Every CX vendor has an agentic AI offering now, including many players you’ve never heard of. With AI evolving so quickly, there is no shortage of choice, and to varying degrees, most any vendor can do the job. That, of course, is the challenge, as you’re only interested in the offerings that are right for your contact center.
Instead of throwing a dart at the wall of vendors and hoping for the best, CX leaders can – and should – take a strategic approach with agentic AI. Whether taking these offerings from your current CX vendor, or shopping the market, you should be thinking first about use cases where agentic AI will bring real business value.
This article is Part 2 for my analysis, where the first article provided a practical definition of agentic AI for CX, along with internally focused use cases. This is one set of AI use cases, but another set would be customer-facing, where agentic AI interacts directly with your customers. Three leading examples will be addressed below, along with strategic considerations for CX leaders.
Use Case 1: Managing simple inquiries
This will be the most common use case, and one where contact centers can get tangible results quickly. Across all verticals, many customer inquiries are for simple, routine tasks that human agents can easily handle, but really shouldn’t be. Typical examples would be: updating customer profiles, confirming hours of operation, address changes, order inquiries, setting appointments, etc.
Automated options – namely IVR and chatbots – have managed these inquiries for years, but not very well, and we’ve all had bad self-service experiences. With agentic AI, the self-service bar gets raised, since these AI agents can handle a broader range of inquiries, and resolve them far more effectively than existing tools.
Being conversational, they can engage more deeply, and not get derailed by human ambiguity that legacy options cannot process. More important, these inquiries can be handled end-to-end by the AI agent, without input from human agents or supervisors. Not only can agentic AI do a better job with these inquiries on their own, but as customers get comfortable engaging this way, they’ll soon be handling more types of inquiries. Strategy-wise, this is a great use case, not just to raise the self-service bar, but to make the workload for human agents more manageable.
Use Case 2: Resolving customer issues
Strong as the above use case may be, the strategic value of agentic AI is even greater here. These are often more complex scenarios that go beyond passive inquiries and require problem resolution. In legacy contact centers, these often start with self-service channels, but quickly need handing off to a human agent. Current self-service applications can confirm some basic details, but mainly to help route the call to the right agent who can actually resolve the problem, rather than to complete the job independently.
With human-like conversational capabilities, AI agents can understand intent, nuance and sentiment in ways that legacy tools cannot, allowing them to take self-service further than ever before. More important, they can resolve some of these issues autonomously, without escalating to a human agent. Proper guardrails must be in place to entrust AI agents end-to-end, so this use case must be carefully managed.
At present, this level of fully automated self-service is limited, and for good reason – to minimize the risk when AI agents fall short. Human agents will intuitively escalate a call when they know it’s needed, but AI agents don’t have enough self-awareness to do that reliably, so CX leaders must be careful picking their spots here.
This is part of the strategy, as AI agents need to build their track record and earn the right to handle more inquiries autonomously. The payoff will be worth the wait, though, as end-to-end problem resolution by AI agents makes life easier for human agents, and takes operational efficiency to new levels.
Use Case 3: Finding best-fit options for customers
This is a more subtle use case of agentic AI, but it shows the value of AI’s real time capabilities. At various points during a call, there will be different best-fit options, either for resolving an inquiry or making the right hand-off decision. For the former, AI can draw from multiple sources on the fly, and use judgment – within parameters defined by CX leaders – to identify and offer a best-fit solution for the customer.
The more refined your datasets, the better this capability will be, as the AI agent would have to consider a multitude of factors and probable outcomes in making a final decision autonomously. Failing that, the next best-fit option would be identifying the right type of human agent or subject matter expert to hand the call off to.
This goes beyond intelligent call routing – the agentic part has to do with assessing customer sentiment and other qualitative factors, and matching them up against options that will yield the best outcome – all in real time, and done autonomously. Since most datasets are not truly AI-ready yet, this use case will largely be aspirational, but the potential business value for automating this form of customer interaction should be clear for CX leaders.
Strategic considerations
Any of the above use cases can be the basis for an agentic AI strategy since they are all customer-facing. There are, in fact, two strategic considerations in play here. The first has to do with deploying AI agents for use cases that will provide tangible results in short order, both of which will be needed to validate AI adoption. The use cases outlined in Part 1 are operational, and while valid in their own right, the impact can be harder to quantify, will take longer to materialize.
Related to that would be the second consideration, namely the focus on outcomes that matter most to top management. CX itself has become strategic for most businesses, so the strategy for agentic AI is to apply it to use cases that have the greatest strategic value. That would be the customer-facing use cases outlined in this article, and in terms of technology partners, you need to choose a vendor that best aligns with these strategic considerations.
Operational use cases for agentic AI still have value, but in terms of getting off to the right start, the strategy for CX leaders should be to begin with customer-facing applications. The caveat, of course, being that your datasets are AI-ready. If that’s not the case, the safer starting point would be with internal use cases, and while doing that, to also uplevel those datasets so follow-on deployments can effectively handle customer-facing use cases.