AI is positively transforming the customer journey. Recent research from the Institute of Customer Service shows that UK consumer satisfaction is continuing its recent ascent. This sustained improvement in end-user sentiment correlates with the rise of AI in Customer Experience (CX).
Intelligent automation in CX isn’t new, of course, but the category has been supercharged in recent years by AI-driven functionality, with generative AI, in particular, powering many of 2025’s advancements. The next phase of CX transformation is being driven by agentic AI, artificial intelligence that can work towards objectives: planning and acting with increasing autonomy. These advancements are redefining the role of AI in CX by keeping humans “in the loop” while automating more and more of the administrative tasks that account for over 50% of a traditional contact centre worker’s time.
Gartner predicts that by 2029, 80% of customer interactions will occur without any human involvement, powered by agentic AI. That sounds like a big figure, however, it is important to recognise that a significant proportion of today’s interactions have already been automated.
The true value of agentic AI lies not in full automation alone, but in its ability to orchestrate seamless collaboration across an entire customer journey. To realise the benefit of how agentic AI can support voice and digital interactions, organisations need to focus on how it can play a role at every stage.
Before the Interaction
Agentic AI is most powerful when implemented at the very start of a customer interaction – either eliminating or shortening any encounter with a human agent. While customers are waiting in a queue, AI agents can gather contextual information, understand the reason for the contact, and assess urgency and complexity.
For non-urgent, low-emotion, and low-complexity enquiries, such as status updates or password resets, the AI agent can usually resolve the issue independently, within the limits of its guardrails. Where an enquiry is urgent or complex, the interaction is intelligently routed to the most appropriate available human agent, with all relevant context passed through to minimise handover time and repetition.
Agentic AI implemented at the start of an interaction is especially valuable for organisations serving multinational audiences. AI Agents can automatically translate customer input and respond in thatcustomer’s preferred language, reducing reliance on translation services and ensuring that non-native English speakers receive a high-quality, consistent customer experience.
During the Interaction
During live interactions, AI Agents can act as real-time assistants for human agents. Agentic assistants can ‘listen in’ on a conversation and use live interaction transcription and sentiment analysis(the latter subject to local legal clearance) coupled with existing data such as customer history and company policies, to recommend next-best-actions and execute back-office tasks during the conversation. While these capabilities are similar to those of generative AI assistants, agentic AI differs in its ability to reason more deeply over the data it consumes before taking action.
Agentic AI’s reasoning capability becomes considerably more powerful when combined with rich datasets held within the Customer Data Platform (CDP). The CDP sits across the enterprise’s other systems of record, such as CRMs, to provide the central repository, coordinating authority and system of action for all AI-related information.
Despite increasing levels of automation, human agents remain critical. Nearly half (42%) of CX professionals believe that human agents will continue to be central to customer experience delivery. This view is reinforced by analysts’ expectations of regulatory developments, with Gartner predicting that by 2028 the EU will mandate a “right to talk to a human” in customer service.
After the Interaction
Post-interaction administration represents a significant operational burden for human agents. Contact centre workers can spend up to 60% of their time on wrap-up activities, such as updating customer records and summarising interactions.
AI can dramatically reduce the post-contact burden by using the real-time transcripts generated during an interaction to automatically populate forms and generate accurate encounter summaries.Agentic AI reasoning can also take the process further, by analysing next steps and carrying out follow-up tasks such as raising a ticket, emailing the customer with a status update, or reminding ahuman agent to follow up with the customer after a set period of time.
Contact Centre supervisors also benefit from agentic AI-enabled Quality Management (AQM), which can score 100% of interactions against predefined criteria, reducing manual review effort and increasing the consistency of Quality Management for voice and other interactions, when compared with traditional ‘random sampling’ techniques.
End-to-End Self-Service and Proactive Outreach
Beyond live agent interactions, agentic AI also enables far more sophisticated self-service capabilities than were previously possible. Customers can check order status, update personal details, obtain refunds (subject to guardrail permissions) and schedule appointments across both digital and voice channels without human intervention. Crucially, these capabilities deliver complete resolutions rather than partial hand-offs. The goal now is containment, rather than deflection.
In addition, agentic AI can monitor customer journey signals, such as repeat contacts or channel-hopping, and proactively trigger outreach before dissatisfaction escalates. Such autonomous decision-making is what differentiates agentic AI from previous generations of intelligent automation.
Agentic-Enabled Human Agents: The Future of CX
Intelligent automation has evolved through a series of leaps over the past two decades, and agentic AI Agents represent the next chapter in that story. Generative AI – whose need for organised, accurate and comprehensive data prompted the rise of the CDP – laid much of the groundwork, with organisations now looking to agentic systems to deliver progressively greater levels of operational efficiency and CX consistency.
Success depends on measured deployment. CX leaders should prioritise use cases where baseline performance metrics are already established, the resultant performance delta enabling a clearer assessment of return on investment (ROI). ROI is critical, given that more than 40% of agentic AI projects are predicted to be cancelled by the end of next year due to the lack of it.
