Bank customers are adopting AI-powered support channels whilst simultaneously demanding human agents remain available for complex or sensitive interactions. This paradox reflects a maturation in customer expectations: AI is acceptable for triage, transaction processing, and routine inquiries, but customers retain deep scepticism about algorithmic decision-making in high-stakes scenarios—particularly in financial services where trust and accountability carry material consequences. The tension exposes a critical gap in how many organisations have architected their support infrastructure. Rather than treating human and AI channels as competing modalities, leading teams are recognising them as complementary layers within a unified routing strategy. The question for CX leaders becomes whether your current platform architecture—whether Zendesk, Freshdesk, or Salesforce—is configured to seamlessly escalate from AI to human agents without forcing customers to repeat context, or whether you're still operating siloed systems that punish customers for requesting human intervention.
This finding carries immediate implications for investment prioritisation. Teams currently deploying generative AI chatbots or Agentforce-style automation must recalibrate their success metrics away from pure deflection rates and towards handoff quality and first-contact resolution on human channels. The data suggests that aggressive AI-first strategies risk eroding customer satisfaction if they create friction around human escalation. Conversely, organisations that position AI as a friction-reducer—handling intake, gathering information, and pre-qualifying issues—rather than a replacement layer are likely to see stronger retention and NPS outcomes. For support leaders evaluating vendor capabilities, this reinforces that the differentiator is no longer whether a platform offers AI, but whether it enables intelligent triage that respects customer agency and preserves human capacity for genuinely complex work.
Human agents must therefore be repositioned as premium resources handling high-value interactions rather than first-line responders to routine queries. This shift demands investment in agent tooling, knowledge management, and training to ensure that when customers reach a human, that interaction delivers disproportionate value. The operational implication is clear: if your team is still measuring success by cost-per-contact or average handle time, you're optimising for the wrong outcome. The market is signalling that customers will tolerate—even prefer—AI for simple tasks, provided the human fallback is frictionless and genuinely capable of resolving complex issues.
Bank customers, while embracing AI, call for human customer support Customer Experience Dive
Bank customers, while embracing AI, call for human customer support Customer Experience Dive