The Pope Got Hung Up On. Your AI Stack Could Be Why.

Pope Leo XIV wrote 42,000 words this year warning the world about AI stripping accountability from human decisions. Then his own bank’s AI-powered customer service line hung up on him.

The two events aren’t directly connected, but together they put a face on something CX leaders have been navigating for years. Enterprises deployed customer service AI tools on top of fragmented infrastructure. That means siloed systems, disconnected data, and no unified view of the customer. AI doesn't fix that fragmentation. It accelerates it. Every bad interaction that fragmentation produces happens faster and at a greater scale. The accountability gap the Pope describes is an AI architectural failure that will define the next phase of customer service AI.

Cognitive Overload From Tool Sprawl Is the Problem

Picture a customer calling about a billing dispute. The agent pulls up the ticketing system to check account history, opens a separate knowledge base article to confirm policy, then switches to the billing platform to process the refund.

Three tabs, no shared data, no unified record. The agent serves as the integration layer between systems that were never designed to talk to each other.

According to recent data, 81 percent of agents juggle more than four tools simultaneously during a single customer interaction. When AI is added to that environment, it inherits the same structural problem: surfacing data from one system while missing critical context sitting in another. The agent still has to piece that information together themselves before responding.

That's the swivel chair: a workaround born from disconnected systems, and one that most enterprise AI deployments have inherited rather than solved. The swivel chair is why the Pope's bank couldn't answer for its own AI. Nobody owned the full picture, so nobody could be held accountable for it.

Distrust Is the Symptom

The consequences show up in how agents actually AI. Despite 100 percent of agents interacting with AI daily, not a single agent considers it critical to their success. An additional 54 percent of agents say AI lacks the context and depth needed for real impact, and 93 percent feel the need to verify AI outputs before acting on them.

That last figure deserves attention. An agent receives an AI-generated summary of a customer's issue, but the summary was built from one system while missing context sitting in two others, so the agent checks it manually before doing anything. The customer waits. The interaction takes longer than it would have without AI involved at all, and the productivity gains the technology was supposed to deliver never materialize. The Pope wanted humans to stay accountable for decisions that AI touches. These agents are already doing that job manually, one verification at a time, because the systems underneath never gave the AI enough to be trusted on its own.

Accountability Requires a Connected Record

The Pope's core concern maps directly onto this dynamic. When AI acts on fragmented data, and something goes wrong (e.g., a fraud alert an AI never escalates because the flag lived in a system it couldn't see, a customer locked out of their account for days because no one owns the full thread, and more). Accountability requires connected data.

The economics reinforce the urgency. Gartner predicts that by 2030, cost per resolution for generative AI in customer service will exceed $3—higher than many offshore human agents—driven by the complexity that compounds when AI navigates fragmented data environments, consuming more tokens, requiring more oversight, and producing outputs that still need human verification before they can be trusted.

Fragmented infrastructure creates a cost problem that gets worse as AI usage scales.

The Architecture Problem Underneath It All

Every failure in this piece traces back to the same root cause: AI deployed on top of fragmented systems instead of a unified one. Fixing that requires giving AI access to a single customer record spanning CRM, ticketing, payment systems, and interaction history before it acts on anything.

With that foundation in place:

• The swivel chair disappears. AI can pull from every system at once, so the agent stops manually stitching together account history, tickets, and payment data mid-call.
• Verification stops being the default. AI summaries carry full context instead of a partial view, so agents can act on them instead of re-checking them 93 percent of the time.
• Accountability becomes possible. When something does go wrong, there's a single connected record to trace the failure back through, instead of three disconnected systems pointing at each other.

The customer stops re-explaining their situation from scratch. Gartner's own guidance points the same direction: lead with AI for customer engagement rather than full automation, and invest in the data infrastructure that makes engagement possible. The orchestration layer has to come first, because AI can only perform to the quality of the data it can reach.

Pope Leo XIV wrote 42,000 words about AI eroding human accountability. Around the same time, he couldn't get a straight answer from his bank’s customer service line. The fix isn't more AI, but a unified record underneath it, so the next answer AI gives is one someone can actually stand behind.

As UJET’s CEO, Vasili Triant oversees all company operations and strategic initiatives. Triant brings more than 20 years of experience in Telecom, Unified Communications (UC), and Contact Center industries, having previously served as VP/GM of Contact Center at Cisco, where he achieved the fastest growth in over a decade through a focus on global alliances and enterprise cloud-readiness.