The article considers automation of modern call centers and offers the use of Amazon Connect as the solution for this issue. The article also examines the components of Amazon Web Services that can help to operate a powerful and flexible customer service system on the basis of Amazon Connect.

Customer service is essential for the success of any business. In order to provide telephone customer support, many companies create call centers. By 2020, call centers will satisfy up to 85% of all the requests without connecting to an operator. The basis of any call center is processing of incoming calls queue and redirecting them to any available operator. More complicated systems support more complex routing and interoperability, including the ability to create custom phone trees among other integrated voice response systems (IVRs – Interactive Voice Response). Respectively, when IVR’s scale increases, its maintenance becomes more and more expensive. Nevertheless, only 7% of the solutions using IVR correspond to the level of service provided by a human operator.

In companies where a number of human operators exceeds 50, there might be some problems. It is necessary to build several software solutions that can maintain the contact center infrastructure, including such issues as queues, working schedule compilation (considering personal skills of each operator, such as languages, experience, etc.), outgoing calls. With a progressive expansion of a call center, engineers tend to enroot all new solutions into the existed system architectrure in order to solve all the new problems occurred. As a result, no one cares about a modernization and maintenance of older solutions. Another problem with these systems is the expansion and support of the infrastructure and technical base.

A solution presented in this article is based on the use of AWS (Amazon Web Services), which provides a possibility to create a cloud-based call center located on the Amazon Connect service. The advantage of this cloud-based service is that the company does not need to have a specialist who would maintain the entire technical base of the call center. Amazon takes care of all the tasks of the scalability of the system. Also, Amazon Connect has a close integration with a lot of other services for building flexible and scalable systems provided by AWS. Each component has a detailed documentation, is supported and expands its capabilities constantly.

AWS Integrated Components

  • Amazon Lambda - allows you to place and execute code in the cloud using Amazon's computing power. This allows you to receive and use information from the database during a call and dynamically influence the conversation scenario.
  • Amazon Lex - allows you to build interactive interfaces with the use of voice or text for interaction with a client during a call. This technology allows the client to simply say what he needs, instead of listening to menu items and thinking about what number to press in order to get the required assistance.
  • AWS Cloud Watch - provides real-time logging and monitoring of metrics of the call center. For example, the number of calls per minute, the number of rejected or missed calls, errors occurred when connecting or during a call.
  • Amazon Kinesis is a platform for events streaming. Data is stored in JSON format and contains information about calls and operator activity. Additionally, this data can be transferred to AmazonElasticsearch for a graphical interface analysis.
  • AWS Identity and Access Management is a service for assigning access rights and policies to data. It creates a hierarchy of access to the AmazonWebServices account.

Call Center Development Using Amazon Connect

The key concept of Amazon Connect is call script (Contact Flow), that means a sequence of blocks that are used during a call. Developing different scenarios, the system gets an opportunity to handle the incoming call and dynamically change the course of events during the conversation. Let's look at the example of the scenario of the automated outgoing call to a customer without the operator's participation.

Amazon Connect Contact Flow

When connected to the customer, the first three blocks of the scenario (Set logging behavior, Set call recording behavior, Set voice) set the logging, recording of the conversation, and the voice type the bot will communicate with the client, accordingly. Get customer input uses the Amazon Lex bot. The bot sounds the text by the template, substituting the customer’s name and all the necessary information that was obtained from the database during the connection established using AWS Lambda. Then the bot listens to the client's response and analyzes it. At the output of the block, we receive a positive or negative response from the client. If the customer has confirmed the order, the confirmation is saved to the database (Set contact attributes block), the summary text is read out, and the system hangs up. If the client does not confirm the order, the system redirects the call to the queue or to any free call center operator.


With proper scheduling of different possible scenarios of outgoing and incoming calls, you can process 20-50% of inquiries without any operator participation, that can be considered a good result. In addition, speech synthesis systems have been developed so well in our time that about 27% of clients are not sure whether they communicate with a robot or a person. The advantage of using Amazon Connect is that it is possible to flexibly configure and compose various services from a single Amazon Web Services ecosystem to achieve the needed result. Such a solution frees operators from the monotonous job, saves human resources that are expended, allowing them to be sent to places where human intellect is really needed.