While generative AI hype has been intense, the most striking thing for me has been how little AI has been adopted in customer service in 2025. The cautionary barriers have been robust but that’s going to change in the next 12 months or so.
Business leaders are going to be applying new pressure on the customer service leadership to increase AI adoption in more meaningful ways than the small and isolated pilots that have tended to happen so far. Any caution about using generative AI and greater automation has been eclipsed by how much business leaders are convinced that inter-weaving AI into their operations will reduce costs and boost service quality. Progress and adoption will vary by industry, with financial services likely to be early adopters.
And yet, the industry will need to guard against any over-reach on embedding more AI into customer service processes. Past much trumpeted technology innovations have landed poorly with customers who have learnt to hate interactive voice recognition systems. There’s a similar disenchantment developing around how AI generates AI slop that’s inauthentic and unpleasant. Trust in how AI is adopted in customer service will need to be carefully nurtured and guarded to make it acceptable and thus actually deliver the business outcomes organisations want for using the technology.
Enter Agentic AI but only on Solid Foundations
The great opportunity and challenge for applying AI in customer service is how it could transform self service during 2026.
Let’s be honest that so far technology has fallen short in its promise to make self service truly effective.
But, the maturity of agentic AI technology could re-write how chat and voice bots become truly autonomous in how they are able to engage with customers with much more personalised and valuable answers. The prize to be won here is how chatbots can answer customer queries without the need to switch to a human agent more successfully and consistently than today. The bottom line payback from customer service organisations is slashing costs per customer interaction from $10 to $2 when agentic AI is used.
That’s the reward but how successful the adoption of agentic AI is in 2026 will depend on how much organisations have laid the right foundations. Those who simply automate existing processes will end 2026 disappointed. What’s going to ensure true transformation of how customer service performs is how the AI-powered revolution rethinks how processes are done, tears up bad to indifferent practices and shakes up the call centre model. The successful customer service agentic AI deployments will be those where AI agents are harnessed to workflows that have been rethought and optimised to deliver value for both the business and the customer.
Don’t Expect Humans Kicked out of the Call Centre
While we expect to see AI make a more significant impact, 2026 will be a transition year for how technology changes the role of human service agents.
There is a definite pathway from how today’s call centres are packed with service generalists to how by 2030 agent roles shift to being much more specialised while routine and triage tasks will be fully automated.
While the fully transformed model is four or so years away, we should expect more service representatives finding themselves receiving more AI-generated assistance and automation of data entry and reporting. There will also be early signs of new AI-related human roles who supervise the use of agentic AI “workers” alongside the human team members.
Indeed, this trend is likely to test how much call centre leaders become adept at overseeing these new digital workers. Traditionally, dealing with human workers in a call centre has stretched leaders who have had to resolve all kinds of disputes and issues. 2026 could see a few of them grapple with ensuring responsible agentic AI behaviours and mitigating the churning out of AI workslop that includes poor to deranged AI-based recommendations to service representatives.
Being more attuned to what makes good AI as distinct from bad AI underlines how customer service leadership must become more technologically aware in supervising the expansion of AI capabilities over the next four or more years.
More Customers Weapon Up with Gen AI to Complain
While the progress of AI into customer service centres has been more cautious than expected, consumer adoption of generative AI has accelerated. One trend that will impact customer service organisation is how much aggrieved customers reach for popular gen AI tools to do the heavy lifting of drawing up complaint letters. How much this puts pressure on case management systems will be something that the industry will need to assess as 2026 unfolds. Already we are seeing an angrier mood about customer service shortcomings so perhaps more customers arming themselves with gen AI will put more pressure on organisations to automate more of their case management functions.
Technology Enables Compassion at Scale
Supporting vulnerable customers in the past has been something of a compliance checkbox rather than a genuine commitment for enterprises. Vulnerable individuals, whether dealing with mental health issues, physical disabilities or financial hardship, have been forced to repeat their painful stories to every organisation they interact with. This repetitive disclosure isn’t just emotionally exhausting for customers; it’s operationally expensive and inefficient for organisations who lack visibility into vulnerability status across their customer base.
The challenge has always been obvious yet persistently unresolved. That’s likely to change in 2026 as organizations wake up to the moral imperative and the business case for improving how they engage with at risk individuals. With regulators such as the Financial Conduct Authority applying more pressure, along with the availability of new technologies such as the Vulnerability Registration Service in the UK, there is likely to be a fundamental shift in how customer service teams identity and support vulnerable customers.
About the Author
Simon Thorpe is a customer service tech expert at Pegasystems.
