As we move further into 2025, the customer service landscape is once again on the brink of transformation — this time, thanks to significant evolution in artificial intelligence capabilities.
Agentic AI is positioned as the next major leap forward, agentic AI goes beyond automating simple tasks or generating content. It promises something far more ambitious: the ability for AI agents to make complex decisions, handle entire workflows, initiate actions, and function independently of human oversight.
But just how close are we to seeing this level of intelligent automation in UK contact centres? And more importantly, what should customer service leaders be doing now to prepare?
To help separate hype from reality, let’s explore what agentic AI really is, where it fits in today’s AI landscape, and how organisations can prepare for its eventual arrival.
What Is Agentic AI?
Before we talk timelines and readiness, it’s worth understanding what agentic AI is and what makes agentic AI different from the traditional automations.
Agentic AI can operate autonomously toward goals, taking initiative, reasoning through tasks, and making decisions with minimal human input. Unlike rule-based bots or narrow AI assistants, agentic AI can adapt and manage complex, dynamic tasks — like those found in customer service. Agentic AI is built on the foundations of generative AI and multimodal AI.
Generative AI enables AI systems to understand, synthesize, and generate human-like content — text, code, and more — based on massive datasets. This has been a core enabler for tools like chatbots and virtual assistants.
Multimodal AI took things a step further. Rather than working with voice or text as input, it can process and act on multiple formats — images, voice, video, and text, together. In customer service, this has already led to breakthroughs like automated claims processing, where AI analyses submitted photos, validates information, and initiates workflows.
Agentic AI builds on all of this. But what sets it apart is its ability to reason, make decisions, and act autonomously across an entire task or workflow, even verifying its own results along the way.
Why the Hype Now?
Part of agentic AI’s growing buzz is driven by falling computing costs and advances in large language models (LLMs), such as DeepSeek, that make it more cost-effective for businesses to deploy increasingly sophisticated AI tools.
Every time the cost of intelligence drops, the business case for automation gets stronger. Agentic AI is exciting because it’s the closest we’ve come to unlocking truly autonomous digital workers — AIs that can manage more complex customer journeys without handholding.
So… How Close Are We?
Not as close as the headlines might have you believe.
According to recent industry data, fewer than 1 percent of enterprise applications were using agentic AI in 2024. But that statistic can be misleading. It helps to think of AI maturity in five stages of “agency,” or how independently an AI system can operate:
- Do as I say – Task-based automation, like robotic process automation (RPA), already in wide use.
- Research and tell me – AI aggregates data and surfaces insights; already happening with multimodal models today.
- Recommend to me – AI synthesizes data to make recommendations, with a human still in the loop.
- Decide and tell me about it – Minimal human oversight; AI executes decisions and provides a report.
- Do it and don’t bother me – Fully autonomous agents executing end-to-end workflows.
At this point, levels one through three are achievable and already delivering value. Levels four and five … that’s where the true promise of agentic AI lies, but they’re not enterprise-ready yet.
What’s Holding It Back?
There are several key roadblocks preventing agentic AI from taking full flight in contact centres.
- Legal and regulatory concerns – As decision-making shifts from humans to machines, accountability becomes a pressing issue. In tightly regulated sectors, this poses significant risks.
- Data and system access – Agentic AI needs access to far more than just customer records. It must interface with internal systems, APIs, and structured processes to act effectively, something many businesses haven’t yet mapped out or made accessible.
- Real-time adaptability – AI models are generally “frozen” at the point of training. Unlike humans, who adjust their decisions based on the latest information, current agentic systems often lack that up-to-the-minute awareness.
How Can Organisations Prepare in 2025?
Even though we’re still a year or more away from widespread agentic AI deployment, there are clear steps forward. The smart move now is to experiment with what’s already available. This means focusing on agentic AI’s early stages — levels one through three — which are already capable of driving measurable return on investment (ROI).
More importantly, businesses should start laying the groundwork for levels four and five:
- Map your customer service workflows – Understand every decision point a human agent makes, the systems involved, and the data required.
- Implement BPM (business process management) – A structured approach to analyzing and optimizing business processes will surface the best opportunities for automation.
- Develop AI governance principles – Ensure your approach to AI adoption is responsible, auditable, and future-ready.
The road to fully autonomous customer service may still be under construction, but the direction is clear. Agentic AI represents a significant evolution in how businesses will support their customers, one that goes beyond scripts and rules, toward truly digital reasoning and action.
Agentic AI might not be ready to run your contact centre today, but the decisions you make in 2025 will determine how ready you are when it is.
For customer service managers across the UK, the message is simple: prepare now or play catch-up later.
About the Author
Mahadevan Meikum Perumal is VP, EMEA – Head of Delivery & Client Success at TTEC Digital.
Mahadevan is an accomplished technology leader with over 26 years of experience across industries such as financial services, TeleCos and Hitech. He currently serves as the Vice President, Head of Delivery and Client Success, EMEA at TTEC Digital, a position he has held since October 2021. Mahadevan’s expertise includes CRM, Data & Analytics and delivering large digital transformation programs. He is known for his customer-first approach, ensuring that client needs are at the forefront of his strategies and operations, which has consistently led to high levels of customer satisfaction and loyalty.
Prior to his current role, Mahadevan led EMEA Microsoft Biz App practice and Data & Analytics practice for a GSI’s. Mahadevan has working knowledge in Agentic AI, a cutting-edge technology that combines advanced AI and automation to create autonomous AI agents capable of complex decision-making and process execution. His understanding of Agentic AI positions him at the forefront of innovation, enabling him to leverage this technology to drive efficiency and transformation for TTEC Digital’s clients.
