View My Portfolio

🤖 How I Built an AWS Lex Chatbot with Google Calendar Integration for Real-Time Bookings

Posted by:

Tanya

|

On:

|

, , , , , , , ,

Have you ever wanted a chatbot that not only answers customer questions but also books appointments for you in real time — complete with a Google Meet link and calendar invite? I did too.

In this post, I’ll walk you through how I built a chatbot using Amazon Lex, AWS Lambda, and the Google Calendar API to do exactly that. I created this project for a demo client called Tomorrow’s AI, and even though some of the media assets have since disappeared, I wanted to document it here so others can learn from what I built.

Whether you’re a beginner developer, a curious business owner, or a recruiter checking out my work — I hope this guide gives you helpful insight into how to build your own AI-powered scheduling assistant.

🧰 Tools, APIs, and Accounts You’ll Need

Before we start building, here’s what you’ll need to follow along:

✅ AWS Account

You’ll use:

  • Amazon Lex (for the chatbot)
  • AWS Lambda (to process bookings and connect to Google)
  • CloudWatch (for logging/debugging)
    👉 Sign up at: https://aws.amazon.com

✅ Google Cloud Account

You’ll need:

Be sure to:

  • Enable the Calendar API
  • Create OAuth 2.0 credentials (Client ID + Secret)
  • Add your redirect URIs (or use testing mode for development)
  • Generate refresh/access tokens for your Google account (I used a service account for simplicity)

🚀 Step 1: Set Up Your Amazon Lex Chatbot

I started by creating a chatbot inside the AWS Console. Here’s how you can do it:

🛠 How to Do It:

  1. Go to Amazon Lex in the AWS Console
  2. Click “Create bot”
  3. Choose:
    • Bot name: e.g., CustomerSupportBot
    • Language: English (or your preference)
    • Voice: None (for text-only)
    • Role: Let AWS create a new one for you
  4. Disable voice interaction unless you want it to speak
  5. Click Create

👉 Image Placeholder:
(pic coming soon!)

🗣 Step 2: Define Intents & Conversation Flows

Next, I created 20+ intents to handle different user requests, like asking about services or booking a consultation.

🛠 Example:

  • Intent Name: ViewServicesIntent
  • Utterances:
    • “What services do you offer?”
    • “Tell me about your services”
  • Response:
    Use response cards with clickable options:
    • Marketing & Business Development
    • Data & AI Solutions
    • Custom Software & Generative AI

👉 Image Placeholder:
(pic coming soon!)

💡 Pro Tip: For multi-turn conversations, enable “slot elicitation” and configure prompts for additional details like name, time, timezone, etc.

📅 Step 3: Add Slots for Booking Consultations

To schedule meetings, the bot needs to collect user data:

🛠 Booking Slot Example:

  • UserName – string
  • PreferredDate – date
  • PreferredTime – time
  • Timezone – string
  • BusinessIssue – string

Amazon Lex guides the user through filling these slots in order.

👉 Image Placeholder:
(pic coming soon!)

🧠 Step 4: Set Up AWS Lambda for Booking Logic

After the user gives their info, Lex passes it to a Lambda function.

This Lambda does the following:

  1. Validates inputs
  2. Connects to Google Calendar API
  3. Creates an event
  4. Adds a Google Meet link
  5. Sends back confirmation details

🛠 Basic Lambda Flow (Python)

pythonCopyEditimport google_auth_oauthlib.flow
import googleapiclient.discovery
from datetime import datetime

def lambda_handler(event, context):
    # Extract slots
    slots = event['sessionState']['intent']['slots']
    user_name = slots['UserName']['value']['interpretedValue']
    preferred_date = slots['PreferredDate']['value']['interpretedValue']
    preferred_time = slots['PreferredTime']['value']['interpretedValue']
    timezone = slots['Timezone']['value']['interpretedValue']
    issue = slots['BusinessIssue']['value']['interpretedValue']

    # Format datetime and build Google Calendar event
    # [calendar logic here using google-api-python-client]

    return {
        "messages": [
            {
                "contentType": "PlainText",
                "content": "Thanks, your consultation has been booked! Check your email for a Google Meet link."
            }
        ],
        "sessionState": {"dialogAction": {"type": "Close"}}
    }

👉 Image Placeholder:
(pic coming soon!)

🌐 Step 5: Enable Google Calendar API and Auth

Inside Google Cloud Console:

  1. Create a new project
  2. Enable Google Calendar API
  3. Go to APIs & Services > Credentials
  4. Click Create Credentials > OAuth 2.0 Client ID
  5. Download the JSON file with client credentials

💡 I used a service account to simplify auth — this allows your Lambda to schedule meetings without user prompts.

👉 Image Placeholder:
(pic coming soon!)

🔌 Step 6: Connect Lex to Lambda

Now, head back to Amazon Lex:

  1. Open your booking intent
  2. Scroll to Lambda Initialization and Validation
  3. Attach the Lambda you created earlier
  4. Test the full flow in the Lex test chat window

👉 Image Placeholder:
(pic coming soon!)

🧪 Step 7: Test, Debug, and Refine

To make sure everything works smoothly:

  • Use the Lex console test window to simulate conversations
  • Check CloudWatch Logs if anything fails
  • Tweak slot confirmations or prompts if users seem confused

👉 Image Placeholder:
(pic coming soon!)

🧠 Challenges I Faced (and Solved)

🔄 Challenge: Handling Multi-Turn Conversations

Solution: I used session attributes to track state across turns and confirm inputs step-by-step.

📅 Challenge: Authenticating with Google Calendar

Solution: I set up a service account and pre-authorized it to manage the target calendar.

📂 Want to Dive Deeper?

You can find the full source code, README instructions, and example intent configurations on my GitHub repo:

👉 View the Project on GitHub
(Click here to view project in my repository)

🙋‍♀️ Final Thoughts

This project taught me how powerful AWS Lex and Lambda can be when combined with external APIs. It was also a great way to practice building conversational UI and automating customer workflows.

If you’re a beginner looking to build your own chatbot or a company looking for AI-powered automation, I hope this walkthrough gives you a solid starting point.

Feel free to leave a comment, connect with me, or fork the project on GitHub!


Posted by

Tanya

in

, , , , , , , ,