Customer AI, Not A Call Center

“Deflection is dead” is a bold claim, but after this conversation it’s hard to argue. I’m joined by Pasquale DeMaio, VP of Amazon Connect Customer at AWS, to talk about why the future of customer service isn’t a shinier contact center dashboard. It’s customer AI that improves outcomes for the customer and also for the humans doing the work, from agents to managers to the business leaders responsible for performance.

We dig into the idea of a human-to-AI continuum: some moments should be fully automated because speed is the kindest thing you can do, while other moments demand empathy and a real person. Pasquale explains how Amazon Connect thinks about using AI for routing, agent assist, recommendations, and automatic summaries so agents stop spending their day on wrap-up codes and busywork. The thread running through it all is simple: make the technology recede so the human can focus on the customer, not the tools.

Then we tackle agentic AI and the practical reality of deploying it safely. The takeaway is not “let the agent invent everything,” but combine agentic capabilities with deterministic workflows for critical steps like payments, identity, and security. We also explore trust, security, and reliability as the foundation for any AI-driven customer experience, plus why disconnected vendors create local improvements but fail to deliver an end-to-end customer journey.

If you care about CX strategy, customer engagement, contact center modernization, and AI in customer service, this one will sharpen your thinking. Subscribe, share this with a CX leader who needs to hear it, and leave a review with your biggest question about agentic AI.

Transcript

The CX Inflection Point

SPEAKER_01 0:01

And so now the technologies are going to allow you to do things to help other human beings in ways you could never have imagined two or three years ago. So let's go on that journey right now. Let's get let's not wait. Let's jump on that train to help make everybody better, everyone have a better outcome across the entire thing from the end customer to that agent, that human being who's trying to help them and to the managers every day who are doing their hardest job to make both the end customers and the human beings who are helping them better. And then the business user trying to drive all that. We can do that right now. It's so exciting to be around. And if you wait, unfortunately, it will be too late before you know it. This is this is that moment in time or that inflection point where all the things you dreamed of probably we started this job and have been so hard for so long are finally becoming possible. This is a wicked time to be to be in this business and trying to help people.

SPEAKER_00 0:54

Today on Adventures in Customer Land, I'm speaking with Pasquale de Mayo, who is VP of Amazon Connect at AWS, which is a short title for what I'm sure is a giant remit, which we'll talk about later. But Pasquale, thank you for joining me. Been looking forward to this for some time. And uh maybe you could do a better job introducing yourself than I just did, what you do in Amazon.

SPEAKER_01 1:17

Well, I'm certainly happy to give it a shot, which is to say, um, as you noted, I'm I'm the VP for Amazon Connect, and I had been really focused on customer service and just fundamentally changing what people think about engaging their customers for the last nine years since we shipped here. Uh, it's been quite a journey. I would say we started with uh a very basic product focused on voice with the idea that AI could bring uh a much better experience to that. And ever since then, we've done nothing but build on that that humble foundation to now be a you know a full world-class engagement tool across every channel using really AI at every touch point. And I tend to talk about it as being really connecting a customer AI product now, not a contact center. And and that transformation has been just an incredible, incredibly exciting thing to be part of. So I've really enjoyed it.

From Voice Bots To Agent Assist

SPEAKER_00 2:03

Um I'm trying to decide where to start this conversation because um you have been in the middle of the biggest transformation in the period in the CX space, and you've been sitting squarely in the middle of the technology that powered that transformation. And it seems unfair to just say, hey, what about AI? Because there was so much that went into kind of the maturation of the space, the thinking, and the kind of intended outcomes before we even got to AI. Um, but maybe just take us back to early days, your personal uh, I would say, goal set with your role there. What was it that you were trying to accomplish? If you can kind of condense that into a couple of sentences.

SPEAKER_01 2:51

Yeah, the when I think about what brought me to this product and this idea was really the idea that um having seen a few other amazing things like the internet and then the cloud takeoff, I thought to myself, where is an area where there's just so much opportunity to make interesting things happen with technology and specifically, you know, how could I make people's lives better? And AI was such an exciting technology, even though we were very nascent then, I feel like, compared to where we are today. Uh, and that revolution has just continued since then. So, from my standpoint, it's a place where human beings and technology meet and the outcomes can be disproportionately exciting. And so that's really what gets me excited every morning to get up is that story hasn't changed. Now the parts of the story have changed massively.

SPEAKER_00 3:32

When you had to map out what uh you could potentially do with AI, which you know has changed dramatically very recently and continues to, but back then, what did it look like? And by the way, I want to acknowledge that uh in doing some of our pre-research to this call, I noticed that you are uh have a patent or two under your name uh related to uh, I believe, how AI routes uh consumer inquiries, which is no small thing.

