AI agents are taking their baby steps, and it’s only a matter of time before they walk with confidence. Take, for example, Amazon’s Alexa+ voice assistant, which can take on the role of a virtual agent. Alexa+ can book a reservation through OpenTable, reserve an Uber for a friend and text people your plans, according to Amazon.
Alexa+ is currently rolling out select devices ahead of a wider release in the coming months, but the implications extend beyond simplifying dinner plans. Once agentic AI reaches its full autonomous potential, it could change how customers expect companies to handle their questions and problems.
Agentic AI — third-party, autonomous systems capable of taking action on natural language queries — are a natural outgrowth of existing virtual assistants, according to Patrick Quinlan, senior director analyst in the Gartner customer service and support practice.
Companies are already exploring AI agents for customer service. For instance, some hotel chains are adopting AI-powered front desk concierges at certain locations. But the bigger potential may be tools like Alexa+, which puts agentic AI directly in the hands of customers.
The evolution of agentic AI applications could realign customer service expectations within the next five years as consumers outsource their problems to the AI in their pockets, accoridng to estimates from Gartner.
“You can imagine if someone has the choice between calling customer service, waiting on hold, explaining their issue, getting transferred, explaining it again, and finally getting to an answer — or just saying, ‘Hey Siri, why is my camera broken?’” Quinlan told CX Dive. The choice, he said, is an easy one.
However, the technology still has a long way to go before it’s ready for prime time.
Gartner predicts that 4 in 5 customer service journeys will begin and be resolved by conversational assistants built into mobile devices by 2029. Jason Maynard, chief technology officer of North America and Asia-Pacific at Zendesk, expects agentic AI to account for a “significant portion” of customer service volume in the next five to 10 years.
“You can imagine if someone has the choice between calling customer service, waiting on hold, explaining their issue, getting transferred, explaining it again, and finally getting to an answer — or just saying, ‘Hey Siri, why is my camera broken?’”
Patrick Quinlan
Senior Director Analyst at Gartner
While agentic AI is in its early days, multiple experts agree that the potential opportunities — and the monetary investments — necessary to make the possibilities a reality in the coming years are already in place.
Businesses have put their investments into deploying technology that automates customer service after they start interacting with the company, according to Maynard.
“Now, you have this equal and opposite side that's very well funded that's [about], ‘How do I turn this device that we all have in our pocket into a personal assistant that can do some of these day-to-day tasks?” Maynard said.
Why third-party agentic AI could take off
The marriage of AI and customer service, in the hands of the customers themselves, could bring convenience to the next level. Agentic AI is capable of handling the drudgework of digging up answers from a FAQ or waiting on hold for an available customer service agent.
For instance, a customer with a question about Walmart currently needs to navigate to the site or app and find the page with the right answer, according to Quinlan. An AI assistant could find it almost instantly.
The time savings could be even more drastic for customer service inquiries.
“It's just dramatically more convenient to use the device that most people are going to have in their pocket already to get an immediate answer with a system that can ask follow-up questions and for clarifications and then, eventually, actually take action on your behalf,” Quinlan said. “I think that's where it can get really interesting.”
“There's just a ton of opportunity for interacting with these systems very conveniently and getting that immediate gratification that people are looking for.
Isabelle Zdatny
Head of Thought Leadership at Qualtrics XM Institute
Another potential benefit is using a voice interface versus typing. Some people are more comfortable asking a question than typing a query into a search engine, which could add to the appeal.
“Voice is our first and most natural way to communicate,” Quinlan said. Imagine “if I can just talk to my watch or to my tablet or to my phone or to the thing hanging on my wall or sitting on my desk.”
These benefits become magnified if an AI agent’s abilities aren’t limited to a single company, according to Isabelle Zdatny, head of thought leadership at Qualtrics XM Institute. Smart devices that can cancel subscriptions or undertake other actions for a multitude of brands present an appealing proposition.
“There's just a ton of opportunity for interacting with these systems very conveniently and getting that immediate gratification that people are looking for,” Zdatny said.
Amazon is already positioning itself at the forefront with Buy for Me, an agentic AI-powered feature that can purchase products from third-party sites without requiring customers to leave the Amazon Shopping app.
Agentic AI presents challenges for companies
Third-party agentic AI has a lot of potential, but the technology is still in its infancy and poses businesses fresh challenges that could outweigh its benefits.
