AWS re:Invent 2025 | Pasquale DeMaio, AWS

Clips
More from AWS re:Invent 2025

Pasquale DeMaio

VP of Amazon Connect

AWS

play_circle_outline John Furrier introduces AWS re:Invent 2025 and highlights the evolution of cloud computing.

play_circle_outline Amazon Connect launched eight years ago, focusing on AI integration from the start.

play_circle_outline Amazon AWS Connect's PQ Reports on Doubling AI-Enhanced Experiences and Growth Success in Customer Engagement

play_circle_outline Emphasis on democratizing AI for diverse customer use cases across industries.

Info
Transcript

Pasquale DeMaio, AWS

Pasquale DeMaio

VP of Amazon Connect AWS

In this interview from AWS re:Invent 2025, Pasquale DeMaio, vice president of Amazon Connect at AWS, joins theCUBE’s John Furrier to discuss the rapid evolution of the contact center into an AI-native powerhouse. DeMaio details the service's explosive growth, highlighting a major milestone of reaching $1 billion in business and doubling customer interaction minutes to 12 billion in just one year. The conversation focuses on how Amazon Connect is democratizing access to advanced technologies, unveiling 29 new announcements designed to streamline operations and... Read more

explore Keep Exploring

What insights were shared about the evolution of cloud computing and the emergence of AI-native capabilities during AWS re:Invent 2025? add

What significant developments and changes have occurred in the service provided by the company over the past eight years? add

What recent achievements and statistics can be shared about the company's growth and customer interactions with their software? add

What is the purpose and impact of the AI solutions being developed and launched? add

bolt Powered by CUBE AI

Pasquale DeMaio, AWS

search

John Furrier

>> Welcome back, everyone, to theCUBE's live coverage of AWS re:Invent 2025. I'm John Furrier, host of theCUBE. This is my 13th re:Invent. Been to most of them, except for the original. And it's been a great journey to see how the transformation of just the basic concept of abstracting away the infrastructure and then higher-level services have really grown the market, obviously made cloud computing. But now, we're at a new area where the agentic infrastructure is abstracting away work and data and new capabilities, AI-native capabilities and new levels of services are coming on. We got a great guest here, PQ, who's the VP of Amazon AWS Connect, one of the most successful projects that was born out of the scale of AWS. PQ, thank you for coming on theCUBE. Appreciate it.

Pasquale DeMaio

>> Yeah, thanks for having me. It's wonderful to be here.

John Furrier

>> Amazon Connect is huge. So, first of all, scope the size of AWS Amazon... It's Amazon Connect, right?

Pasquale DeMaio

>> Amazon Connect.

John Furrier

>> Amazon Connect?

Pasquale DeMaio

>> Yeah.

John Furrier

>> AWS Amazon. Scope the size of it because it is really one of the working examples of what has come out of the scale of AWS. As a new service and COVID hit, and then the rest is history.

Pasquale DeMaio

>> Yeah, it's been quite a ride. I mean, we launched eight years ago and I always joke it's another overnight success, but we launched with a very simple service. It was just about giving the power of AI to customers natively. That's only accelerated in the last few years. And now, with our launches this week, and we had 29 of them, we've moved to a size of a system where AI is touching every touchpoint across the entire journey of a customer, whether that starts in some place where they're self-servicing and getting helped by AI and now they can do it with a fully agentic AI. All the way to a human being who are on the other side. And really, we think of those as AI teammates for those humans to be able to then take that full context there. And then, after everything's wrapped up and hopefully they've solved the customer's problem or helped the customer achieve their goals. At that point, then you can even take on all that lame work that nobody wants to do afterwards, like taking notes or trying to do codes. And you probably didn't hire those agents because they were good at taking notes in the first place, so the AI does a much better job and it never gets bored. So, between those two things, we're really taking AI, making it better by the human side of it, and then making the human superhuman with the AI.

John Furrier

>> And this came out of Amazon's internal IP. Revenue, you guys had a milestone that was talked about here. I'll let you deliver the number.

Pasquale DeMaio

>> Yeah. Well, it's very exciting for us. A billion dollars is... A few businesses like that, it starts to add up as they say, you got something really going on. But for us, we're very mission-driven in the team. And so, we'll take a moment to celebrate the fact that that's an interesting one. And that's just the software, that's not any of the other ancillary services and things like that and support or anything like that. But what we're really excited about is just the scope of what people are doing with Connect. And so, if I look at the numbers I'm most excited to report on, last year here, I told people we did six billion minutes of AI-enhanced experiences for end customers. Our customers did those with their end customers, I should say. And this year, that number jumped all the way to 12 billion. So, in just one year, it more than doubled and on a very big number already, that's very impressive. I've got 20,000 to 30,000 customers daily doing more than a billion interactions a day on Connect now. That, to me, is mind-blowing. And we don't have the final numbers in yet, but our number on Cyber Monday this week was by far a new record of usage because we've seen such rapid growth of the business in the last year.

