Words matter.
The Collins Dictionary defines agent as a person who looks after someone else’s business affairs or does business on their behalf.
If only things were that clear and simple. While most other dictionary sources specify an agent only as a person, others include entity or business with the definition. This may sound like splitting hairs, but it opens the door to non-human agents, and that takes us to a term that is coming into wider use now – AI agents.
My focus here is specifically on how the term “agent” is used in the contact centers, where it has unequivocally meant human agents – people who dealt directly with other humans, namely customers. The term AI agent has been gaining currency in other areas of the enterprise since 2023, but only now is it turning up in the CX narrative. Before addressing the broader use of AI agents, it’s worth noting how hard-wired the term agent is in the business world.
Outside the contact center, there are many other types of agents, and their role is basically the same as customer service agents – insurance agents, travel agents, sales agents, etc. Since all of these roles pre-date the Internet, it’s not necessary to specify that these are human agents – there have been no other types of agents. Until now.
English is a constantly evolving language, and as AI strives to emulate every facet of humanity, the broader definition of agent to encompass AI’s expanding role doesn’t seem so jarring. We are well along the path now whereby non-human, AI-based agents are taking on a wider range of roles – interacting both with humans and non-humans - as in other AI-driven agents. As this becomes normalized, the distinction between human and non-human agents will become less clear – presuming, of course, that non-human agents can perform their tasks effectively.
This gradual disappearance of the human agent/non-human agent distinction is a subtle example of how AI is transformative by nature, and this really crystallized for me at Verint’s annual customer event, Engage 2024, which I just attended. Verint’s focus is CX automation, where the role of AI is to power purpose-built bots to automate specific CX-impacting tasks. Using a multitude of bots to tackle a multitude of specific problems is very much a bottom-up approach, one that adds up to substantive outcomes, both for CX and operations.
While not fully autonomous, these bots are performing many of the tasks that agents do (or doing things that help agents do them better), and being AI-driven, they are referred to now as AI agents. The pairing of these two words is new, and represents a shift in language, where AI is the paradigm that re-defines the meaning of words.
Previously, contact center vendors have termed this very differently – bots, virtual assistants, intelligent virtual agents, etc. While all of these are AI-driven, they were not referred to – or considered as agents. Their role is to support human agents – take on simple workflows or handle routine customer inquiries – not to be agents, handling customer inquiries directly in an autonomous fashion. The human agent always had the last word, but that is on the verge of changing.
Again, this is where words matter. Verint is not explicitly talking about AI agents as being autonomous, and on a path to being on par with human agents; at least for now. Even from a year ago, Verint has made great strides in the capabilities of their bots, and increased forms of autonomy are not really a big stretch now for them. The very thought of fully autonomous AI agents should trigger all kinds of scenarios, both good and bad. This why AI is transformative for the contact center, as it will inevitably lead to replacing some human agents, but it could also be a game-changer for improving CX.
I should also note that Verint is not alone here, as other contact center vendors are now talking about AI agents, and some like NICE have forward-thinking visions of fully autonomous AI agents. In the spirit of AI-for-good, these AI agents are positioned as being complementary to human agents – not as a replacement – where they work hand-in-hand, so to speak, supporting each other and learning from each other.
For now, most use cases for autonomous or semi-autonomous AI agents are internal, where they can’t do much harm with customer-facing interactions that go awry. They can be very good at automating tasks and workflows internally, and to some degree, handling simple or routine customer inquiries with minimal or even no human assistance.
As such, you’d be right to still think about bots as having limited capabilities, and there only being one type agent – a human customer service agent. With references to AI agents being pretty nascent in the contact center space, this may not register very high on your priority list.
However, the topic of AI agents is more top-of-mind now for other enterprise applications, some of which are quite adjacent to the world of CX. None more so than Salesforce, who loudly announced Agentforce at their recent Dreamforce event. This is their next-level form of autonomous AI agents that can handle a wide variety of tasks end-to-end, relieving human workers of routine workflows, and allowing them focus on higher-value needs.
The natural use cases would be for business functions that have a lot of routinized processes and workflows – sales, marketing, commerce, and yes, customer service. In terms of automation, these AI agents are more advanced than what existing virtual assistants can already do. The big change is their ability to function autonomously, without human supervision, to do things such as make decisions, create plans, execute plans, and even reason – just like humans do. Furthermore, they are designed to do so in a personalized manner, again, just like humans do.
With CRM being such a big adjacency to contact center vendors, the halo effect from Agentforce should be substantial, and will likely force these vendors to up their game with AI agents. The story is much bigger however, as Salesforce is just one of many large AI players and hyperscalers who see AI agents as their entry into the next wave of enterprise tech adoption.
The AI agent ecosystem is taking shape now, and is poised to be a leading tech trend in 2025. Prime examples would be ChatGPT Enterprise from OpenAI, Claude for Enterprise from Anthropic, and Agent for Bedrock from Amazon. While much of this is about reducing the cost and time for coding to develop new applications, the applications are focused on making people and teams more creative, and better able to manage complex challenges in their business. This goes well beyond what first-generation bots and virtual assistants were built to do.
These companies are setting the bar high for an AI-driven future, but all the normal cautions very much apply here – accuracy, transparency, ethics, privacy, security, etc. Presuming these can be credibly addressed, the prospect of doing all this at scale and wire speed will surely change our thinking about what an agent is and can be. Even if they get it half-right, I don’t think there’s any turning back now, and that the term AI agent is going to soon pass into the CX vernacular as if it was always there.
This post is written on behalf of BCStrategies, an industry resource for enterprises, vendors, system integrators, and anyone interested in the growing business communications arena. A supplier of objective information on business communications, BCStrategies is supported by an alliance of leading communication industry advisors, analysts, and consultants who have worked in the various segments of the dynamic business communications market.
