From Cost Center to Revenue Engine: How Amazon Connect is Revolutionizing Contact Centers with 25 Years of AI

I’m wandering through the halls of the Venetian at AWS re:Invent, and I’ve got this nagging thought about contact centers. We’ve always called them cost centers, haven’t we? Places where we send money to disappear, or at least to be spent apologizing to people for late deliveries.

But sitting in session BIZ219 at this year’s AWS re:Invent with Ayesha Borker, Principal Solution Architect and Jack Hutton, Principal Product Manager within the Amazon Connect at AWS, I began to wonder if we’ve got it all wrong.

They’re pitching Amazon Connect not just as a way to fix problems but as a growth engine. It’s a bit of a pivot. We are moving from the era of “How quickly can I get you off the phone?” to “What else can I help you buy?” And it’s powered by AI.

Importing 25 Years of Retail Wisdom

The most compelling part of the pitch wasn’t actually the new features. It was the history lesson.

Ayesha Borker reminded the room that Amazon has been rehearsing for this moment since 1998. That was when they launched their first recommendation engine—the “customers who bought this also bought that” feature that we now take for granted.

For 25 years, the Amazon Marketplace has been a massive laboratory for understanding human intent. They have learned how to predict what you want before you know you want it. They have learned how to reduce churn by showing you the right alternative when your size is out of stock.

Now, they are simply lifting that entire infrastructure—the algorithms, the predictive attributes, the “item-to-item” filtering—and dropping it directly into Amazon Connect.

It solves the classic problem Ayesha highlighted:

“I’ve got the data, but I don’t know what to do with it.”

By baking models like “Frequently Paired,” “Trending Now,” and “Recommended for You” into the contact center, they are giving agents the same superpowers that power the Amazon homepage.

The “Stella” Effect

In the session, they showed a demo involving an AI stylist named “Stella.” A customer, Maya, was looking for a dress for a wedding in Napa. Stella didn’t just search for “dress.” She checked the context. She realized it was an outdoor wedding. She suggested a lighter fabric.

When Maya picked the dress, the “Frequently Paired” algorithm immediately popped up with a matching necklace.

It was smooth.

For the agent on the other end, this changes the game. Instead of scrambling to find a solution, they have a “whisper” in their ear offering a revenue opportunity. It turns a service recovery call into a potential sale.

The “Help Me Decide” Factor

One of the more human touches Jack mentioned was the “Help Me Decide” feature. We all suffer from decision fatigue. You look for a flight. There are fifty options. You end up closing the laptop in frustration.

Amazon Connect is trying to replicate that old-school shop assistant who knew your taste. By embedding generative AI agents into the app or web chat, they are moving from passive search to active curation. It removes the mental load for the customer. It’s less about “Here is a list of ten things” and more about “Here is the one thing you actually need.”

Cleaning the Plumbing Before Painting the House

Of course, none of this works if your data is a mess. And let’s be honest, it usually is.

Jack dropped a rather startling stat from United Airlines. During a review, they found duplicate entries for 50% of their loyal customers. That is a bit of a muddle. You can’t offer hyper-personalized service if you don’t know that the person calling is the same person who emailed yesterday.

This is where the new “Entity Resolution” feature comes in. It’s the unsexy plumbing that makes the magic possible. By using AI to scan and resolve these duplicates, United reduced those errors by 35% and cut infrastructure costs by 30%. It’s a quiet win, but a necessary one.

The numbers Ayesha and Jack shared were rather compelling. They noted that organizations implementing this level of personalization are seeing a 10-15% lift in revenue. GoStudent, another customer cited, saw a 20% increase in engagement activity by better matching students to tutors using these profiles.

It leaves you with the sense that the contact center is finally growing up. It’s no longer just the department that cleans up the mess. It’s becoming the department that drives the business forward. And for the agents on the frontline, that might just make the job a little more interesting.

For all our event coverage, announcements and exclusive interviews visit our AWS re:Invent news hub here