How to Think About AI for Customer Support

AI, or artificial intelligence, seems to be everywhere these days. In movies, AI is often portrayed as the villain deciding what's best for people and — in the most extreme cases — deciding that people aren't needed at all. Thankfully, this isn't our reality yet.

In the real world (or "IRL," as my kids say), AI first came to prominence with IBM Deep Blue winning at chess and later when Watson became a Jeopardy champion. With games as a proving ground, AI has been evolving quickly and is now widely deployed in many parts of our daily lives. It helps Amazon optimize warehouse and delivery operations, Netflix recommend what shows to watch (and create), and JP Morgan execute complex trading strategies.

In each of these areas, AI is being used for its ability to find patterns, evaluate options, and suggest actions more quickly than a person can using traditional analytics. For a long time, the world of customer support was not a focus for AI. Customers used help sites to look for answers to questions or called customer support agents to help them resolve issues. While the systems customers and agents used became more sophisticated, the approach they followed was still analog — it relied on the intuition of the user to know where to go.

Becoming Mainstream in Customer Support

In the past couple of years, AI has exploded onto the customer support scene, especially in the world of agent-delivered assistance. Support leaders are inundated by an overwhelming array of solutions that promise that AI will transform how agents are staffed, coached, and managed, while influencing which calls they get, what products and services to recommend, and how best to deal with angry customers.

Covid-19 continues to make staffing contact centers a challenge, so AI-powered virtual agents are even starting to take over for live agents by handling simpler questions and requests. For many support leaders, the reality of what AI can do hasn't lived up to the promise and organizations are suffering from AI confusion, fatigue, and disillusionment. Regardless of these frustrations, it's clear that AI in support is here to stay.

So what should a great AI experience for support look like? It starts earlier in the customer's journey than support leaders might traditionally believe — with search. Customers should be able to find answers to most of their questions without having to contact an agent. This means creating the content users are looking for, ensuring it's accurate, and making it easy to find. AI helps this process by understanding the underlying meaning of a user's search, no matter which words they use or how they're ordered. AI-powered search delivers the content best suited to answer a particular question, learns with each use which content delivered the best self-service, and recommends content that answers the next question a user is likely to ask.

AI's Role as a Complementary Piece

In the contact center, Support leaders have found that AI still has a ways to go before it meets customer expectations for a natural, empathetic, and effective support experience. Accordingly, AI is finding its niche in augmenting the agent experience, rather than replacing it. Because of Covid, many support teams have been forced into work from home and distributed work environments. In practical terms, this means that supervisors and fellow agents are no longer readily available to answer questions and provide real-time guidance on how to help customers. As support leaders continue to work through these challenges, AI can create a positive impact by helping to arm remote agents with the information they need to solve inquiries, making them feel empowered instead of discouraged and disconnected.

Additionally, AI tools are being used to monitor an agent's conversation with a customer, identify when the interaction isn't going well, and recommend how to adjust the agent's tone or messaging to deliver a positive outcome. What's important to understand here is that it's not "Big Brother" watching an agent's every action and looking to penalize them for poor performance. Rather, successful AI deployments are based on trust, communication, and a desire to help agents perform difficult tasks more effectively and confidently.

As AI for support continues to evolve, we'll see it start to match the level of effectiveness and transparency we see in the best consumer situations. It will understand the personal preferences, history, current context, and desired outcome from the beginning of the support experience, and tailor the right content and messages for each situation. An AI-powered virtual agent will be able to handle more complex tasks with ease and know when to hand the interaction off to a live agent, without the customer asking. Eventually, AI will even be able to create the content, workflows, and coaching strategies that work for each individual agent so they can deliver the best experience possible.

To learn more about Yext's AI-power support solutions, click here.