Conversational AI for Customer Service: An Appealing Boost to Transformation

The advent of chatbots and voicebots a few years back was the first indication of both the impact conversational AI would have on customer service and the added value it would deliver. With the emergence of generative AI (GenAI) and the introduction of virtual assistants for contact center employees, this effect has been amplified dramatically. Used correctly, the technology makes customer service more attractive for all involved – from the customer to the agent to the head of the contact center. The prerequisite for this, however, is a far-reaching organizational and technological transformation of the entire customer service. 

Tackling this change is currently the biggest challenge facing customer service managers, according to a recent McKinsey study. Even more so as it goes hand in hand with ever higher commercial targets and customer expectations.

In this blog, we explain the key features and added value of conversational AI for customer service and the impact this technology has on jobs in the contact center. We show real examples of companies that have taken the step towards conversational AI and present best practices for successful implementation.

What is conversational AI?

Conversational AI is a subarea of artificial Intelligence that allows interaction with computer systems in natural language. It uses advanced algorithms and machine learning (ML) to decode speech patterns, understand the intention behind the words, and answer in a way that corresponds to human dialog. This technology is increasingly being complemented by generative AI, enabling the creation of dynamic and contextual content that makes interactions even more natural and adaptable.

What is a possible application of conversational AI in customer service?

From changing your address to taking out life insurance

To illustrate the potential benefits of conversational AI for customer service, let’s start with the story of Ms. Müller:

1. Ms. Müller has been a customer of Plus Insurance for many years. She has just moved into her new house with her husband and newborn child. It’s 7 pm and she realizes that she hasn't informed her insurance company of her new home address yet.
2. She calls the insurance company, a voicebot answers, notes the new address, and immediately asks whether Ms. Müller would also like to check the cover of her household insurance with an agent.
3. She would like to, as such a big change in life also has an impact on the cover. She gives the bot a few key details and immediately arranges a callback appointment.
4. At the scheduled time, an agent calls Ms. Müller. The agent already has the relevant information at hand thanks to the data collected by the bot. With the support of the Agent Co-Pilot (see image below), she can efficiently and systematically check and adjust the cover.
5. The agent feels great because she can provide comprehensive advice. She asks whether Ms. Müller and her husband have already considered life insurance after the birth of their child, and whether she would be interested in making an appointment for a consultation. Ms. Müller agrees. The Agent Co-Pilot guides the agent through the most important questions and schedules an appointment with a local advisor.
6. The advisor is notified and receives both an automated summary of the conversation and the relevant information about the Müller family. In preparation, he uses Advisor Assist (see image below) to draw up a proposal for a life insurance policy with various investment strategies based on the Müllers' profile. After the conversation, the Müller family takes out a life insurance policy.

Ms. Müller's story illustrates that conversational AI goes far beyond the mere automation of customer inquiries. Its added value extends across the entire customer journey: Conversational AI improves efficiency, increases customer and contact center employee satisfaction, strengthens customer loyalty, and boosts sales. More than enough reasons to boost conversational AI for your customer service.

Conversational AI solutions and their added value for customer service

Conversational AI platforms include advanced technologies, such as virtual assistants that can have human-like conversations with users thanks to the integration of natural language processing (NLP) and machine learning (ML). Modern systems such as generative AI expand these capabilities. They incorporate generative artificial intelligence that can not only respond to human speech, but also proactively generate content and respond to complex requests in various media forms (including text, image, and potentially also audio or video). These AI-supported systems offer an intuitive and dynamic user experience that simplifies and personalizes interaction with technical devices.

In Ms. Müller's story, various complementary tools cooperate along the customer journey:

The diagram shows how these elements interlock and complement each other. While the voicebot processes part of the request, such as the change of address, it initiates and prepares the dialog with the agent by suggesting an appointment.

Based on the collected information and the data from the CRM, the Co-Pilot in turn supports the agent in improving the quality of the dialog – both in terms of efficiency and advice.

Finally, «Advisor Assist» enables the advisor to provide personalized advice.

