Dutch telecommunications company KPN has partnered with McKinsey & Company and its AI division, QuantumBlack, to transform its customer service operations using agentic artificial intelligence, as businesses increasingly explore AI-powered systems capable of handling end-to-end customer interactions while working alongside human employees.
The initiative marks KPN's shift from a traditional contact centre model to an AI-enabled operating framework where specialised digital agents perform routine customer service tasks under human supervision. According to the companies, the programme has achieved an employee adoption success rate of 86 percent, while enabling faster issue resolution and improving operational efficiency.
KPN, the Netherlands' largest telecommunications provider, receives around five million customer calls annually. To identify opportunities for automation, project teams analysed large volumes of customer call transcripts and chat logs using large language model-based frameworks. The analysis grouped interactions into recurring customer needs, allowing the company to identify processes suitable for AI while retaining human involvement in more sensitive scenarios such as home relocations and complex service requests.
The first set of AI agents has been deployed across customer verification, order status enquiries, technician appointment scheduling and internet troubleshooting. Rather than functioning as standalone chatbots, the digital agents are designed to complete entire customer journeys by connecting directly with KPN's core telecommunications systems.
To support natural voice conversations, the companies built a centralised conversational platform capable of maintaining response times below two seconds per interaction. The system also allows customers to interrupt conversations while preserving context, a capability intended to make AI-powered voice interactions more natural. The platform was designed as a reusable technology stack that can support future AI applications across the organisation instead of isolated use cases.
KPN said quality control remains a key component of the deployment. The company has implemented AI guardrails, automated monitoring tools and human evaluation processes to ensure conversations remain accurate and aligned with company policies. Teams reportedly review up to 100 real customer conversations each day, refining prompts and updating AI models before releasing improvements into production.
Beyond technology, the telecom operator has also invested in workforce readiness by updating operating procedures and training programmes. As AI agents assume repetitive activities, customer service employees are being prepared to focus on higher-value responsibilities that require empathy, judgement and complex problem-solving. Frontline staff also participated extensively in testing before deployment, contributing to the programme's adoption rate.
Early operational results indicate improvements in call handling efficiency while maintaining customer experience. According to McKinsey, KPN's customer satisfaction score currently stands at 83, comparable to interactions handled entirely by human agents. The company aims for agentic AI to manage between 10 and 20 percent of customer service calls by 2027, while ensuring customers can continue to access human support whenever required.
The partnership reflects a broader shift in enterprise AI adoption, with organisations moving beyond chatbot-based automation towards agentic AI systems capable of executing complete business workflows. Consulting firms increasingly view such deployments as a step towards AI-native operating models, where digital agents augment employees rather than replace them. For KPN, the collaboration also establishes an internal foundation for expanding agentic AI into additional business functions over the coming years.
Related Articles
