Ten years ago, one word dominated the contact center conversation: omnichannel.
Yet, the word became so overused that it created a lot of confusion in the market.
Is omnichannel simply offering lots of communication channels beyond voice? No, that’s multichannel.
Should omnichannel allow customers and agents to shift between channels mid-conversation without starting the entire conversation over? In theory, yes, an omnichannel contact center should ensure that the context of a conversation follows the customer as they switch between channels.
However, many solutions that are marketed as omnichannel tend to fit the multichannel definition much more closely.
Unfortunately, the over-marketing of these solutions resulted in many service leaders taking a long time to fully grasp what such an omnichannel contact center could look like in practice.
In many cases, that confusion has stuck, continuing to hamper customer service experiences.
2023 ContactBabel research backs this up. It found that fewer than a third of contact centers are truly omnichannel.
With agentic AI, the contact center industry should be wary of history repeating itself.
What Agentic AI Is Not
‘Agentic AI’ and ‘AI Agents’ are not new terms for ‘chatbots’ or ‘virtual agents’.
Nevertheless, as with ‘omnichannel’, many contact center vendors have already confused the market by co-opting these terms for their existing self-service solutions - whether or not they meet the criteria for true agentic AI.
Liz Miller, VP & Principal Analyst at Constellation Research, recently made this point at the Adobe Digital Experience Conference. She stated:
Definitions are important, and – right now – we do not have a good definition of agents, and we are not holding people accountable to those definitions of agents or “agentic”. We should be.
The risk of letting these definitions slide is that the market confusion ultimately prevents customers from realizing the technology’s full potential.
So, How Should Contact Centers Define AI Agents?
There are three core characteristics of an AI agent.
First, it should be interactive and able to interpret communications.
Second, it should be able to reason, taking those communications, understanding context, and making decisions.
Third - and most importantly - is the concept of agency - meaning it should be autonomous and able to analyze information, plan, and take actions on its own.
Let’s consider the use case of a customer-facing AI agent looking to resolve incoming queries.
If it’s following a series of rules, that’s not an AI agent. If it’s scouring knowledge articles and spotlighting relevant information, that’s also not an AI agent.
Instead, an AI agent will think through the problem, only respond when confident in their answer, and – when it’s not – escalate the contact.