Specifically, voicebot and Co-Pilot work as follows:

Voicebot

The voicebot, also known as the voice assistant, on the one hand enables the automated processing of inquiries and, on the other, the preparation of a call with an agent if it is unable to fully process the customer's request. In both cases, it relieves the contact center agent. The main functions of the voicebot can be summarized as follows:

• Self-service
  The voicebot automatically analyzes and answers frequently asked customer queries and takes over complete business cases with the involvement of peripheral systems (e.g., CRM, ERP).
• Intelligent callback arrangement
  The voicebot schedules the ideal time for a callback with the customer, records the reason for the call, and shares the information with the agent for preparation.
• Routing
  The voicebot identifies the customer and recognizes their request, which it automatically forwards to the right contact point with the relevant information.
• Biometric authentication
  With the customer's express consent, their voiceprint is recorded and checked during the next call for authentication.
• Protocols and analysis
  Conversations with bots or employees are transcribed and analyzed to gain valuable insights.

The added value of the voicebot:

• Greater customer satisfaction thanks to 24/7 availability, the elimination of waiting times, and the direct handling of queries
• Higher employee satisfaction thanks to a lower workload and fewer repetitive questions
• Increased efficiency and lower costs thanks to the automated processing of customer queries or efficient preparation of customer consulting

Co-Pilot (Agent Assist)

The Co-Pilot, often called Agent Assist, serves as a virtual assistant that supports the agent by handling administrative tasks and providing immediate AI-based guidance. This occurs along the following four phases:

In this phase, the relevant information is obtained from the customer, analyzed, and forwarded to the right contact point («triage»). This task is performed by a voicebot instead of a classic IVR (interactive voice response) and comprises the following main functions:

• Intent recognition

• Triage

• Intelligent callback management

• Identification and authentication

Takeoff: Information overview to prepare for the consultation
This phase takes place immediately before the customer dialog. The aim is to provide the agent with the relevant information in concise form so that they are optimally prepared for the conversation. The information is displayed in the form of a cockpit, either as part of the contact center software or as a separate window. It includes:

• Intent of the current call if already known
• Customer profile (360-degree customer view)
  • Master data
  • Purchased and still open products
  • Past customer interactions
  • Notes and recorded tasks for the customer
• Contextual information such as the customer's waiting time or NPS (net promoter score)
  

In flight: Real-time support during the call
In this phase, the Agent Assist function enters the stage. The Co-Pilot listens in on the conversation and provides its assistance, supporting the agent during the call with the following, freely combinable functions:

• Live transcription of the dialog
• Simultaneous translation (e.g., from Italian to German)
• Live biometric authentication (if not done before)
• Analysis of the customer request with intent recognition and extraction of information (for example, the customer number)
• Next-best action recommendations for
  • conversation techniques
  • quality assurance
  • up/cross-selling
• Information search and provision (e.g., via RAG)
• Live sentiment analysis
• Task queuing to record tasks for the follow-up to the customer call

Landing: Follow-up to the conversation
This phase takes place after the consultation has been completed. It is designed to help the contact center agent complete typical follow-up processes as efficiently as possible. This includes such tasks as:

• Summarizing the customer dialog
• Processing the task queue (e.g., e-mails, sending material such as brochures, and updating master data)
• Input support when entering data, for example, in a CRM
• Recording customer and agent feedback
• Dialog analysis and metrics for evaluation

The added value of the Co-Pilot:

• Optimal preparation for consultations through a 360-degree view of the customer with relevant contextual information in real time
• Improved customer experience by allowing the agent to focus on value-adding activities
• Enhanced dialog quality thanks to the recommendation of optimal next actions (e.g., checking data quality, cross-selling and upselling)
• Increased efficiency by taking on administrative tasks (e.g., summarizing, editing e-mails and reports)
• Easy fulfillment of quality and compliance requirements thanks to systematic documentation
• Increased employee satisfaction thanks to support, expanded tasks, and continuous improvement

Note: The use of Co-Pilots in contact centers is yet in its infancy. However, companies are gradually beginning to introduce such systems with success. The described Co-Pilot functions can be implemented step by step and as required.

Take CSS Insurance, for example: As a first step, they implemented a GenAI-supported information search for products and services, and integrated it into the existing contact center software and CRM.

The impact of AI on contact center jobs

With the introduction of conversational AI in customer service, the role of contact center agents and advisors will change significantly. This comes with a paradigm shift:

1. Thanks to the automation of standard tasks, employees will focus on higher-value and interaction-oriented tasks.
2. Humans and AI will work ever more closely together and form a «collaborative intelligence».

