Agentic AI has exploded into our collective consciousness. PwC reports that it’s driving nearly 9 in 10 US companies to increase their AI budgets. Three-quarters of respondents even (allegedly) believe that these AI agents will reshape the workplace “more than the internet did”.
It’s easy to see why. The concept of autonomous, goal-oriented and logic-driven AI agents capable of completing complex, multistep tasks and making their own decisions sounds like the stuff of sci-fi novels. So why then is Gartner forecasting that 2 in 5 agentic AI projects will be cancelled in the next two years?
Spiraling costs, unclear business value, or inadequate risk controls are often to blame. The technology clearly holds vast potential. But success requires focused, meaningful, and deliberate application, a true appreciation of the efforts involved in both establishing and maintaining AI agents, and a proven operating model.
A New Conversational Model
Customer service and support is by far the most mature use case. It promises a significant return against measurable, value-driving metrics. Confidence in this use case is so high, in fact, that Gartner also predicts that Agentic AI will resolve up to 80% of common customer service issues without human intervention by 2029, resulting in a 30% reduction in operational costs.
But achieving this level of success isn’t a simple one-and-done task. Don’t think of agentic AI as another add-on, a supplementary tool to tack on to the side of your customer call center. What we’re talking about is a reorchestration of your customer service and support structure to one led by AI – a fundamental reimagining of the customer experience model.
Human-first models are almost invariably seen as cost centers. These phone-centric operations are measured by their brevity, their containment, where scale is minimized to keep overheads down. There is often a hard disconnect between communications channels, while fragmented systems require information to be duplicated in parallel. This all adds up to repetitive, costly, manual work.
AI-first contact centers take a different approach. Everything from CRM to payments to social media integrates into a unified orchestration layer in which the AI acts as the meta-orchestrator, connecting interactions across each channel and system.
Conversational AI provides business logic, structure, and guardrails, and enables AI agents to engage in real-time dialogues with users and understand the context, even across multi-turn conversations. This combines with generative AI to produce human-like and personalized interactions, not bound by limited pre-programmed responses.
As a result, AI agents can proactively carry out processes and communication, both customer-facing and backend. Every action taps all relevant data, workflows, and systems in real-time. So, regardless of whether a customer starts with a website chat, phone call, or social media message, each interaction is unified, consistent, and smooth, with no repetition when switching between channels.
By taking on the majority of straightforward tasks, human agents are empowered to focus on complex, high-value tasks. Centers can move from reactive to proactive engagement, contacting customers to remedy situations as they arise, reminding them of upcoming invoices and payment due dates, and even inquiring about abandoned purchases and offering incentives.
It’s a hugely compelling proposition, but not without its challenges.
No Silver Bullet
The strength of Agentic AI lies in its integration. Being able to proactively draw on and execute across multiple systems is intrinsic to how it delivers value and distinguishes itself from advanced chatbots. However, enterprises with fragmented and legacy infrastructure are prone to data silos which can significantly hinder an agent’s ability to perform end-to-end tasks or provide relevant responses. System access must be possible. Any potential pilot requires a thorough audit of the tech stack for integration feasibility first.
For much the same reasons, Agentic AI should not be considered a silver bullet. There will be limits to its scope, especially when rolling out initial proofs of concept. Focus on delivering against specific, measurable outcomes that build internal confidence and prove ROI, then gradually scale from there.
Starting small and scaling in this manner also helps with managing complexity and support levels. Whilst the goal may be to help reduce team workloads through AI agents, they themselves require management. Consider the ongoing development and testing, security and privacy, data stewardship, brand voice, and regulatory requirements and oversight required, especially in use cases that collect and use personal information, such as payments and financial information.
A Bold Future
Despite these challenges, the promise of conversational agentic AI is undeniable. When focused on value-driving applications like customer experience and support, it has the potential to transform entire functions from sunk costs to growth drivers. Those who embrace AI-first strategies, who are bold enough to reimagine their approach, stand to gain a decisive advantage.
The payoff is more than operational efficiency. More than improved experiences, even. More than new sales opportunities, and loyalty drivers, and morale boosters. It’s a redefinition of what customer service can be in the age of intelligent automation. And it will determine true winners and those who will be left behind.
