can also be manually annotated to collect the parameter values you want to extract from the end-user's interaction with your agent.

Recommended: in order to reuse intents as well as make maintenance easier, name your intents with clear and explicit names.

Format of intent: category.some_description

Example of formatting:

• Core Intents: main.book_a_flight
• Common intent but not core: supplemental.flight_emissions
• Reusable intents: confirmation.yes, confirmation_no, redirect.live_agent.

Create your first intent

1. To create a new intent, navigate to Flows, then click Manage tab, select Intents, and finally click + Create.

   

2. For the Display name enter main.book_a_flight.

3. Under the Training Phrases header, add each of the following phrases into Custom training phrase field and press Enter after each phrase:

   • Book a flight
   • Can you book my flight to San Francisco next month
   • I want to use my reward points to book a flight from Milan in October
   • My family is visiting next week and we need to book 6 round trip tickets
   • Four business class tickets from Taiwan to Dubai for June 2nd to 30th
   • I need a flight Saturday from LAX to San Jose
   • Book SFO to MIA on August 10th one way
   • Help me book a ticket from 4/10 to 4/15 from Mexico City to Medellin Colombia please
   • I am booking a surprise trip for my mom, can you help arrange that for May 10th to May 25th to Costa Rica
   • Do you have any cheap flights to NYC for this weekend
   • I want to fly in my cousin from Montreal on August 8th
   • I want to find two seats to Panama City on July 4th
   • For my wedding anniversary we want to go to Seattle for Christmas
   Note: For higher model accuracy, using 20-50 training phrases with short and long response options is recommended.

4. Click Save.

   

Note: You can also add training phrases in bulk by creating a training phrase CSV file and uploading it to Conversational Agents.

Task 3. Flows and pages

Flows are used to define topics and the associated conversational paths. Every agent has one flow called the Default Start Flow. This single flow may be all you need for a simple agent.

More complicated agents may require additional flows, and different development team members can be responsible for building and maintaining these flows.

Every flow starts with a Page, and is made of one or multiple different pages thereafter to handle the conversation within a particular flow. The current page an end-user is on is considered the active page. Each page can be configured to collect any required information from the end-user.

Build from your Default Start Flow

The page your agent starts from is called the Default Start Flow. Pages store routing logic, responses (known as Fulfillment), specific actions to take if an intent cannot be matched (known as no-match) or receives no-input (which is when the agent does not receive a response in time).

1. In Flows console, click Build tab.

2. Click Start Page to open the page.

3. From the expanded options on the Start Page, select the + icon next to Routes.

   

4. Select the intent main.book_a_flight from the drop-down, then click Save.

   

5. Next, in the Routes section, click the main.book_a_flight route.

   

6. Scroll down to the Transition section, choose the Page radio option, and then select + new Page from the drop-down menu.

7. Name the page Ticket information and click Save.

   

8. Exit out of the windows to return to the main display of flows to see your new Ticket information page connected to the Start page.

   

   The beginning of the flow now includes a greeting, and will then proceed to the Ticket information page when the main.book_a_flight intent is matched. On the Ticket information page you will collect parameters from the end-user so they can book their flight.

Task 4. Entities and parameters

• Entities define the type of information you wish to extract from an end-user, ex: city you want to fly to. Use Conversational Agents built-in "system entities" for matching dates, times, colors, email addresses, and so on.

• System entities can also be “extended” to include values that are not part of the default system values. If you need to create a fully customized entity, you can do so by creating a Custom Entity type for matching data that is custom to your business and not found as a system entity.

• Parameters are information supplied by the end-user during a session, such as date, time, and destination city. Each parameter has a name and an entity type. They are written in snake_case (lowercase with underscores between words)

Create your first set of parameters

Next, you will use an entity to extract a required parameter from the end-user.

1. Click on the page Ticket Information, then the + by Parameters to collect flight data.

2. Enter departure_city in the Display name field.

3. Choose @sys.geo-city from the Entity type drop-down.

4. Scroll down to the Initial prompt Fulfillment section. Under Agent Responses, click on + Add dialogue response, select Agent dialogue and then add What city would you like the flight to depart from?.

   

5. Click Save.

6. Exit out of this window to make another parameter.

7. Click the + by Parameters again to create 4 additional parameters one by one with the following name, entity type, and how the agent will prompt the end-user.

   Display name	Entity type	Agent dialogue
   departure_date	@sys.date	What is the month and day of the departure?
   destination_city	@sys.geo-city	What is your destination city?
   return_date	@sys.date	What is the month and day for the returning flight?
   passenger_name	@sys.any	What is the passenger's name?

   When finished they are listed like this:

   

Note: The order in which the parameters are listed affects the order in which the flight booking agent will ask for each. You can easily change the order by dragging parameters up or down.

Knowledge check

Task 5. Conditions

Once the agent has collected the necessary 5 flight booking parameters, you want to route the end user to another page using a routing condition, which you will create next.

1. Exit out of the parameter window to return to the Ticket information page again.

2. Scroll down to locate Routes and click the + sign next to it.

3. Scroll down to Condition > Condition rules > select "Match AT LEAST ONE rule (OR)"

4. In the Parameter field enter $page.params.status.

5. Choose the = sign in the Operand drop-down.

6. In the Value field enter: "FINAL" (ensure you include the double quotes).

   

7. Click Save.

Knowledge check

Task 6. Fulfillment

Now add a response to say to the end-user when all 5 of their booking parameters are collected. These responses are called Fulfillment.

