AI Agents in Customer Support
An increase in new sophisticated AI agents may finally rescue us from frustrating chatbot experiences
It was the week between Christmas and New Year's Day. I ordered a pair of shoes online - a last-minute treat. When the package arrived, I unboxed the product, and slid the shoes on my feet. Darn, wrong size. Undeterred, I decided to exchange them. However, the retailer had not included a return label in the box. Still optimistic, I thought, “No problem, I’ll just print a shipping label”—that’s when the real headache began.
Their website ensured a seamless return process, but when I tried to print a label myself, the link to generate one kept returning an error message. The customer support line went straight to voicemail. It was, after all, the one week out of the year that a sizable portion of companies’ workforces are off the clock. That’s when I was met with the brand’s ‘trusty’ website chatbot, conveniently named ‘Rob,’ as if this attempt to humanize the robot would drastically improve its ability to help:
Me: “Hey Rob, I need to return a pair of shoes I ordered. Can you help me print a shipping label with the correct return address?”
Rob: “Hi! I’d be happy to help. First, can you tell me your order number?”
Me: “Sure, it’s #1639403”
Rob: “Got it. Please hold on while I look this up.”
[ 3 minutes go by]
Rob: “Sorry, I couldn’t find that order number. Are you sure it’s correct?”
Me: “Yes, I’m looking at the confirmation email right now.”
Rob: “Okay, let me try again.”
[A few more minutes go by]
Rob: “I am still unable to find your order. For assistance, please contact our customer support team at 1-800-555-1234.”
Me: *begrudging sigh*
—
After seven or eight minutes spent with Rob, I had gotten nowhere. Like me, if you’ve ever bought something from a large company that has an online presence, you’ve probably experienced the ineffectiveness of conversational AI chatbots.
Enterprises often have numerous products, business lines, and a diverse customer base, which leads to customer inquiries that vary greatly and are highly nuanced for their support teams. Incumbent AI chatbots, like ‘Rob,’ were constructed1 using basic decision tree algorithms. They are only trained to support a limited array of customer inquiries and if they face a question outside their scope, they fail. Needless to say, ‘Rob’ eroded my trust in the vendor. However, new advances in AI model development and agentic AI are addressing this problem.
Automated support, but make it effective
In contrast to ‘Rob’-like chatbots, AI agents are characterized2 by their ability to reason through complex problems, complete actions without help from humans, and their ability to continually improve on a task. This technology is used to enhance the customer support experience, reducing how frequently consumers like me encounter a frustrating experience. With agentic solutions that have emerged in the past 12 months, enterprises are achieving improved deflection rates (the percentage of customer inquiries resolved without human intervention) between 70%3 to 90%4 -- subsequently driving greatly enhanced5 customer satisfaction.
Looking forward to the future of customer support:
More personalized customer experiences
Providing tailored experiences was customer experience leaders’ #1 priority according to a 2024 survey.6 New AI chat agents enable organizations5 to deliver these tailored experiences as the agents learn more about each customer through subsequent interactions. For example, they could greet the customer by their name, reference topics they mentioned in previous conversations, and more effectively answer their questions based on what products and systems the customer uses.
Driving continuity between sales and customer support
Additionally, as AI agents already have knowledge7 of a company’s offerings and learn about customers’ shopping preferences through interactions with them, I anticipate that AI customer support agents could also bridge sales and customer support, providing personalized experiences for customers from the pre-purchase value exploration stage all the way through post-purchase revenue retention and upsell.
When I was initially considering which shoes to buy, imagine what it would be like if I had my own personal sales assistant on the website that knew my style, size, and price preferences that could recommend me the best options. Further, as I bought a pair and found that I wanted to return them, I could interact with the same AI agent to help with post-purchase support inquiries, instead of ‘Rob.’. Clay Bavor, co-founder of Sierra, a startup founded in 2024 that offers an AI chat agent for customer support, remarked8 on this new reality: “What if in every call with your customers you had the equivalent of the most knowledgeable, veteran, grizzled support person who’s seen everything, and yet is still patient and friendly, and the sales associate who knows everything about your products because he or she has followed your company for two decades and knows everything, including the history of those products themselves?”
Conclusion
Although ‘I had to wait until the human support staff returned from the holidays to receive a return label for my shoes, it seems that AI agents are beginning to spare customers of the ‘Robs’ of the world (no offense to any real-life Robs, of course). In addition, this technology is in its nascent stages of bridging sales and customer support, providing continuity throughout the customer journey from pre-purchase value discovery through post-purchase revenue retention and upsell support. The future is looking a lot more, shall we say, supportive!
- Jack Marks | Clients & Market Growth | Deloitte Services LP