Amazon Web Services Inc. said today it’s adding more machine learning capabilities to its cloud contact center platform Amazon Connect, making it easier for companies to forecast demand and analyze conversations in real time.

Launched in 2017, Amazon Connect is an omnichannel cloud contact center service that businesses can use to enable customer service and customer engagement. Customer service representatives can respond to phone calls or chat inquiries from customers in the same manner as if the contact center infrastructure were set up and managed on-premises.

Announced on day two of AWS re:Invent 2022, the first of the updates pertains to the general availability of the forecasting, capacity planning and scheduling features that were revealed in preview in March 2022. With the update, contact center managers now have the tools at their disposal to accurately predict demand and determine optimal staffing levels, while carefully organizing agents’ schedules to ensure they have the right experts on hand to deal with customer’s calls.

The Forecasting, Capacity Planning and Scheduling can now be enabled through the Amazon Connect console for all customers in Amazon’s US East (N. Virginia), US West (Oregon), Asia Pacific (Sydney) and Europe (London) Regions. The new functionality relies on a machine learning model that’s tailored to analyze contact center operations and predict future contact volume and average call handling times, based on historical data. The forecasts are expansive, covering inbound, transfer and callback contacts through both voice and chat channels.

With the long-term forecasts it enables, managers can leverage planning scenarios and metrics such as maximum occupancy and daily attrition to predict more accurately how many agents they’ll need to meet service level targets on any given day. The long-term capacity plans they create can then be shared with stakeholders. Another benefit of the forecasts is that they enable more optimal scheduling, AWS said.

The forecasts can be used together with shift profiles, staffing groups, human resources and business rules to create more efficient schedules that are optimized to hit a specific service level or targeted average speed of answer. Schedulers can then review and edit the schedules they create, if necessary, then notify agents and supervisors whenever their schedule changes.

Constellation Research Inc. analyst Liz Miller said the new capability is a good example of how AWS is focusing its AI efforts in service of boosting operational efficiencies for customers.

“The new scheduling functionality gives managers and teams the opportunity to change, adjust and optimize schedules and resources in real-time,” she said. “Normally, schedule and resource management can be a painful task that causes lots of headaches due to all the moving pieces. So this is actually a terrific place to deploy AI and ML models that can make automated, real-time adjustments to schedules based on business rules, demand, plus labor rules and requirements.”

In a second update, Amazon Connect is getting beefier analytics capabilities with the introduction of Contact Lens for Amazon Connect. Available in all the AWS Regions where Contact Lens’ conversational analytics for speech is already available, the service adds conversational analytics to all chat-based interactions. This capability was already available for calls, and relies on natural language processing models to help agents better assess what customers need and their sentiments.

It can also help to redact sensitive customer information and monitor agent compliance with company guidelines. Other features it enables include contact search, which makes it possible for managers to identify chats where customers had issues based on a specific keyword. Finally, it powers a new chat summarization feature that uses machine learning to classify and highlight key parts of a conversation, such as their issue, the outcome or an action item.

In addition, Contact Lens for Amazon Connect introduces evaluation forms for agent performance in preview in all regions where the service is available. According to Amazon, this can help contact center managers to evaluate agents based on specific criteria, such as their adherence to required scripts or compliance with sensitive data collection policies.

Finally, Amazon said, the agent workspace is getting a new step-by-step experience that guides agents through the process of resolving customer issues. The agent workspace is a unified application that provides agents with all the tools they need to handle customer’s inquiries. When they accept a call or respond to a chat, it will provide them with up to date customer information, knowledge base articles and real-time recommendations on how to resolve issues.

Lastly, managers will be able to use a new no-code, drag-and-drop interface to create custom workflows for their agents, guiding them through the resolution of common issues with one click actions. The agent workspace updates are available in preview now in the US East (N. Virginia), US West (Oregon), Asia Pacific (Sydney), and Europe (London) regions.

Miller said today’s updates are in line with Amazon’s broader goal of delivering ambient intelligence throughout its products and services.

“We can expect that a good deal of the announcements at AWS re:Invent will be focused on this goal,” Miller said. “Amazon is looking to identify places where data, workflows and AI can be inserted into a business function, such as a contact center, to fluidly work on behalf of agents. It’s looking to do this without disrupting anyone’s work, yet deliver tangible and immediately accessible value, with the technology itself seamlessly melting into the background. The tech’s very purpose is to fade away and make it possible for agents to have more meaningful and hopefully more effective engagements with customers.”

Featured image: DCStudio/Freepik; graphics: Amazon