How Next-Generation AI Will Make Customer Service More Customer-Friendly

A brand’s approach to customer service is a make-or-break opportunity in developing lifetime customer loyalty, but the flood of inquiries inundating contact centers since the pandemic has left brands struggling to keep up with the pace of issues across channels. In fact, addressing agent burnout is a top two priority for contact center executives. While AI-powered chatbots and Interactive Voice Response (IVR) technologies have helped contact centers scale with customer self-service models, complex issues compound the demands on agents.

The advancements in AI with Large Language Models (LLMs) like OpenAI’s GPT technology are going to revolutionize customer service and redefine the modern contact center. In doing so, this new type of AI will help brands generate new engagement models and revenue opportunities.

It’s important to first establish how LLMs—often referred to as generative AI—differ from the type of AI that we’ve experienced in customer service to date. Trained in vast amounts of data, LLMs excel at summarizing, synthesizing and composing content. As OpenAI’s ChatGPT showed the world, these models can be interacted with in a very natural, conversational way.

This level of interaction is a significant divergence from the type of AI we’ve historically seen, where chatbots have been trained to respond to pre-designated key words and phrases, triggering workflows that return what is often generic information. This can be a helpful starting point in offloading repetitive inquiries, but it doesn’t account for the nuance of human language and corpus of relevant information to address the issue. In a controlled experiment, Brynjolfsson and colleagues found that customer service agents with access to a generative AI-based conversational assistant were able to resolve 14% more issues per hour than those without access.

In the enterprise, LLM-powered applications are customized to work in context of your business data and optimize your business process, in alignment with security and privacy requirements. The enterprise application of LLMs is revolutionary because it enables organizations to scale AI across vast amounts of information and business data to quickly surface insights, information and recommendations, and develop a draft response using context from historical interactions with the customer, such as recent purchases and preferences, to help compose responses in the context of an email or over a messaging channel like chat. Ultimately, it’s about getting to business insights and impact faster—and improving the customer and employee experience along the way.

The new LLM-powered customer service bot

LLMs give customer service chatbots incredible new and supercharged skills—natural language understanding, conversational user interface and the ability to quickly source and reason over internal and external information. In short, these new customer service bots are programmed to generate both answers and actions, making them more resourceful and useful.

The “copilot” era of generative AI represents a paradigm shift in chatbot programming. Previously, developers had to program a chatbot with both information and answers. The path from question to answer is pre-defined and limited by common customer questions and pre-determined answers. For example, “if the customer asks about business hours, provide X information.” Now, instead of pre-defined workflows, developers will design apps to connect to various sources of information and data for the chatbot to reference and create a contextual reply—such as the customer service handbook, the local weather, the scheduling calendar, the customer data platform and so on.

Consider the example of rebooking a cruise. Prior to generative AI, a chatbot might direct the customer to the company FAQ on rebooking policies and a link to the website for scheduling. An LLM-powered chatbot can recommend options for the customer by bringing in contextual insights about the trip, advise on dates with preferable weather and consult the schedule for availability and price ranges. This saves the customer significant time in researching new dates, weather, pricing and working with the cruise line to rebook over the phone, ultimately improving the customer experience.

LLMs will boost agent job satisfaction and efficiency

How does this new type of AI change the role of customer service agents if chatbots are able to handle a much higher case load? As LLM-powered bots enable more timely and satisfactory self-service, the most complex issues will still surface to agents, and this next-generation AI will help them resolve those faster, freeing them to focus on creating exceptional customer experiences, improving brand loyalty and creating space for more revenue-generating opportunities.

Customer service agents often need to navigate internal knowledge sources, team chats and external web sites to find the right answer for customers, but AI can quickly look across a variety of sources and synthesize a suggested answer for the agent. This can be especially powerful for new agents that have yet to learn where all the information is stored or which solution might be best. AI can also expedite an agent onramp to a new case by quickly summarizing the numerous data points and conversations that have already taken place. These summaries allow agents to save reading time and focus on resolution, resulting in an improved end customer experience.

The brand opportunity

As companies begin to adopt next-generation AI into their customer service departments, we expect this to have a positive ripple effect connecting service, sales and marketing organizations more seamlessly. Historically, contact centers have been cost centers for brands. But, with agents spending less time fighting fires, they can become part of your brand loyalty strategy. One-on-one engagement with a customer is a high cost to the organization, but AI-powered CRM and CDP solutions can help the agent not only solve the issue but upsell or recommend an additional item the customer may not have considered.

Leaders can set themselves up for this shift by readying their contact center for the future of LLMs by experimenting now—starting with external chatbots and agent-assisting copilots capable of grounding themselves on your trusted knowledge sources and helping with specific tasks like scheduling an appointment or drafting a customer response for the agent to personalize and send. With these advancements, the next era of customer service will undoubtedly be more customer-friendly.