When Vercel first adopted AI for customer support in early 2024, the technology company was able to use it to resolve about 30% of its cases. That figure has since climbed to over 70% and is projected to exceed 80% by early next year, according to Matthew Sweeney, senior director of customer success at Vercel.
“The primary use of AI — for me — is the resolution of cases,” Sweeney said.
It’s a rare instance of AI living up to its promises, as approximately 2 in 5 companies abandon most of their AI initiatives, with nearly half of initiatives seeing no return on investment, according to S&P Global Market Intelligence.
However, the tools themselves are not the problem.
“The challenge is that tools don’t change behavior,” said Patricia Camden, EY managing director and EY Americas loyalty leader.
Change management is the key to encouraging the adoption of AI in customer experience, but many leaders still view the issue as a technology implementation problem, according to experts.
“It’s one of the biggest misconceptions among our clients,” Camden said.
Most businesses assume that once AI tools are in place, employees will start using them automatically and that “the transformation will happen naturally,” Camden said.
“But the adoption of AI is not about technology. It’s about people,” Camden said. "Organizations are adopting AI technically, but they're not enabling it culturally. They roll out these systems, but they don't create the time, space or the emotional permission for their people to learn, practice and build trust in this new way of working.”
AI isn’t magic
The core issue is simple: Most organizations struggle with AI adoption because they want to “go too fast, too soon,” Sweeney said.
“They expect magic out of the box,” Sweeney said.
AI chatbots and agent assistants can help resolve about 20% to 30% of all customer support cases with minimal investment, and basic prompt engineering and indexing documentation can increase that up to about 65%, according to Sweeney.
“But it’s not really going to get there on its own,” Sweeney said.
Further improvements require significantly more investment and face three main challenges: organizational silos, operational disconnect, as well as fear and mistrust.
AI initiatives often launch across different departments, including CX and IT, without effective AI governance, which can damage customer experience, loyalty and brand reputation. That can result in a lack of alignment, as teams frequently disagree about what AI should accomplish, leading to unclear objectives and ineffective outcome measures.
Unclear ownership also leads to a lack of accountability, as CX owns customer relationships but may not own the AI implementation.
That’s not to mention that many AI tools exist outside normal workflows. They’re not integrated within them, creating “complex handoffs” and “operational clunkiness” as agents struggle to jump between their core systems and AI tools, according to KJ Kusch, global field CTO at WalkMe, a digital adoption platform.
And perhaps most importantly, businesses must overcome staff skepticism of AI. Many employees, especially agents, worry that AI will replace their jobs. That often stems from a lack of understanding about AI’s actual purposes, resulting in short-term use and abandonment rather than sustained adoption.
Getting buy-in
Bringing staff onboard is easier said than done.
Experts say in addition to providing all stakeholders a seat at the table from the start, focusing on five key areas can help ensure AI adoption:
- Value alignment: CX leaders must ensure that everyone is aligned on the technology’s business value, such as cost savings, first-contact resolution and escalation prevention. “Value means different things to different people,” Kusch said. “If we don’t agree on the value, we're never going to get to adoption."
- Phased rollouts: Starting with small pilots allows for gradual trust building through demonstrated wins and creates positive feedback loops before wider deployment.
- Communication and feedback: Businesses must proactively communicate their AI plans and engage in ongoing discussions with all stakeholders, including customer support workers, to successfully address cultural barriers to adoption.
- Measure screen-level activity: Tracking usage patterns rather than high-level metrics can help businesses identify when and where users abandon AI tools, as well as ensure adherence to compliance and processes. “If you don’t know how you’re doing, you won’t necessarily see if somebody got the wrong answer five times before moving on,” Kusch said.
- Build an AI assistant, not a replacement: Positioning AI as an assistant can help employees buy into and adopt the technology. Seamlessly integrating it into existing workflows also ensures that agents get the help they need at the “right time, the right moment,” Kusch said.
Vercel has taken a continuous learning approach to AI implementation and adoption, viewing every escalated case as an opportunity to improve.
The company created an internal dashboard that utilizes customer data to automatically review where AI went wrong and feed corrections into its models, thereby speeding up the learning process without requiring employees to review thousands of tickets manually.
“You know exactly where in the interaction it went wrong,” Sweeney said. “That’s incredible insight.”
It’s made a massive difference in adoption, as both customers and employees experience the real-world benefits of automation.
Ultimately, however, CX is about people, and AI is no different.
"We're implementing these technology platforms to make our lives better, but it takes a healthy dose of humanity to actually make it successful,” Camden said. "I was talking to a client last week, and she said, 'My goal for AI is that it gives my team a little more breathing room so we can be more human.'”
