UK organisations are racing to deploy AI in customer service. More than half are using it to help frontline teams become more efficient, and almost 40% are using virtual assistants to handle inquiries instead of people.1 Investment is flowing, pilots are launching, and the headlines keep celebrating ever-more capable AI agents.
But the future of customer service can’t be won by technology alone. A poorly-designed customer journey will still be poorly-designed, even if you add a layer of AI into it.Without the right operating model, skills and governance built around it, even the most advanced tools will struggle to deliver better customer experiences or sustainable outcomes.
The companies that get this right – and make change stick – will be those that think about this new organisational design, with both human and digital agents working together responsibly and effectively, from the start.
The bigger challenge behind customer service AI
The real challenge in deploying AI for customer service is everything that surrounds it. Who designs customer conversations when context, ambiguity and tone matter just as much as speed? Who checks whether the right answer is being given on a Tuesday afternoon, not just in a test environment? How do organisations manage digital agents day-to-day, maintaining performance, compliance and service levels? These questions are rarely addressed with the same urgency as the technology selection process.
Workforce planning also needs a rethink. When human and digital agents work side by side, traditional resourcing models break down. Predicting demand, deciding when AI should hand over to a person, and maintaining service quality all need to change. Organisations that skip this will end up with powerful AI sitting inside a service model that was built for a different era. The technology may be ready, but they are not.
It’s time for leaders to lead
When AI deployments fall short of expectations, the instinct is often to blame the technology. The platform wasn’t right, the model needs more tuning, or the data wasn’t good enough. Sometimes that’s true, but rarely does the diagnosis point to leadership decisions made much earlier in the process. Yet it’s those decisions – about how quickly to move, whether to invest in organisational redesign alongside technology, who owns accountability for AI performance, and if, or when, to hand over to a human agent – that typically determine the outcome.
Senior leaders set the conditions for AI to succeed or fail long before a single customer interaction. If they treat it as a tool roll-out, that’s what they will get. The organisations achieving meaningful results are those asking harder questions from the outset: how will this change the way our teams are structured? What new capabilities do we need to build, and how quickly? And, most importantly, who is accountable when a digital agent gets it wrong?
Building the team behind the AI
To close this gap, organisations need to rethink how customer service is designed and managed in an AI-driven world. The industry is moving beyond traditional contact centres towards integrated orchestration centres which combine AI and customer expertise. Businesses need to ask: how do we build an organisation that can coordinate and continuously improve these interactions at scale? What structure do we need to make it work effectively in the real world?
We believe the answer is a dedicated customer orchestration function, a Digital Centre of Excellence. It will breakdown the traditional boundaries between business teams and functions, drawing them together to shape how customer interactions flow across digital and human channels. That means looking at how conversations are designed, when AI should hand over to a person, how service quality is monitored, and how systems adapt as customer behaviour changes over time.
This is not a traditional technology team, or a side project for an overstretched service leader. It sits at the intersection of customer experience, operations and technology, bringing together the people responsible for designing, managing, and improving customer interactions from the moment of first contact to final resolution. And it requires a new mix of skills and roles that most organisations don’t yet have.
The new frontline of customer service
These are the roles we see a gap for, and which we’re working on creating.
First up is the conversation designer. This role shapes the conversations customers have with digital agents. That means writing flows that account for ambiguity, hold context across a multi-turn interaction and reflect brand tone, which feels human and unscripted, consistently. The focus is on reducing customer friction and ensuring that automated interactions feel intuitive rather than transactional. Crucially, this role needs to work closely with ethics, risk and compliance functions, because a conversation that looks good on paper needs to also be safe to use with real customers.
The second is the digital agent builder – the person who translates conversation design into working reality. They implement prompts, set guardrails and connect the agent to backend systems, as well as making sure it can only act within clear limits. The Builder also takes responsibility for training and refining language models for specific service areas. For example, a digital agent handling council tax needs different rules and escalation points, from one handling a car finance renewal.
The third role is the digital agent analyst, focused on real-time management and continuous optimisation. They keep an eye on how digital agents are performing once live, tracking and overseeing containment rates, escalation patterns, service levels, and data handling. This role keeps digital operations running safely and effectively.
Their job is to spot where the agent is helping and where it is creating risk or extra work. If customers keep asking the same question twice, or if a response is technically correct but unhelpful, the analyst feeds that evidence back to the designers and builders.
Your best AI talent may already work for you
It would be a mistake to assume these roles need to be external recruits. We’ve found that some of the best candidates are already within our contact centres. Team leaders and experienced advisors can be successfully upskilled, as they are the people that know what good customer conversations look like and what good outcomes feel like. They know where customers get confused and where process break down. That knowledge can be hard for an external hire to replicate quickly.
This matters because AI deployment in customer service is a service quality challenge. The people who have spent years handling customer interactions are often best placed to design and govern digital ones. Supporting existing employees, through the right training, as they move into these new roles can help organisations build capability more quickly and sustainably. And allow service teams to shape and own this transformation.
Most businesses are only scratching the surface
Many organisations – particularly those in regulated or high trust sectors – are currently using AI only in limited, internal ways: retrieving documents, surfacing information, or supporting back-office processes. This is a reasonable starting point, but a fraction of the full potential. The real opportunity lies in customer-facing AI that genuinely improves experience, and getting there safely requires both the right structure and clear commitment from leaders.
A well-designed Digital Centre of Excellence gives organisations a practical way to build that structure. It creates ownership for how digital agents are designed, managed, checked and improved responsibly once they are live. But this Centre only works if leaders give it authority and investment. Without visible senior sponsorship and clear accountability, even well-designed functions will struggle.
The structure and software
If AI is to play a meaningful role in transforming customer service, the focus must shift: from the agent to the organisation around it, and from technology investment to leadership accountability. Businesses must treat conversation design, digital operations, governance and workforce development as core strategic priorities, not afterthoughts. The organisations that get this right won’t be the ones with the most advanced tools, but those that build the roles, skills, support functions and accountability needed to run AI well in live services.
That is how AI moves from experiment to everyday performance. And it is how organisations build change that lands and lasts.
