logo_header
  • Topics
  • Research & Analysis
  • Features & Opinion
  • Webinars & Podcasts
  • Videos
  • Event videos

Catalyst taps AI to improve the customer contact center

The Cognitive Contact Center Catalyst project, is using AI and voice recognition to cut contact center costs and increase customer satisfaction.

Arti Mehta
06 Sep 2018
Catalyst taps AI to improve the customer contact center

Catalyst taps AI to improve the customer contact center

Artificial intelligence (AI) and voice recognition are key to cutting contact center costs and increasing customer satisfaction. A TM Forum Catalyst proof of concept, the Cognitive Contact Center, is laying the foundations for contact centers to use these tools.
The contact center channel represents a substantial operating expense for the customer service area of many operators, and can often struggle with new revenue generation and customer satisfaction. Operators need to take advantage of the fact that this is the first point of contact for customers, and understand the right type, level and channel of engagement can drive profit and customer advocacy.

The Catalyst, which was demonstrated at Digital Transformation World in May and will be shown again at Digital Transformation North America in Dallas later this month, aims to surpass the contact center issues by using natural language processing to personalize automated services (virtual agent and social media) for customers. The project is championed by Telefónica and includes CA Technologies, Everis, Microsoft, Nokia and Verbio as participants.

Researching customers’ preferences


Everis began by conducting some research for the team, surveying 3,600 customers of communications service providers (CSPs), finding that telecom customers expect to be able receive customer service across all digital channels, but at the same time, consider the contact center as the preferred channel for customer care interaction.
In separate research conducted by Everis, the company found that customers seek care at every point of their journey which requires CSPs to be proactive and anticipate issues, and prevent or resolve them quickly.
“We need to find ways to improve engagement with native digital customers – they want to talk to a virtual agent,” says Marta Besteiro Martínez, Enterprise Architecture Senior Manager, Telefónica. “We also need to focus on customer interaction in this channel.”

The project is using AI for two primary reasons:

  • Improving customer experience – natural language processing is an area of computer science concerned with the interactions between computers and human languages. It allows customers to have a more humanized and natural interaction when interacting with digital agents or bots. In addition, a more personalized experience comes from being able to process huge amounts of customer profile information and identifying the best next step.

  • Reducing operational costs – shorter and more efficient interactions using a high level of automation will improve operational indicators such as ‘average handling time’ or ‘first call resolution’. This in turn reduces the amount CSPs spend on human resources and infrastructure.


Telefónica’s transformation


Telefónica is in the midst of a massive digital transformation project to transform not only the way engages with customers but also operations. The company is working to transform its mobile app and boost its social media channel but knows that it must also transform call center channels, a major part of its service and engagement activity.

Telefónica wants weave a more personalized customer experience through all these channels by coordinating conversation and interaction flows. The goals are for conversation to be less stilted and to produce fewer incorrect customer interactions. The company also wants to reduce time spent on operational tasks. For example, wait times can be reduced by using interactive voice response technology, which means less staff time is taken up with common, menial queries.

The left side of the image below shows an example of a contact center agent using a lot of resources to interact with the customer, such as business support systems (BSS) and management tools. The Catalyst’s approach is to place a virtual agent at the beginning of the customer’s service journey and have the virtual agent use BSS to manage customer interactions using cognitive technologies which use natural language processing technologies (NPL) to understand the customer intent and to predict the next-best- action or operation. When issues can’t be resolved by the virtual assistant and must be transferred to a human agent, the human agent has a unified service desk to get a 360-degree view of the service and interactions the customer has had thus far, including the conversation with the virtual assistant.

Who participated?


For the purposes of the Catalyst demonstration, each participant provided a different piece of the overall solution, and the TM Forum Business Process Framework (eTOM) was used to map conversational flow processes.

everis

everis provided its Virtual Agent (eVA) solution, which integrates three major components:

  • Broker – This module orchestrates all conversations and interaction flows. It uses the cognitive engine to identify the issue and determine the response to the customer. It accesses the customer’s information through using CA Technologies’ to understand the context, decide the next action to take and complete the answer to the initial query through automated online tools within the solution minimizing complexities such a creating code to enable these activities, as they already exist within the solution.

  • Content Manager – This module stores all the content needed provide the answer to the customer. It can adapt the content to the channel that the interaction comes through (for example, providing short, sharp responses when using social media).

  • Analytics – This capability provides statistics about virtual agent usage and contributes to continuous training of the underlying cognitive engine.


Verbio

The company’s speech-to-text and text-to-speech technology is used to enable natural and intelligible communication with users. Text-to-speech expresses feeling and emotion within the user interaction, while speech-to-text allows any verbal interactions to be recorded as text in the eVA’s orchestrated conversation/interaction flows.

Microsoft

Microsoft’s cognitive engine is based on LUIS (language understanding intelligence service), a machine learning-based service that builds natural language into bots (as well as apps and IoT devices). The system identifies the intent of any conversation by analyzing how sentences are constructed and applies rules to answer the customer with humanized responses, making the customers feel that they are interacting with a real human agent.

CA Technologies

CA’s API management tool extracts user information from the business support system (BSS) about the query at hand. It also provides an integration layer to connect with the social media channel. For this purpose, CA exposes a set of TM Forum Open APIs: Party Management API, Resource Inventory API, Product Catalog Management API, Product Inventory API, Billing Management API, Service Problem Management API, Service Problem Management API, Usage Consumption API and Product Order Management API. The APIs were implemented exactly as defined without any modifications.

Nokia

Nokia’s service management platform diagnoses any technical issues the customer has. For example, if a customer cannot navigate through internet, Nokia diagnoses the problem, providing information about what the problem is such as whether it’s a commercial issue to do with the customer’s plan, or that they just have the wrong navigation parameter. Nokia provides this information to the virtual agent whereby the conversational flow adapts accordingly.

How does it work?


The graphic below shows the interaction among participants. The modular architecture is based on the three typical layers of a digital IT architecture and in this case include engagement, cognitive and core layers.
At Digital Transformation World, the team was able to demonstrate a number of use cases to visitors including:

The journey of a customer calling the center with connectivity problems:
A virtual agent proactively calling a customer who has been detected as being in another country and has activated roaming, to suggest the next best step for the customer:
A customer who doesn’t understand their billing information and needs to be passed on to a human agent by the virtual agent:

Next steps


In future iterations of the project, the team intends to explore:

  • More complex use cases the development of robotic process automation.

  • Using data intelligence to increase outbound interactions, and upselling and cross-selling products as appropriate.

  • Integrating an identity management mechanism such as voice biometric authentication.

  • Enabling user lifecycle management, allowing users to access to customer services and manage them.

  • The omnichannel experience by extending the channels the customer may interact and ensuring a unique an integrated experience for the customer.

  • Augmented reality, allowing the customer to interact with physical devices (for example, a home router) through the support of a virtual agent.

  • New revenues streams – finding alternative ways for CSPs to generate revenue by potentially providing the cognitive contact center as a service.


Watch team members discuss the project in this video recorded at Digital Transformation World: