Our Take
KPN's move signals that European telecoms see agent-based AI as a near-term cost lever, not a long-term R&D bet, but the McKinsey framing obscures whether this actually reduced call volume or just redistributed it.
Why it matters
Contact centers remain one of the largest cost centers in telecom. If KPN's deployment works, it becomes a template other carriers will copy within 18 months, reshaping staffing and vendor strategy across the sector.
Do this week
CTO/Enterprise AI lead: Map your contact center call flows and hand-off rules before Q2 2025 so your procurement team can negotiate agentic AI pilots with realistic SLAs.
KPN deploys agentic AI in customer support operations
KPN, the largest telecom operator in the Netherlands, is building an agentic AI system for its contact center (per McKinsey Insights). The deployment focuses on improving service quality and operational efficiency. The company is also building internal AI capabilities to sustain competitive advantage long-term.
Agentic AI differs from traditional chatbots in that it can make decisions, take actions, and handle multi-step tasks without handing every edge case back to a human operator. In a contact center, this means agents can resolve billing disputes, provision services, or escalate intelligently rather than simply classify and queue.
Cost pressure and capability building collide
Contact center labor is one of telecom's stickiest cost categories. A typical mid-size carrier spends 8–12% of revenue on customer care. Agentic AI directly attacks that line item by handling routine and semi-routine interactions without hiring.
KPN's framing, however, reveals two distinct motivations. The immediate story is efficiency: fewer agents per call, faster resolution, lower cost-per-contact. The longer-term story is capability building. Companies that can operate agentic systems in-house avoid vendor lock-in and maintain optionality as the technology matures. This is why KPN is investing in internal AI infrastructure, not just bolting on a third-party solution.
For competitors watching KPN, the stakes are both tactical and strategic. If the deployment succeeds and becomes visible in KPN's customer satisfaction metrics and earnings reports, other carriers will feel obligated to follow. That creates a buyer's market for agentic AI vendors in telecom for the next 12–24 months.
What to prepare now
If you run customer care operations at a carrier or large enterprise, you should audit your call routing and escalation rules now. Most contact centers have tangled decision trees built over years. Agentic AI will expose that mess immediately. Agents cannot route intelligently if humans cannot articulate the routing logic.
Second, benchmark your current handle time, first-contact resolution, and customer satisfaction scores. You will need baselines before pilot deployment, and vendors will push aggressive targets. Independent measurement matters.
Third, clarify your vendor strategy. Are you building an in-house agentic engine like KPN or buying a packaged solution? The choice determines hiring, architecture, and contract terms. Buying looks faster but locks you into a vendor's roadmap. Building is slower and capital-intensive but gives you control. KPN is betting on the latter; most firms will not have that luxury.