Claudeforce—Salesforce's integration with Anthropic's Claude—exemplifies a systemic misdiagnosis in how enterprises approach customer experience improvement. Rather than addressing the root cause of friction, organisations are layering AI onto already-fragmented technology stacks that have accumulated over decades of point-solution buying. UJET CEO Vasili Triant argues that the industry has fundamentally misidentified the problem: it is not the contact center agent that needs replacing, but the underlying infrastructure that forces agents to navigate multiple disconnected systems during live conversations. When companies bolt AI onto legacy CRM and ticketing platforms without consolidating their data architecture, they create another interface, another workflow dependency, and another source of operational drag. For teams already managing complex Zendesk, Salesforce, and Freshdesk implementations, this raises an uncomfortable question: does adding an AI copilot to your existing stack solve customer experience problems, or does it simply obscure them beneath another layer of automation?
The core issue traces back to how SaaS procurement has evolved. Individual tools appear manageable in isolation, but collectively they create what Triant describes as a "massively complex stack"—comparable to the fragmentation that occurred when streaming services replaced cable bundles. Ticketing platforms and CRM systems, designed in an era before data warehouses and customer data platforms existed, have become the gravitational centre of service operations by default rather than design. This architectural debt now constrains how effectively AI can operate. When customer context is scattered across multiple systems rather than unified in a single data lake, AI agents and agent-assist tools cannot surface relevant information efficiently; instead, they amplify the problem by adding another system to the maze. The implication for CX leaders is stark: the strongest ROI may come not from implementing more technology, but from removing it. Triant explicitly questions whether ticketing platforms and certain CRM layers will survive as enterprises mature their data strategies and build intelligence across enterprise systems rather than confining it to contact center tools.
The human agent's role is being reframed in this context, not eliminated. As customer fatigue with poorly designed automation grows—evidenced by users hanging up when encountering virtual agents—the competitive advantage shifts toward brands that preserve emotional context and first-contact resolution. This suggests that teams should evaluate their AI investments not by how much automation they enable, but by whether they reduce the cognitive load on agents and simplify the customer journey. The question facing support leaders is whether their current stack can support this simpler model, or whether it has become so entrenched that adding intelligence merely perpetuates complexity under a different name.
Are AI agents fixing customer experience, or are they becoming the latest layer in a stack that already asks too much of agents, customers, and IT teams? That is the question facing many enterprises as customer relationship management (CRM) platforms, contact center systems, ticketing tools, and AI