Las Vegas — It is rather odd, when you think about it, that the conversation around AI in the contact center has become so binary. It’s either the robots take over, or the humans stay entrenched. It feels a bit like a zero-sum game.
But sitting down with Pasquale DeMaio, Vice President of Amazon Connect, at AWS re:Invent 2025, you get the sense that the reality is far more nuanced. DeMaio calls it the “gentle continuum.”
The premise is simple yet relieving for the anxious CX leader: You don’t have to choose.
The End of the Binary Choice
During his keynote, DeMaio outlined a vision where businesses can be 100% agentic, 100% human, or—more likely—somewhere messy and effective in the middle.
“We believe that AI and people will get a lot better together,” DeMaio told the audience.
Later, when we spoke one-on-one, he expanded on this. “You want that true concierge experience that brings all the context up front from those automated experiences… all the way to the agent to make them superhuman.”
This is a significant shift. For years, the industry has treated self-service as a way to deflect calls away from expensive humans. Amazon Connect’s approach, bolstered by 29 new features launched this week, treats AI as a teammate that carries the luggage—or in this case, the context—so the human agent doesn’t have to ask you for your account number for the third time.
Escaping the “Franken-stack”
The problem, of course, is that most contact centers are currently held together by hope and duct tape.
DeMaio described the plight of leaders stuck between “legacy platforms that don’t really understand AI” and flashy point solutions that promise to fix everything but fail at complexity. It puts leaders between a rock and a hard place.
“The hard part is the hard part”
DeMaio noted with deadpan accuracy.
The solution isn’t necessarily to rip everything out on a Tuesday. DeMaio’s advice is pragmatic: “Pick a workload. Move fast, learn, iterate.” You can move your automated self-service to Connect while leaving other parts on legacy systems, creating a bridge rather than a cliff edge.
Action Over Chat
The buzzword of the week is “Agentic,” but for once, it seems to mean something substantial. It is not just about a bot chatting politely. It is about a bot doing things.
During the keynote, we saw demos of AI agents that didn’t just apologize for a delayed flight but proactively rebooked it based on the traveler’s history. It’s the difference between a chatbot that offers sympathy and one that offers a boarding pass.
The Centrica Proof Point
If this all sounds a bit theoretical, the numbers from Centrica suggest otherwise. James Boswell, from the 200-year-old energy giant, took the stage to share how they pivoted to the cloud.
The results were rather impressive. By using Bedrock to analyze transcripts and “auto-wrap” calls, they reduced talk time by 30 seconds.
But here is the interesting part. They didn’t just bank the savings. “What we’ve been able to do with the extra 30 seconds is cross-sell,” Boswell explained.
They also saw a 28% reduction in complaints and an 89% increase in NPS. It turns out that when you remove the “cognitive load” from agents—stopping them from having to “brain Alt-Tab” between 10 different systems—they become much better at actually talking to people.
A “Teammate” Mentality
I began to wonder if we have been looking at the AI problem upside down. We worry about control and security—valid concerns that DeMaio acknowledges are keeping leaders up at night.
But the “gentle continuum” offers a way to innovate without losing that control. It allows for “deterministic workflows” (the rules you absolutely need to follow) to coexist with generative capabilities.
“The hardest part of working out is putting on your sneakers”
DeMaio quipped during the session.
Amazon Connect seems determined to make sure that when companies finally do put those sneakers on, they aren’t running toward a cliff, but toward a model where humans and AI actually get along.
