Liam Dunne is the Co-founder and CEO of Klearcom, a global leader in customer call path testing & real-time contact center optimization.
One of the foremost concerns about customer service AI is whether it can resolve a customer’s problem without trapping them inside automation. Nobody wants to be stuck in a whirlpool, do they? New Irish government rules now allow “consumers buying financial products or services online or over the phone” to “request human intervention rather than rely solely on automated systems.” Although their legal scope is specific, the wider message should matter to every contact center leader.
The timing matters. Eurostat reported that “20% of EU enterprises with 10 or more employees used AI technologies” in 2025, up by 13.5% from the prior year. Adoption is moving quickly, so the customer experience around AI is very much a growing concern, as I’m sure you’ve read recently.
It would be easy to see a right to human intervention as a rejection of AI. I see it as evidence that AI is maturing and becoming more customer-friendly. Strong systems won’t be judged by how effectively they keep people away from agents but by whether they understand intent, complete the task and recognize when a person should take over.
Containment Is The Wrong North Star
For years, many contact centers have measured automation through containment: how many interactions ended without reaching a human. That metric has value, but it becomes dangerous when treated as the goal. A customer who gives up, calls back or changes channels may still appear to have been “contained.” Any apparent saving disappears if the unresolved problem returns as a complaint or repeat contact.
Leaders should instead examine first-contact resolution, repeat attempts, abandonment after failed recognition, time to recovery and transfer quality. The cost of an AI interaction must include what happens when automation misunderstands, loops or blocks access to help.
Human Handoff Must Be Designed, Not Added
In my experience testing and optimizing customer call paths, a system can recognize a customer’s intent correctly and still fail. Authentication can break. A back-end connection can return the wrong status. Routing rules can send the call to the wrong team. The agent may receive none of the context already provided. Customers don’t experience the AI model in isolation. They experience the entire chain.
That is why human handoff should be a core capability, not an emergency exit. When escalation happens, the system should pass the reason for contact, verified identity, steps already completed, relevant history and point of failure. Asking a customer to start again isn’t a handoff. It’s a restart.
Escalation shouldn’t depend only on the customer repeating “agent.” Confidence scores, repeated intent failures, urgent language, vulnerability indicators and high consequence scenarios can signal that automation has reached its limit. The objective is to identify when human judgment has greater value than another automated turn.
The Human Agent Becomes More Valuable
As AI agents absorb more high volume, repetitive work, I expect the human role to become more specialized. Status checks, appointment confirmations, simple account queries and routine routing are suited to automation when the journey is reliable. Human agents can then focus on exceptions, complaints, fraud concerns, bereavement, financial distress and other situations where judgment, empathy and accountability matter.
This raises the standard of the role. The contact center agent of the future will need to understand how AI reaches decisions, how to challenge or correct it, how to diagnose journey failures and how to take control without losing the customer’s trust. Working effectively with AI becomes a skill in its own right.
Contact center experience may also become more valuable across technology organizations. People who have handled real customer conversations understand where scripts fail, where policy collides with reality and where a minor system issue becomes a serious human problem. Those insights are essential when enterprises design, test and govern AI agents.
Regulation Is A Floor, Not A Service Strategy
A 2025 Eurobarometer survey captured the balance businesses must strike: More than 70% of Europeans believed robots and AI improve productivity, while 84% said “AI requires careful management to protect privacy and ensure transparency in the workplace.” People can recognize AI’s value and still expect safeguards, choice and human accountability.
Ireland’s 2026 Presidency of the Council of the European Union runs from July 1 through December 31, with a program centered on “competitiveness, values and security.” Europe’s wider AI framework is intended to “foster trustworthy AI.” The lesson for business leaders is that consumer protection and innovation aren’t opposites. Systems that preserve choice and accountability can earn the trust required for wider adoption.
But compliance is only the floor. A company can technically offer a human option and still make it hard to find, slow to access and essentially useless once reached. Leaders should test whether customers can ask for a person in natural language; whether the request works across accents, languages, channels and background noise; whether context survives the transfer; and whether the right specialist receives the interaction. They should also test what happens when a downstream system is unavailable, not only when everything works perfectly.
The Best Of Both Worlds
The future of customer service isn’t a contest between AI and people. AI is well suited for speed, scale and consistency. Humans remain essential when a situation involves ambiguity, emotion, judgment or accountability. The strongest operating model uses each where it adds the most value.
The right to request a human shouldn’t weaken the business case for automation but should sharpen it instead. The next stage will be defined by systems that solve more problems and know their own limits. That’s when AI stops being a barrier between the customer and the company and becomes a better route to the right outcome.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?