John Furrier

>> It's interesting, a lot of people in the call center sector, only a couple of years ago just came to the religion of, "Wow, we have a lot of data." You guys had this from day one. Share the first principles around how you guys laid out Connect because you guys had done the AI-native work kind of before it was fashionable.

Pasquale DeMaio

>> I think that's true and it's a great point. The thing you bring up that I think is you have to start on a foundation that understands that... First off, you have to have a uniform system that works across all your channels, all the way you build it. So, we didn't buy anything to create Connect. We built everything from the ground up to be able to treat every chat, every voice interaction, every task that may not even have a customer component on it as the same thing. And that means every time we innovate with AI, it touches those things the same way. And so, one of the reasons why that really matters, if you think about an interaction that might move across those channels or have follow-ups, you can then have all your data just make sense. So, it's not just the integration points that can be hard if you're using disjoint systems, but your data doesn't match up, then you don't have the ability to actually understand it. And so, I mean half the job is just being able to see the information you need. So, we've made that super simple with Connect, and so that's why customers can move so quickly.

John Furrier

>> PQ, you guys work backwards from the customer, but you got now so many customers. There's so many use cases. There's a lot of slicing and dicing the UX or CX experience. Take me through your news you just guys... 29 announcements here at re:Invent. That's a lot. I mean, Amazon always has great... But your group only had 29. What are you working backwards from? What's the pattern that you're seeing? Because again, you have a very diverse use case. Big call centers, big companies, small, medium-sized businesses. I mean, it's a service.

Pasquale DeMaio

>> Yep. The thing about it is we wanted to democratize AI for folks. And so, with the launches right now, Connect now actually just is AI. And that's what I tell people. I say, "This is an AI solution." I don't talk about context anymore because we have such a broad set of use cases. Internal to Amazon, some of our largest use cases have always been around tasks with no customer component. Now, it's hard to explain that too. And then, some of our businesses when they first see it because they really want to fix their contact center challenges. So, we'll start with the context and other things, but we move well beyond that. Whether it's for a slice pizza who is a relatively small business compared to say someone like a Barclays and some of the world's largest airlines like Delta, Air France, those folks are incredibly excited to be able to scale up their access to AI. And so, the principle is we work back from are get past checkpoint features, get into actually delivering solutions and then work with those customers as partners to make sure they're really getting the results.

John Furrier

>> What was some of the news? Just break it down the 29 news announcement. Do you categorize them as some technical? Were they more front-end? I mean, just lay out the categorical. It's a lot of news.

Pasquale DeMaio

>> Everything from stuff that might be boring and mundane like performance, but even more interestingly, like people love better performance, but it's not exciting, all the way up to being able to now create fully-agentic experiences and going agentic only if you want to. So, someone can have a completely agentic solution for all of their customer engagement. And also, then using that same AI to power and act as a concierge that actually supports both the end customer and the agent to make a fully, if you want to, a fully all human agents experience. Now, almost nobody's going to do just either one of those things. What they're really going to do is they're going to use some deterministic AI and workflows and plug the whole thing together. But now, with just a few edits inside of Connect, you can actually bring generative AI to help transform the experiences you've already built in Connect with those legacy deterministic flows and also empower your agents. So, I would say the number one thing we've done is like democratize AI across that entire experience and people can move into the agentic era at their own pace now.

John Furrier

>> What's interesting, you mentioned the deterministic piece, that's check the box, nail those workflows. What came out of the frontier agent portion of Matt Garman's keynote was the agents can do longer thinking things and do experiments. So, that's like a non-deterministic view. So, take us through the distinction, how you see the AI playing in that, okay, I locked down all the workflows or some things that are deterministic, but now the new opportunity is going to be how do I leverage some of these black swan type workflows that you might not think about, but like there could be magic in there?

Pasquale DeMaio

>> That's exactly right. And the thing is, in two years, I don't think people will talk about agentic AI at all, really just like people don't say they're an internet company anymore, but in the interim, we're all going to become agentic AI companies, where that agentic AI is acting as a teammate to help improve the performance of the people. And one of the big things, as you point out, is that folks are finding themselves awash and drowning in information and data that they really can't leverage very easily. One of the big launches we had actually let you fully understand your contact center just by actually typing in and asking questions about what's happening there and then doing configuration. So, we made all of Connect AI accessible, so you can actually manage it through AI too. That way every touchpoint inside of Connect, whether you're an end customer, an agent or even a manager is now improved by it. And you won't notice that. You don't need to understand it. We're going to come and the AI is going to meet you where you are. So, you don't even have to, as a user of this, on any one of those levels, you don't have to understand whether it's agentic or not. Just have an awesome experience.

