In a recent demonstration of AI capabilities, Google DeepMind AI system AlphaStar rose to the challenge of defeating several top pro video gamers in the Real-Time Strategy game Starcraft II. Starcraft II is a game that involves gathering resources and building military units that battle against another player or players.

Many large companies have already adopted some form of automation into their Customer Service chats and help forums. The smallest of companies can also implement AI into their current customer service department.  AI is a powerful element that can be used in a customer service department to lower costs and provide a quicker and more meaningful customer service experience. This is an example of deploying the AWS Lex chat bot into a fictional dental practice named Happy Dental Care.

Identifying and Setting Goals

The first step of any organizational department is to clearly identify what performance goals are to be met. This starts with the current department KPI metrics as a baseline. The Average Call Time below is an example service KPI for a dental office practice. The goal is to improve efficiency of the department by eliminating repetitive chat that can be automated and by allowing employees to focus on backend processes.

Happy Dental Care

Average Call Time: 3 minutes

If 2 additional minutes are added for idle time per call, then a total of 5 minutes per call can be assumed. Approximately every 12 calls will cost 1 hour of labor. Many companies who are just beginning to implement AI often think of it as a tool to deflect calls and chats. This is true, but the majority of the value from it is automating the processes that customer service employees have to perform manually.

Start by signing into AWS and navigating to Amazon Lex. Click on Create and you will be presented with the menu in Figure 1.

Figure 1

Select the Sample Bot named ScheduleAppointment and rename it. This preconfigured sample bot is designed to schedule appointments.

There is already a collection of Sample Utterances that are statements of intent.  Add additional statements as needed that are specific to your business. For this example I added different variations of the phrase, "Schedule an appointment."

Select the Sample Bot named ScheduleAppointment and rename it. This preconfigured sample bot is designed to schedule appointments.

There is already a collection of Sample Utterances that are statements of intent and these can be seen in Figure 2. Add additional statements as needed that are specific to your business. For this example I added different variations of the phrase, "Schedule an appointment."

The Slots section is the data the user must provide to fulfill the intent. These follow a logical flow a patient would use in creating an appointment. An appointment type, date, and time need to be provided. These required data fields are already present in this template.