Amazon Connect: Optimize Your Customer Experience with Chatbots


Amazon Connect’s AI innovations give you new ways to deliver stellar customer service

Amazon Connect brings an enterprise-level call-center experience within reach of even the smallest businesses. Designed from day one to be open-source and user-friendly, it puts all its capabilities within the scope of the non-expert end user. One such capability worth exploring is chatbots, and how you can use them to enhance your customer service options.

Leveraging AI to streamline your customer’s service journey

Artificial intelligence, once a subject for science fiction, is now a part of many systems we interact with every day. AI is at the core of what makes Amazon Connect powerful as well as easy to use. Two key Amazon Web Services (AWS) technologies – Amazon Lex and Amazon Polly – use the power of AI to make your customer’s call experiences more efficient and pleasant.

Amazon Lex and Polly are based on the technology that underpins Alexa, one of the most popular intelligent personal-assistant services in the world. In conjunction with Amazon Connect, the two can be used to create a chatbot that can handle many of your most common requests, freeing up your agents to take more complex service tasks. Amazon Connect’s ability to interface with many existing systems, such as your customer relationship management (CRM) system and other systems on the AWS platform, gives Lex and Polly even more power to address your needs.

Amazon Lex and Polly empower chatbots to handle simple requests so your agents don’t have to

Using automatic speech recognition (ASR) and natural language understanding (NLU), a Lex chatbot can understand natural human speech and, through Polly, respond appropriately. The chatbot can take a customer’s call, greet them and proceed to a question such as, “What can I do for you today?” Customers can respond in natural speech and, for example, make a dinner reservation.

         “Hello? Yes, I’d like to make a dinner reservation for tomorrow night at 7:00.”

The Amazon Lex chatbot reaches out to your customer-scheduling application via a request routed through another AWS service called Lambda. The scheduling software finds open seating for the requested time (Polly recognizes a request for a 7:00 p.m. reservation from the words “dinner” and “night” and the date from the word “tomorrow”) and sends the information back to Lex.

The chatbot responds to the customer, “Yes, I do have a table available. Can I have your first and last name?”

“Jane Smith,” the customer answers.

The chatbot confirms the name and spelling with the customer. It then asks, “Do you have any special requests?”

The Lex chatbot converts speech to text and sends the customer’s reservation information, including the text of their special request, back to the scheduling software. The reservation information will appear on the maître d’s tablet the following evening, and he will determine whether the special request can be accommodated.

A chatbot can provide services 24 hours a day, seven days a week. It can provide services like stock and price checks, patient appointments, account balances, driving direction requests – with backup responses sent to the customer’s mobile device – or nearly any other interaction with a clearly defined set of variables. This, of course, includes handing off the call to the correct agent queue (perhaps accounts receivable, tech support, or shipping) in an expedited fashion. It can tackle the minor tasks that are often all it takes to get the customer the help they need as quickly as possible, using the resources at hand.

And Lex and Polly are always getting better at what they do. Every single AWS interaction using each of the complementary services contributes to the ongoing development and, indeed, education, of both platforms.

Creating a simple Amazon Lex bot

Building an Amazon Lex bot is a straightforward step-by-step process. While chatbots can be programmed to handle a variety of tasks, for this example we’ll go through the steps for a simple bot that will schedule appointments.

You’ll build your bot in the Amazon Lex Console (which requires a login). When creating your bot, you can select from a variety of premade bot blueprints. We’re going to choose the ScheduleAppointment blueprint. All bots have intents, which are tasks that the bot is intended to fulfill. For our purposes, the intent is to make an appointment. This intent is already configured in the blueprint as MakeAppointment. When you select the MakeAppointment intent for your bot, Amazon Lex builds a learning model for the bot so it can learn to recognize user input.

To accomplish an intent, a bot is looking for specific information in the input from the user. This information goes in what Amazon calls slots. For this bot, the slots would be:

  • What does the user want to accomplish?
  • Type of appointment to schedule
  • Date of appointment to schedule
  • Time of appointment to schedule

To obtain each of these, you need to configure a prompt that the bot will use to get the slot information. The corresponding prompts to the slots above could be:

  1. “Yes, how may I help you today?”
    a. User: “Book an appointment.”
    2. “What kind of appointment would you like to make?”
    a. User: “I’d like to get a haircut.”
    3. “When should I schedule your haircut?”
    a. “March 15th.”
    4. “At what time would you like me to schedule the haircut on March 15th?”
    a. “2:30 p.m.”
    5. “2:30 p.m. on March 15th is available. Should I go ahead and book your appointment?”
    a. “Yes.”

 At this time, the bot will generate a line with the information collected and send that to the scheduling software. It would look something like this:

           “AppointmentType:haircut Date:2019-3-15 Time14:30”

The bot would then confirm the appointment with the user:

“Thank you. You have an appointment for a haircut at 2:30 p.m. on March 15th, 2019.”

Imagine making a list of all the frequent requests your agents take daily. If those requests can be broken down into a specific set of variables (the slots) and questions to ask to get the necessary information (the prompts) you can build a bot to handle those requests.

See how Amazon Connect and the power of AI can take your customer satisfaction to the next level

Apply the latest in artificial intelligence and machine learning and make your life simpler. See how Amazon Connect (with a little help from Lex and Polly) can take some of the customer service traffic off your agents’ hands while taking care of your customers’ needs faster than ever. Learn more by getting in touch with CloudHesive at 800-860-2040 or through our online contact form.

Related Blogs

  • Exploring the Role of Amazon Web Services (AWS) Tools for DevOps Implementation in Cloud Projects

    Integrating DevOps best practices into cloud projects presents a few inherent challenges. With the help of AWS Tools for DevOps, processes can be streamlined for better cloud project management....

    Learn More
  • Optimizing Cloud Operations and Cost with DevOps Planning

    DevOps planning tips and tricks can help your organization balance operational efficiency and cost management. Even though implementing DevOps planning comes with many well-known benefits within the...

    Learn More
  • Key DevOps Trends: How They Shape the Future of Cloud Computing

    Staying on top of the rapidly evolving world of DevOps is challenging. Using prevalent DevOps trends can significantly impact project success in the evolution of cloud computing.  Considering the...

    Learn More