Real-life examples of conversational AI at companies

The use of conversational AI for customer service can’t be stopped. It is playing an increasingly important role in large companies, as the following examples show:

• CSS
  CSS, one of Switzerland's largest health insurers, receives about 1.7 million phone calls every year. Customers' questions range from information on premiums and benefit statements to claims and supplementary insurance. To avoid waiting times, CSS has introduced an intelligent callback solution. The customer does not have to wait but receives a callback at the scheduled time. Another benefit: The service center employees are prepared for the conversation. This further increases service quality and thus customer satisfaction. 
  
  This solution is the first step towards systematically introducing further voicebots (for identifying customers, applying for or adjusting policies), as well as Co-Pilot functions such as a CRM-integrated search via a retrieval-augmented generation (RAG) solution.
• Swiss property insurance company
  A leading Swiss property insurance company has integrated conversational AI into its helpdesk and customer service. One service is the self-service voicebot for reporting claims. It enables automatic processing of claims, which simplifies and speeds up the process for the customer. Another service is the intelligent IVR. Thanks to improved triage, over 92% of all calls are forwarded to the right point of contact. 
  
  The self-service voicebot for Windows Password Reset has also been introduced to relieve the IT helpdesk. It verifies the caller number and secret information before resetting the password. This offers greater security than manually processed calls.
• A large Swiss retail bank
  One of Switzerland's leading banks systematically uses conversational AI for its various customer channels. The focus here is on improving customer experience by rapidly processing requests, recognizing potential for cross-selling and upselling, and supporting agents in customer care. The bank consistently equipped the relevant channels – public website, e-banking, and customer service by telephone – with chatbots and voicebots to handle simple customer concerns in self-service. For more in-depth queries and customers with cross-selling and upselling potential, agents responsible for customer care are called in.

Best practices for introducing conversational AI

When a company opts for conversational AI, it also initiates a paradigm shift. A systematic approach to implementation is therefore key. The following best practices will help you make the switch in six steps:

1. Vision and strategy coordination
   Define a clear vision for the implementation of conversational AI that aligns with the corporate strategy and sets concrete goals.
2. Analysis along the customer journey
   Identify potential use cases for conversational AI along the customer journey. Estimate the economic added value (higher efficiency, lower fluctuation, more sales), as well as the qualitative added value (higher availability, convenience, and faster problem-solving for customers, higher data quality). 
   
   Now put these values in relation to the complexity of implementation, for example, in the form of a matrix, and use it to develop a roadmap. Start with an automation case (chatbot or voicebot). This requires less training or organizational changes and serves as an introduction to raising awareness within the company. Then systematically add additional use cases according to the roadmap.
3. Inclusive design
   Involve all relevant stakeholders in the design process and create the solution according to their needs. For example, involve contact center agents in the development of a voicebot, as they are affected directly or indirectly.
4. Selection of the AI platform
   Choose an AI platform that allows you to implement individual use cases separately but is also scalable in terms of use case diversity and volume. A multichannel platform that supports different formats (e-mail, tickets, chat, voice) and forms of interaction (automation of the request or human support by AI) makes it possible to easily reuse data and adapt use cases from one channel to another (e.g., from voicebot to chatbot).
5. Capacity building and implementation
   Develop the necessary skills in-house or supplement them with external support. Form a well-coordinated, multidisciplinary core team (UX, architecture, business representatives, data scientists, GenAI and MLOps specialists, etc.) that systematically implements the use cases. Consider aspects such as conversational user experience (ConvUX), security and data protection, and operational requirements right from the start – and never lose sight of your vision.
6. Learning and optimization cycle
   Establish a clear cycle for learning and optimization with defined key performance indicators (KPIs). A conversational AI project often only really begins after launch. Optimize the solution on a continual basis, whether in terms of user experience, data quality, or model performance. 

Convince your customers with fast, personal service

Conversational AI is increasingly becoming the standard for contact centers. Such advantages as high availability, well-prepared and personalized conversations, greater satisfaction among employees, and greater efficiency are compelling indeed, illustrated by companies that have already taken the plunge.

If you want to drive digital transformation forward, conversational AI is a must. But it is crucial that you take a careful and systematic approach to the introduction. Of course, there are a few challenges along the way. After all, we are talking about a far-reaching organizational and technological change of your entire customer service.

Would you like to know in detail what requirements a company should fulfill to offer successful and scalable conversational AI services? Check back soon, we'll tell you about it here.