As AI agents become embedded in CX, deploying them on fragmented data and disconnected systems turns them into a liability at enterprise scale.
The urgency to ship AI-powered experiences is creating a dangerous gap between what organizations think they are delivering and what the business can stand behind.
For CX leaders, that means stepping back from the deployment-first mindset that has defined the early AI era, confronting the foundational weaknesses that has persisted beneath their customer-facing technology.
Balaji Balasubramanian, President and CPO at SAP CX, told CX Today that as brands race to deploy AI agents, CX leaders must question whether the foundations beneath those experiences are strong enough to hold.
“The danger with ‘vibe coding’ CX applications is that it can make software appear useful before it is safe, reliable, or capable of delivering on customer promises,” he explained.
“Applying AI on top of weak or fragmented foundations doesn’t just introduce risk; it amplifies those weaknesses at scale.”
AI Cannot Fix What It Can’t See
Today, AI is accelerating the weaknesses that have existed in CX technology for years, but with many organizations focusing on quick AI agent deployment, the underlying customer data remains fragmented.
When business rules are inconsistent and core systems remain disconnected, AI simply operates on incomplete or inaccurate information.
“Automation alone doesn’t deliver a step-change in value, it’s the intelligence behind those actions, grounded in enterprise context, that drives meaningful outcomes,” Balasubramanian said.
“Without that layer of context, an AI agent may act without the relevance and security that enterprises require.”
If an AI agent lacks visibility into customer service information, it may create inconsistent support experiences that are incorrect or even breach privacy expectations.
He emphasized:
“Without that foundation, brands risk offering promotions they can’t fulfill, recommending products that aren’t available, violating consent preferences, escalating the wrong service case, or creating inconsistent pricing and billing experiences.”
These technical issues directly affect customer trust and operational efficiency, because when AI executes decisions at a scale and speed, that can rapidly amplify existing weaknesses within an organization.
This reflects the growing SaaSpocalypse narrative online that AI interfaces will make conventional SaaS applications largely unnecessary, enabling users to complete tasks through conversational agents.
AI cannot compensate for disconnected enterprise systems because its effectiveness depends on updated and governed business context.
“AI amplifies both strengths and weaknesses, which is why a strong enterprise foundation is essential,” he continued.
The next priority for CX organizations will therefore require AI agents to be grounded in trusted data and connected systems that enable the business to consistently deliver on promises.
The New Rules of Agentic CX
For CX leaders, the next phase of AI adoption requires it operate effectively and responsibly, ensuring those agents have the right context and permissions to take meaningful action.
Any AI agent can generate a conversation, but its value depends on whether it can access trusted customer data and connect with effective operational systems.
“The winners will be the brands that combine AI-generated experiences with trusted business data, governed processes, and enterprise systems that can deliver on every customer promise,” Balasubramanian noted.
“That’s what transforms AI from simply generating conversations into driving reliable business outcomes.”
For CX teams, personalization in agentic environments requires the right action for the right customer with the appropriate permissions and context, ensuring the organization can follow through.
More recently, governance is becoming a critical requirement as AI agents take on more autonomous decision-making, requiring organizations to provide clear controls around data access, actions, and when human intervention is necessary.
“As AI agents become more autonomous, they need to understand what they’re allowed to do, what data they can use, when consent is required, when a human should be involved, and how their decisions and actions are monitored or audited.”
As a result, CX leaders that build these capabilities will be better positioned to use AI to create consistent, reliable experiences that strengthen customer trust.