John Furrier

>> Do you hear a lot around the domain-specific intelligence? Because one of the quotes in my interview with Matt Garman, he didn't say this in the keynote, but he said it to me, his quote was, "Generic tokens are useless, unless they know your business." That's from Matt. What he was basically saying is, "You got to tokenize." This is where Nova set up, this is the domain-specific intelligence frame. Take me through your thoughts on this and your vision because this is where these companies have domain-specific things. The workflows are domain-specific, the products, the customers.

Pasquale DeMaio

>> Yes. So, when you think about agentic, I think of it as being an understanding portion, which adds value just by itself. There's reasoning, which then takes it and makes it into something more interesting. Then, there's the ability to act, which then changes the fact that instead of just saying, "Hey, here's some information. I'm actually going to take action on your behalf." And finally, you want to be able to measure and utilize that to keep growing and that memory piece you're talking about. Connect builds all of that into our core system. And then, we also allow people to bring new models in as well. And so, later this year we'll actually fully orchestrate third-party and first-party models in here, that's coming soon. And the thing about that is to your point, people are going to need special case models that do certain things, like one that really understands how you want to do something as mundane as baggage transfers or something as interesting as changing an appointment, but you have to understand how those appointments have to line up with the different people in it. Those types of things are stuff where an agent might be specialized and then it plugs into all the other ones. Connect is starting from there because we've always been where AI rubber meets the road-

John Furrier

>> It's real world too.

Pasquale DeMaio

>> Yeah, it's real world. And so, it's not great to just talk about it. You got to be about it.

John Furrier

>> Yeah. Take me through the real-world example, because I think that's the killer case study. And if you think about the treasure trove of data, just the synthesis. We get the innovation for the human who's on the point of support or point of leveraging the value, there's also you can abstract away the knowledge from the corpus.

Pasquale DeMaio

>> Yep. I can give you a great real-world example with someone like DoorDash. And the thing that they did is they said, "We've got these folks who wake up every morning and they decide they're going to drive for me or they're going to go drive for some competitor. And they have no requirement to do either. They can do whatever they want. And so, if they have a question or they need something fixed and they have to wait on hold, they're not going to sit there and wait for that. They're just going to go pick someone else and go do it there." So, DoorDash said to us, "How can we build this into connect and really accelerate the answers to those questions?" So, we did a ton of work with them to go understand the business and the needs of those folks and we started powering answers to questions, basic updates to things. And then, as soon as there's an issue that can't be solved that way, then we take all of that context and hand it right in a beautiful little package to the people who can help them, a human being. And then after that's all done, again, all that information goes back. We learn from it. It's all recorded in a way that's much more detailed and much more useful than if a human being tried to take notes while they're being told, "Get back to work."

John Furrier

>> Okay, this might be a question that hits the third rail inside the Amazon, but I'll ask it anyway. Matt Garman said, "Three to six months, these agents are part of your team." How are you using some of the internal Nova, Kiro? Because one thing I like about Kiro for instance is that, from a coding standpoint, it learns your environment, it knows your preferences. I'm assuming Connect has a similar vibe where it's like, "Hey, I know the environment. I get smarter."

Pasquale DeMaio

>> That's exactly right. We in fact an integrated Nova speech-to-speech is GA in Connect as of a couple of days ago. So, that is another one of your big announcements and also-

John Furrier

>> That's huge.

Pasquale DeMaio

>> Yeah, it's huge. It's amazing and it sounds great.

John Furrier

>> Bidirectional?

Pasquale DeMaio

>> Yep, and it's awesome. It's a huge improvement and it's only going to get better. And as it gets better, our customers just get it for free. They don't have to do any work to go integrate new stuff. We just keep making it better-

John Furrier

>> Talk about the bidirection because the latency is a huge... You don't want speech to be delayed.

Pasquale DeMaio

>> Yeah. The thing that's interesting about that is honestly, the problems around latency are so often the integration points, And that's why I talked about having the system be built from the ground up to work that way because it's not a lot of times when you call into any of these systems, they're not slow because they can't talk fast enough. It's because they're waiting for those backend systems. So, we pull all that data from all over the enterprise into one place so you can get it. And then, the speech-to-speech makes it sound really nice. It can even responsively react to say... In my demo I showed earlier that actually I say, "Hey, I'm in a rush. Can we hurry this up?" And it starts talking faster. I mean, these are magical things that even if it told a person to do that, they might struggle to respond to, right?