1. From the condition you just made, scroll down a bit and locate the section called Fulfillment.

2. Scroll down to the Fulfillment section. Under Agent Responses, click on + Add dialogue response, select Agent dialogue and then add Thank you for that information. Let me check on the availability of your ticket.

   (Now stay on this page while you read on to the next step of confirming information.)

   

3. Click Save.

Task 7. Confirming information

After offering a response (or fulfillment), you need to create a transition to a new page that will repeat back to the end-user if the travel information collected (parameters) are correct.

1. Continue to scroll down (past the fulfillment you just created) until you reach Transition section.

2. Select the Page radio option, and then choose + new Page from the drop-down menu.

3. Type Confirm trip in the field called Page name.

   

4. Click Save.

5. Exit out of the window.

6. Take a look at the flow of your 3 pages.

   

Repeating back the parameters collected from end-users

Session Parameters store information previously collected from the end-user and are active throughout the session. They also help you repeat information back to the end-user.

For example, we can have the agent repeat back a passenger's name: Thanks for providing that information, $session.params.passenger_name. This displays to the end-user as Thanks for providing information, John Day.

They are formatted as follows:

• Prefix: $session.params.
• Entity Name: passenger_name

So referencing the departure city would look like: $session.params.departure_city

7. Starting from the Build view, click on the Confirm Trip page > Entry fulfillment > Edit fulfillment field.

   

8. Since you used 5 parameters, you can repeat them back to the user via the following session parameters. Paste the following text within the Agent responses > + Add dialogue response > Agent dialogue section:

   This is to confirm that $session.params.passenger_name will fly From: $session.params.departure_city To: $session.params.destination_city Leaving on: $session.params.departure_date Returning on: $session.params.return_date Is this correct?

9. Then click Save.

10. This is what it will look like to the end-user when the conversational agent repeats back the collected session parameters:

    

Positive confirmation route

11. Exit out of the window to return to your Confirm Trip page. Click + next to Routes.

    

12. Click the Intents drop-down, then click + new Intent.

    

13. In Display name enter confirmation.yes.

14. In Training phrases enter yes then click on Add (you can add more phrases like "correct", "yup", etc., to improve the NLU matching for this intent).

    

15. Click Save.

16. After saving, scroll down to the Fulfillment section. Under Agent Responses, click on + Add dialogue response, select Agent dialogue and then enter Great, your flight is booked!

    

17. Then click Save.

18. Click the back arrow, next to Route.

    

Negative confirmation route

Now add logic to route an end-user to recollect their flight parameters if they say the information is incorrect.

19. Still on the Routes section select Add route.

    

20. From the Intents drop-down choose + new Intent.

21. Name the intent confirmation.no in the Display name field.

    

22. Scroll down to the Training phrases section type no then click Add.

23. Click Save.

24. Next, scroll down to the section called Transition > Page, then choose Ticket information from the drop-down.

    Note: This is to prompt the user again for their flight information.

25. Scroll up to Parameter presets and click Add parameter. Enter the following 5 values and assign their value to null without the quotation marks.

    Note: You will need to delete the quotation marks in the value column and type null. This is to delete the parameters collected from the end-user.

    Parameter	Value
    departure_city	null
    destination_city	null
    departure_date	null
    return_date	null
    passenger_name	null

    

    The purpose of this is to remove the value that was previously collected from the end user to allow them to submit a new value. If this step is missed, it might result in an infinite loop scenario in your bot!

26. Click Save.

27. Exit out of the window to return to the Build view, you will now see how all 3 pages flow. Note that the last page has two arrows between the Confirm trip and Ticket information pages because the confirmation.no intent is linked back.

    

    Knowledge check

Task 8. Testing

1. To test that your agent works as intended, click on the Toggle Simulator icon located at the upper right corner of the screen.

   

2. In the Start Resource section of the Preview: Default Start Flow page, select Default Start Flow and test your agent.

   

3. Interact with the agent as if you were the end-user. As you move through the main flow, notice the pages, intents, and transitions you created.

   Depending on how you arranged your parameter collection, you can try typing in the following sample dialogue:

   • I'd like to book a flight
   • Austin
   • Tomorrow
   • Boston
   • Next Friday
   • Mickey Mouse
   • Yes

   This should result in a successful transaction through your agent, commonly known as the happy path.

   Here is an example of the above agent testing in the Test Agent console:

   

Click Check my progress to verify the objective. Test the agent

Task 9. Exporting your agent

When you build an agent for one project, you can export it to use in a different project. You can export your agent and save it to use in future labs or to continue building in your own personal project!

1. In the Agent drop down at the top of the Conversational Agents console, click View all agents.

   

2. On the Agent list screen, click the context menu next to your agent and then click Export.

   

3. On the Export Agent screen, choose Download to local file, then click Export.

   

Congratulations!

You've successfully built a complete Conversational Agent Flow! You created your agent, defined intents, structured flows and pages, and used entities and parameters for data extraction. You also defined conditions, fulfillment responses, and thoroughly tested the agent. This lab has truly solidified your understanding of conversational agent development.

Next steps / learn more

• Bookmark the Conversational Agents Documentation
• Go through the entire Conversational Agents Video playlist

Manual Last Updated: May 29, 2025

Lab Last Tested: May 29, 2025

Copyright 2023 Google LLC All rights reserved. Google and the Google logo are trademarks of Google LLC. All other company and product names may be trademarks of the respective companies with which they are associated.