SPEAKER_01 4:04

Yeah, the the uh in the in the first days when it started up, we were very much focused on taking what had been a bit of a dark science where folks would have to hand off to an army of technologists to go implement stuff. And we thought there was an opportunity to democratize access to that kind of capabilities to folks. Now, back then the technology was still pretty nascent, you know, with a lot of NLU-based, you know, intent detection, things like that. And understanding intents is still important, obviously. But we went and said, Hey, can we put that in the hands of our customers, the folks who want to deliver great customer service and let them go quickly? And if the answer was yes, then we felt very comfortable that we'd be able to, uh, in addition, um, be able to continue to grow that over time. And those folks who were who are on that early train with us have, as you said, noted a significant change in the way we can approach that today. Uh, I don't think anybody looks back at those as the good old days, but I certainly look back with them fondly as a as a foundation. Um, the next thing we added from an AI standpoint was after understanding basic intents and allowing people to sort of solve problems, you mentioned that routing challenge. Well, the next thing was how could I use this information to much better understand what the actual customer's goal was? And then I could start thinking about agents differently, how the human agents would then talk to the customer, get them set up with the right information. And then we were able to then also have AI really paying attention to that interaction with the agent and then bringing forward, you know, recommendations, opportunities, offerings for that human agent to improve that experience with the customer as well. And even after the call was wrapped up, then make it sure, make sure that the human agent wasn't spending a lot of time taking notes. I see so much work put onto human beings that aren't things that they are good at or enjoy particularly. I'm not a great note-taker. I doubt many people are hiring their agents based on their note-taking capabilities. And so why are we making human beings do that when AI can do a great job is is sort of an obvious and easy one. But at each step of the way, can you bring AI to make that experience you know meaningfully better? And the answer is definitely become yes.

Building Scrappy Inside AWS

SPEAKER_00 6:02

I'm interested, and we actually don't have to talk about this if this is not within a your comfort zone. But you know, you work for a company with seemingly unlimited resources, um, as especially as compared to some of the even some of the other big players in the CX space who are developing some pretty neat technologies. But that had to have been kind of empowering or uh let me just say delimiting in your ability to think through what could be done. Do you have any sense of what the kind of internal innovation culture in your space has been as compared to what it might look like outside of AWS? Is that even a fair question?

SPEAKER_01 6:43

Well, I I can't tell you about anything from outside of AWS because I that's not my lived experience, but I can I can tell you a little bit about my lived experience inside of AWS. Um, and and and really the reason why I originally came to Amazon was because I was excited about web services. And it seems like quaint to say this now, but at the time it wasn't a foregone conclusion of people that web services were going to be exciting. And so I I thought this is the place that's doing it, and this is the place to be at the center of it all. And one thing I will say is, you know, I can understand the idea that it might feel like we can do anything we want, but we're actually a company that prides ourselves on being very scrappy. And so when we started Connect, it was a very small set of folks, and many of those leaders are still on the team now, more than 10 years later. Uh, in fact, the person who wrote the first line of code for the internal tool, this has got to be now 16, 17 years ago, is still in the team. And we were just all very mission-driven around this. And the one thing I knew we did have was a company that was willing to invest in AI. And so I felt like we would always have access to great AI capabilities. And to me, that was a very, very lucky thing. And, you know, they say in life, sometimes you make your own luck. I it was no accident that I that I was looking for an area where AI could bring to bear. And then I was also thinking to myself, how can I take advantage of those opportunities here to have access to some of the best um, you know, applied scientists and leaders in that thinking? Uh, and and that said, that that has turned out to be very, very lucky. Um, but every day when we think about funding something, we we have a very measured approach to this, and we think very carefully about every single headcount and every person to we bring to bear on a problem. Because for one thing, you know, scappy products I think are usually better. It really forces you to focus and to into picking up important things.

The Human To AI Continuum

SPEAKER_00 8:19

I want to just go to something that that it was a phrase that popped up in some of the materials that your team and I were tossing back and forth. And this is the idea of a gentle continuum where you describe uh the the relationship between uh AI and humans relative to each other in customer service. And I find that to be fascinating because I don't know anybody else who's really thinking of it as a in a kind of continuum format. So I'd like to unpack that a little bit.

