Lots of things sound good on paper but fail to deliver in practice.
Unfortunately, contact center AI is at risk of falling into this category.
That isn’t because the technology is lacking. At its best, contact center AI is intuitive, relevant, and easy to leverage without task-switching.
However, far too often, the tools fail to live up to this promise. Indeed, the sheer number of different features that customer service agents need to navigate can actually become more of a hindrance than a help.
There is no denying that a dashboard full of insights, a copilot surfacing suggestions, and a transcript updating in real time are all highly effective features. However, if the integration and design are not up to scratch, they can actually add cognitive load rather than reducing it.
Rémi Guinier, Head of AI Product at Diabolocom, has spent a lot of time watching what actually happens when agents use these tools in live environments, and his conclusion is not what most vendors want to hear:
"We have AI implementations on the market right now that I think are distracting.”
"You're reading your transcript, reading your notes, reading your quick answer, and you're not really thinking about what the customer actually wants, which is against the point. Customer service quality is ultimately a function of where an agent's attention is spent."
That observation sits at the center of how Diabolocom approaches product design, and it connects directly to a broader argument Frédéric Durand, the company's CEO, made in his recent manifesto:
The goal of AI in a contact center is not to remove your people; it is to remove everything standing between the agent and the customer.
Attention is the Scarce Resource
It is worth being precise about what the problem actually is. Agents are not at risk of burnout because customer service is hard; they are at risk because the environment around customer service makes it unnecessarily harder.
Now, that isn’t to say that customer service is easy. It is a difficult profession that requires a great deal of skills and experience. But many organizations are actually making things more complicated for their teams.
For example, 74% of contact center agents are required to navigate multiple screens within a single live interaction. Every context switch, manual note, and knowledge lookup is a withdrawal from the attention the customer deserves.
"It's not intelligence that's scarce," Guinier says. "You have lots of tools that provide intelligence at any given time. It's attention.
"Customer service is about being at the service of the customer, and we're built to talk to each other in a way that AI still can't replicate. We say attention is the product."
There is also a second side to this. When agents spend the majority of their day on repetitive, low-complexity interactions, their ability to handle the hard ones deteriorates, as Guinier explains:
"If 95% of the calls you take are password resets, what are you actually developing?"
Automating routine work does not just free up time; it improves the quality of the cases that reach human agents and, by extension, the quality of attention those agents can bring.
What Unobtrusive AI Actually Looks Like
The distinction Diabolocom draws between AI that assists and AI that distracts is essential to the vendor's product suite.
