We are all familiar with customer service chatbots. Whether it's via web messaging or SMS conversations, chatbots seem to be everywhere. This prevalence is in part due to the COVID-19 pandemic and the rise of remote interactions between customers and businesses and also because customer call centers are a huge cost center to companies. Chatbots were devised to keep expenses down.
Luckily, according to Salesforce, 69 percent of consumers prefer to use chatbots because they provide instant responses. From a business perspective, 57 percent of executives in a study by Accenture said that chatbots bring significant ROI with minimal effort.
Even so, the chatbot experience has often been far from extraordinary, and generative artificial intelligence is going to change that.
Still, most chatbots today are for customer FAQs.
When it comes to company FAQs, automated chatbots can be very useful to customers because they can automatically mine companies' knowledge resources and report back store policies, for example. No one really wants to have to wait on hold for a customer service representative to answer basic questions.
However, when it comes to resolving more complex issues, chatbots usually don't cut it. Most aren't intelligent enough to help customers with multi-part or time-sensitive questions. Chatbots too often resolve the issue by sending a how-to article. That can be maddening and far from action-oriented.
When a customer's problem is a bit more complex than an FAQ, such as "I think I might miss my flight, now what?" traditional chatbots are less than ideal.
But what if your business had a generative AI-powered chatbot—better described as an AI agent—that could actually resolve customers' issues by taking the same action a human agent would?
A sample conversation might look like this:
Customer:"I'm running late for flight ADA7190; am I going to make it?"
AI Chatbot: "Your flight's departure is delayed 45 minutes. If you don't make it, the next flight with seats is leaving at 12:15pm EST. Let me know if you would like me to book you on the later flight."
An action-oriented response like this would be refreshing and unexpected to most customers and is leaps and bounds better than sending an article on how to rebook your flight. This type of inquiry traditionally would have required a human agent to coordinate with multiple back-end business systems, if they can even connect with one another in time.
Often, a human simply isn't available to intervene at all, especially with most companies trying to deflect customers away from agents due to high costs. Ever try to call an airline or an online travel broker? Good luck getting someone live on the phone.
But what if an AI agent could tell you your flight is running 30 minutes behind, giving you more time and offer to rebook you on the next available flight if you still think you'll miss it?&
That would be a game changer in customer service. People might actually learn to love customer service again and prefer dealing with AI agents.
The AI Agent: Best Practices and Back-End Needs
Generative AI can transform a scripted chatbot into a customer service superstar. Here's what you need on the back end:
- Formalized, up-to-date company knowledge bases that are ready to be communicated externally.
- Experienced customer service agents who can be elevated to AI agent coaches (also known as bot managers), facilitating integrations with different business systems and typing in plain English which information can be retrieved and when the bot should use it.
- Instead of hard-coded answers, every reply should be as unique as the customers to whom it's sent. Specific replies should be generated live from pre-existing, trusted sources, like your knowledge base or help center.
- Continuous feedback through coaching so the AI agent can learn and improve its performance.
- Domain expertise, iterative learning and testing, because an AI's generated responses must be engineered and deployed in a safe and reliable manner.
So, it's time to break the script. Are you ready to let your chatbots enter a new era and mimic the reasoning ability of live agents through generative AI? The time is now. Your customers will thank you for it.
Kristal Lam is senior director of product management at Ada.