For the past two years, the artificial intelligence debate has been dominated by capability.
AI has moved quickly from innovation teams into boardroom conversations, every week has brought another model, benchmark or prediction, with attention focused on what AI can do.
Across industries, executives are being asked what their AI strategy is, where they are investing and what results they expect to see. Customer service is often one of the first areas under scrutiny, because the case for improvement is easy to understand: faster responses, better customer experiences and lower operating costs.
Indeed, the implications of better customer service for business’ bottom lines is clear, research from PwC suggests that 86% of buyers are willing to pay more for a better customer experience.
Statistics like this make the pressure to move understandable. Customers expect to be able to contact organisations across voice, chat and messaging without having to repeat themselves or wait unnecessarily. At the same time, businesses are trying to improve efficiency and manage costs. AI appears to offer an answer to both. The technology has also matured quickly, with conversational AI, automated knowledge management, real-time agent assistance and emerging agentic AI applications no longer being theoretical. Most large organisations now have access to tools that would have felt experimental only a few years ago.
But having access to the technology and the capability is not the same as understanding how it should be applied within your business. Board-level enthusiasm has triggered a rush to invest, yet many businesses are still starting with the tool rather than the problem. They launch pilots, test new platforms and deploy isolated use cases, only to find that scaling them across complex customer operations is harder than expected. Gartner research reveals that up to 72% of AI initiatives struggle to fully meet ROI expectations, with only 1 in 5 achieving clear financial returns.
This is where the conversation around AI is starting to change. Success is no longer measured by whether a business has implemented AI, but by whether it improves customer experience, raises productivity and delivers a measurable return.
The organisations seeing the strongest results are not necessarily buying the most advanced technology, instead they’re applying it to specific operational problems.
That requires a clear understanding of how customer service actually works. These are complex environments, often involving multiple systems, regulatory obligations, fragmented data and customer journeys that stretch across several channels. Introducing AI into that environment is rarely just a technology project. It usually requires changes to workflow design, systems integration, knowledge management and governance.
In consumer finance, NewDay integrated generative AI into its contact centre to improve both efficiency and customer experience. Its initial proof of concept focused on live call transcription and knowledge base access, but high costs and infrastructure limitations forced a pivot to a chatbot solution. Within ten weeks, the system reached 80% accuracy, later rising to 90% in a pilot with agents, alongside reported reductions in handle and hold times.
A similar lesson can be seen in the insurance industry. SegurCaixa Adeslas applied generative AI to medical authorisations, a sensitive and high-volume process. The system interprets customers’ natural language requests, maps them to the right medical speciality and passes the required information into transactional systems. It now auto-authorises around 15% of authorisation calls, while transferring unresolved requests to agents with a transcript attached.
The most effective deployments start with a clear view of where value can be created – and agreed outcome. That might be reducing repetitive administration, improving self-service, shortening response times or helping agents access the right information more quickly. The technology matters, but so does the operating model around it.
These foundations become even more important as organisations begin to explore agentic AI. Unlike tools that respond to a prompt or assist with one task, agentic AI can act across systems and complete multi-step processes. The opportunity is significant, but so is the need for control. Speed and efficiency cannot come at the expense of trust, compliance or accountability.
That is why many organisations will need credible partners who understand both the technology and the environment it is being deployed into. Successful organisations operating within the next phase of AI adoption will know which AI technologies fit which problem and will not be businesses that blindly implement automation tools. They will also know how to integrate those tools properly and how to measure whether they are working.
The boardroom has decided that AI matters, but now the task is turning that investment into better experiences for customers, better tools for employees and stronger business performance.
