KPN's development of an agentic AI engine for customer care represents a deliberate shift from reactive automation toward autonomous decision-making systems that can handle complex customer interactions without human intervention at every step. Rather than deploying chatbots constrained to predefined workflows, KPN is building agents capable of reasoning through problems, accessing multiple systems, and determining appropriate next actions—a meaningful distinction that separates this approach from traditional IVR or rule-based automation. The telecommunications operator's investment signals that mature enterprises are moving beyond the "AI as efficiency layer" narrative toward genuine autonomy, where agents operate with delegated authority to resolve issues end-to-end. This aligns with broader industry momentum: 70% of companies deploying customer service AI agents see ROI in 60 days, suggesting the business case is no longer speculative.
For CX teams, this development creates an immediate strategic question: how does agentic AI reshape the role of your existing support infrastructure? Teams currently optimising for first-contact resolution and agent efficiency may find those metrics become secondary to agent autonomy and decision quality. The implication is substantial—your platform selection, team structure, and performance frameworks may require recalibration if you're not already architecting for systems that operate independently rather than augmenting human agents. Organisations like Verint and Conduent are already shipping agentic products, meaning the competitive pressure isn't theoretical. The critical question becomes whether your current platform—whether Zendesk, Salesforce, or another vendor—can evolve quickly enough to support genuinely agentic workflows, or whether you'll need to integrate specialist tools that operate alongside your existing stack.
The KPN case also exposes a capability gap between enterprise-scale deployments and mid-market operations. Building agentic systems requires sophisticated data architecture, robust governance frameworks, and the ability to manage agent failures gracefully—capabilities that demand investment beyond software licensing. This suggests a widening divide: larger organisations with dedicated AI teams will extract disproportionate value from agentic systems, whilst smaller teams may find themselves managing hybrid models where human agents remain essential for edge cases and complex reasoning. The real competitive advantage won't accrue to those who simply deploy agents, but to those who can operationalise them effectively—which means your team's ability to monitor, audit, and continuously improve agent decision-making becomes as critical as the technology itself.
How KPN is building an agentic AI engine for customer care McKinsey & Company