The realistic outcome of businesses’ investment in AI this year is the less exciting, but no less important, groundwork, according to Zdatny. Agentic AI needs massive amounts of data behind it, and generative AI is well-suited to collecting and organizing information while establishing the foundation for agentic applications.
“I've been interviewing a ton of companies about the work they're doing, and I almost always end with the question, ‘Where do you see your customer experience in three to five years based on these AI capabilities?’” Zdatny said. “And almost to a person everyone says, ‘Who can say?’”
Instead, many companies are focusing on the roadmap for the next 18 months, which is much easier to see, according to Zdatny. In some cases this includes developing first generation AI agents for customer service — which come with their own challenges.
Once agentic AI comes into its own, its widespread use could present a new set of challenges, according to Quinlan. One important consideration is potentially losing a direct line between the brand and the customer.
“As friction is diminished, support interactions will rise. If a customer doesn't have to wait on the phone for 20 minutes to get something done, it's much more likely they'll do it.”
Jason Maynard
Chief Technology Officer of North America and Asia-Pacific at Zendesk
If a customer has a bad third-party support experience, that could lead to them leaving the brand they told the AI agent to contact, according to Quinlan. The customer never contacted the brand directly, leaving the brand without an opportunity to make that experience right.
“There's a loss of control over the experience, and there's certainly a loss of data and insights about what your customers are doing or struggling with or feeling,” Quinlan said.
Contact centers should expect massively increased volume — as much as three to times more in the next five years, according to Maynard. As consumers gain access to personal agentic AI agents, the convenience of the tools will make it more likely for them to reach out for customer service support.
“This has huge implications for how businesses will think about scale, channels and integrations,” Maynard said. “As friction is diminished, support interactions will rise. If a customer doesn't have to wait on the phone for 20 minutes to get something done, it's much more likely they'll do it.”
Trust may be a barrier to consumer adoption
Business leaders aren’t the only ones who may feel wary about agentic AI. While the technology has the potential to make customers’ lives easier, they will likewise seek to balance its benefits against the potential loss of control.
An AI assistant not only needs personal information to do its job, it needs permission to share that information with businesses, according to Maynard. This will make trust paramount, and some tech companies are in a better position than others to earn that trust.
“On that front, I feel like Apple is really well positioned,” Maynard said. “You have so much of that personal information already on your phone. They've taken a really strong stance around data privacy and protection.”
Apple already warns users if they’re about to unknowingly share information with an app, which is another point in their favor, according to Maynard.
Even if agentic AI becomes mainstream, that doesn’t mean every customer will give up privacy for convenience. A portion will be uncomfortable with the idea of handing off important tasks to technology they don’t fully understand, according to Zdatny.
Many people are already expressing hesitation. Two-thirds of consumers say they would refuse to let an AI agent make a purchase on their behalf even if it would result in a better deal, according to a February Omnisend survey.
“Some of that's going to come down to how comfortable people feel with the AI black box, of just having the model pop out recommendations or actions without seeing what's going on behind the scenes,” Zdatny said.
The time to prepare starts now
Agentic AI is just beginning to take its first steps, but customer experience leaders can still make preparations for when the technology becomes more mainstream.
Implementing proper knowledge management procedures will be an important first step, according to Quinlan. Companies need to treat their knowledge as data and make that data accessible to internal or third-party AI to power agentic features.
Leaders also have an opportunity to put guardrails on data access, which can help them maintain control over the customer experience by limiting what information AI agents can access.
“They'll need to decide what types of questions they want third-party AI to answer, what types they want their own first-party AI to answer, and what questions should not be answered by AI at all,” Quinlan said.
Companies will also want to push for connection with third-party agentic AI systems to further refine their control over what answers are provided, according to Quinlan. They may also seek opportunities to collect data from third-party tools when they interact with the brand.
The tech companies providing agentic AI will likely be willing to assist with this integration, according to Quinlan. Vendors could make control over interactions with their conversational assistants into a differentiating feature.
The true potential of third-party conversational agents is years away, but companies will want to start considering how the technology could affect their customer experience, as the technology holds the potential to change customer service at scale.
“I think it's going to pretty dramatically change how people interact with companies and how they interact with their devices,” Quinlan said.
Kristen Doerer contributed reporting to this story.
