Acting on the right customer signal before an experience deteriorates is becoming the next CX challenge, even after an AI is deployed.
Whilst the technology is raising the bar for CX, generic models are unable to reliably resolve customer issues without an understanding of the business behind the interaction.
By becoming proactive with an AI deployment, CX leaders should be aware of where AI sources evidence, how decisions are governed, or whether their outcomes can be measured post-interaction.
Speaking with CX Today, Mark Hammond, SVP for Core AI at Qualtrics, explained why generic AI models are unable to effectively support customers alone.
“The general-purpose large language models know everything about everything except your business," he explained.
"They’re not grounded in your business’s operations. They’re not grounded in the details of the service that you specifically offer.”
The Cost of Playing Catch-Up in an Interaction
Customer expectations are outpacing many organizations' ability to adapt their service operations.
As customers become accustomed to the speed and convenience offered by consumer technology, this creates global-wide standards for every business interaction.
“The modern set of AI technologies is really driving a rapid increase in what customers expect out of their interactions,” Hammond noted.
However, many CX programs still rely on established, traditional methods of collecting customer feedback, often leaving a larger-than-hoped gap between issue and resolution.
By the time feedback reaches the relevant team, a customer may already have experienced repeated frustration or decided to take their business elsewhere.
Organizations therefore face growing pressure to move beyond retrospective reporting and use customer signals to inform decisions while interactions are still taking place.
However, introducing AI does not automatically close this gap, as many general-purpose LLMs often lack the needed operational knowledge.
“The general-purpose AI models can tell you a lot about the industry, the space," he acknowledged.
"They can give you a plausible answer to a problem, but not a specific answer to that specific customer’s problem in that specific moment.”
For CX leaders adopting AI, access to relevant business information, policies, customer histories, and operational processes is essential to provide recommendations that reflect the circumstances of an individual interaction.
Without this context, organizations risk automating responses that fail to close the experience gap.
Neurosymbolic AI, Combining the New with the Old
To make AI a reliable part of CX, organizations will need neurosymbolic AI, an approach that combines generative AI with structured business information and established reasoning techniques.