The central tension animating current CX strategy is whether AI tools are genuinely improving customer outcomes or whether support teams are primarily generating training data that refines AI models at the expense of service quality. Organisations like Vodafone and emerging vendors such as Kim.cc are betting heavily on AI-native architectures, consolidating operations onto platforms designed around machine learning from the ground up rather than bolting AI onto legacy systems. Yet the parallel trend of companies bringing humans back after betting big on AI suggests the premise itself may be flawed—that every interaction logged, tagged, and resolved by support teams becomes proprietary training data that makes the AI layer more valuable, whilst the actual customer experience stagnates or deteriorates. For teams already running Agentforce or similar agentic systems, this raises an uncomfortable question: are you optimising for customer resolution or for model improvement?
The implications cut across vendor strategy and team resourcing. Mid-market and enterprise CX leaders face a choice between committing to AI-first platforms that promise efficiency gains through continuous learning loops, or maintaining hybrid models where humans retain decision authority and AI serves as augmentation rather than replacement. The legal and operational risks compound this calculus—AI adoption carries significant legal exposure, and teams must now account for whether their support interactions are being used to train third-party models or locked within proprietary systems. Smaller vendors and in-house teams should be particularly attentive: the consolidation trend toward AI-powered CRM platforms suggests that scale and data volume will increasingly determine competitive advantage, making the question not whether to adopt AI, but whether to do so as a customer or as a data supplier.
Is AI Enhancing Customer Support, or is Customer Support Enhancing AI? Hospitality Net