John Furrier

>> Yeah. I have to ask, because there's so much going on, but we need an hour just to unpack Connect. What's the coolest thing that you're working on right now or have announced that you're showing here? Or coolest thing that you launched and-

Pasquale DeMaio

>> The coolest thing I have announced?...

John Furrier

>> here at re:Invent and the coolest thing that you're working on now?

Pasquale DeMaio

>> Yeah. I mean, I think it's a game-changer to be able to have an interactive experience like the one I showed in my demo and I encourage folks to go take a look at it online where it just shows how this thing can be human-like, but not try and trick you into thinking it's a human and take all that information and pass it through the entire experience. So, the whole thing is a thing of delight, as opposed to a thing of pain, the operator-operator experience and and I love that. The thing I'm most excited about is taking all of the capabilities we've built around this to make human beings in the contact center exceptional and making folks in the back office all over the different parts of business who do lots of high volume, sometimes repetitive work. Some of it's not very much fun and eliminating all the boring stuff that's not useful, turning them into human beings who have all the information at their fingertips and even can walk them through every step of what they're doing, so they can actually relax and just achieve the goals as a human being you want to do every day. Almost everyone wants to be great at their jobs, how often is it the technology that's holding them back? Let's get rid of all of that.

John Furrier

>> Well, Connect actually hits the core ethos of the principle of undifferentiated heavy lifting because there's a lot of stuff there that you don't want to do and then that gives the superpower to the human to either think creatively or do some high-value task.

Pasquale DeMaio

>> It's amazing you say that because one of the big things is generative AI, and now agentic, has really stepped in and changed fundamentally the conversations I have with customers in a way that's wonderful, because a lot of times I'd sit down with them and we'd be talking about speeds and feed stuff and I'm like, "That's not going to change your business. We've got to get you into a world where your customers are becoming more loyal to you. Every time someone contacts you, it's not a chance to get rid of them, it's a chance to build a relationship." And to that point, now that's the conversation we're having, I used to say, "Pick three things," now I just say, "Let's fix everything," because we can go so much faster.

John Furrier

>> All right. So, what's your focus now? What are you optimizing for for next year? Obviously, re:Invents going to happen. People Take a vacation, a couple days. Amazon, maybe a day off.

Pasquale DeMaio

>> Yeah, no days off.

John Furrier

>> What's your focus? What are you optimizing for next year? What's your big agenda?

Pasquale DeMaio

>> Time to outcome. The whole thing I want to build on is we've now built these things where we can get these incredible results, I want to have people getting those results in an hour. You can set up Connect to take a call in five minutes right now. Literally, you could have your own contact center. I want to have it, so you can set up and take a call and that call feels magical in less than an hour. That's what I'm challenging my team to go do.

John Furrier

>> Awesome.

Pasquale DeMaio

>> Magic. I want magic.

John Furrier

>> Final question. The shirt. Show the shirt. What's the meaning behind the shirt? That's a clever shirt there. It's one of the most unique on theCUBE.

Pasquale DeMaio

>> Yeah, so my demo was all about my secret love of K-pop. And so, I was showing how I had a problem and I needed to get some stuff fixed and how we could make that a good experience versus a bad one. And at the end of my demo, I come out in a shirt that I had ordered from and had redirected to the Venetian, so I could wear it for my demo. And so, all week I've been wearing different K-pop shirts too. And this is a celebration of one of them. So, it's been a great fun time. If you're not having fun doing your job, I don't know why you do it, but I do love my job and my team loves what we're doing.

John Furrier

>> Yeah, great shirt. Thanks for coming on. Really appreciate it. Again, Connect is really a huge success story. And it's one of those things where companies now, as they come into the AI scale, you see opportunities that you can go after. And I think this is not just an opportunity for Amazon, your customers.

Pasquale DeMaio

>> Oh, yeah. I mean-

John Furrier

>> Once they get there... Scale is a huge advantage.

Pasquale DeMaio

>> Yeah, we see them as a team. We don't see ourselves as a vendor, we see ourselves as a partner with them to achieve that. And one of the things that makes it seamless for them is they never have to worry about the fact that we can go to, essentially, an infinitely-large contact center, an infinitely-large workload, and it just goes with them. They don't have to go buy more capacity or worry about any planning. We just make it happen for them. It's freeing for them. It lets their brain think about the exciting stuff, not the lame stuff that's not productive.

John Furrier

>> PQ, thank you for sharing and congratulations on the continued success, a lot more work to do. Appreciate it.

Pasquale DeMaio

>> Yeah. So, much more work to do, but we're having so much fun doing it and we're getting great results right now.

John Furrier

>> 29 announcements puts stress on the system. They've had thousands of announcements here at re:Invent. theCUBE can barely keep up. Of course, go to siliconagle.com, a lot of great stories there. We're doing our part, bringing you the data here. Thanks for watching.