SPEAKER_01 8:48

Yeah, the I I I think when you talk about a continuum, obviously you have to think about what's on either end of it. And on one end of the continuum, you've got people who believe that that it should be 100% agentic AI doing everything on the other end of the continuum. There's 100% human and no assistance whatsoever. And my belief has always been that the reality has got to be across that continuum, you want every touch point with the customer to be made better by the way you've decided to handle it and not be looking at it from a technology standpoint and instead be looking at it from a personal how do I interact better standpoint. And what we've found pretty quickly as I as I talked to folks who are in the industry, as I thought about the way our own our own customer service folks at Amazon, and we have a number of customer service organizations here, uh, by some counts, around 150 folks are using Connect internally. And so when I think about that problem, I think to myself, well, what if I said, in the moments where a customer is looking to get a very simple answer, I should automate that in a very fast way and get them that answer right away. If I have a complex challenge, something where human empathy and the and the interaction with a human being is going to help, well, obviously I want to bring a human being to bear that as fast as possible. Before that interaction even starts, I can likely predict a lot of the outcomes. For example, I might know you have a case open, I might see that your plane's been delayed or you've missed your connection. And in those scenarios, why not start with my leg up and say, starting with an automated experience, oh, I see you've missed your flight. I can book you on the next one right now. Does that sound good? And if the person says yes, go ahead and book them on the flight and maybe you bring a person to that, but maybe you don't. It's not something where a human being is probably going to help a lot. Now, if the person says no and they have a different challenge, now you engage with the customer and you pretty quickly can think to yourself, well, I want to bring this to a human being and say, okay, I've got a challenging issue here. There's something where this person is upset, they need some help, they're not feeling comfortable, whatever that answer is, how do I make that better? And I can use AI to help that too, because I can understand every part of the conversation that's happened so far. I can understand the last years of relationship better with that customer. And I can use all of that to inform the engagement, no matter how the interaction is going. And there are going to be times where I'm going to prefer to talk to someone who's, you know, an automated solution. And the answer I would say is in that scenario, if I'm if you're talking to me about my bank statement, I'm not sure I want a person looking over my shoulder at my bank statement. But if I'm talking to you, you know, but there are so many times when obviously human beings bring so much to bear on the conversation. In that scenario, certainly I would love to have a human being there. And most of the times when I talk to a lot of our biggest customers, the ones who are really forward-thinking here, they're not saying, hey, we're looking to reduce agents or have fewer human interactions or have human fewer calls. What they're telling me is I need every interaction to be 10 times better than it was before because my customers have so many choices. The switching costs are going to zero for everything. And there's so many folks out there who are interested in disintermediating me from my customer. How do I stop that from happening? Well, the way to do it is to have that customer dying to interact with you. And every time they interact, you make that situation better. You learn from every experience, you, and then you use that understanding to then improve every every additional experience after that. And so those conversations are the ones that get me really excited.

SPEAKER_00 11:53

I'm interested in the external view. I mean, there are uh I'm I sit on the sidelines of all this stuff, and I get to speak with people like you, but you're actually doing the work with the actual customers here. And uh, I can tell you, even from my perspective, there is a whole spectrum of uh progressiveness and innovation as it relates to customer interaction and customer engagement. You know, uh and I I think it's hard to just say um that it's an antiquated view to uh just try and minimize call time. I mean, that's certainly a factor, but um but then you know, the other end of the spectrum, people are looking at uh um kind of a spectrum of sentiment analyses to really get a handle on what the customers are feeling at the moment and what that could mean for the ongoing customer engagement journey. And I'd love to hear from your perspective, uh, you know, who are the companies? Are there any, and I'm sure there are, who are taking that more progressive view of what customer engagement should look like as opposed to singularly focused on cost center costs or call center costs?

SPEAKER_01 14:19

Yeah, I you know, I would say the majority of large companies are taking a new view to this. I'm not seeing a lot of customers stuck in that old mental model. Now, getting out of the mental model and actually turning that into a reality has been where the challenge has been historically.

SPEAKER_00 14:32

Maybe that's a better question.

Outcomes Over Tech And Revenue

SPEAKER_01 14:34

And you know, I I look no further than Amazon.com and think we are trying to make every interaction better all the time. And we don't think about an individual call as something to shorten. We certainly aren't trying to have artificially long calls. And I always will tell people, you know, faster customer service is usually cheaper and better, but that's not the way to measure it. The way to measure it is understand what is better and then work your way back into how you deliver that as best as possible. Uh, and but the challenge has been that you're you're finding yourselves having to cobble together all the pieces here to get there. You mentioned sentiment analysis, and I see some of these sentiment analysis things where the the sentiment they're telling you is the customer's angry. And I'm like, if the human being who's listening to this call can't tell the customer is angry, that's only because you've made the technology they're using so hard to use that they can't focus on the customer, right? Right. And so now give them some more information. That'll definitely help their performance, right? Just overwhelm them with more stuff, right? More stuff. Obviously, that's a you know, that that that is a that is a path to failure to just keep you know putting more and more stuff in front of that person's face. What I keep saying to my customers and and and they love about it is how do we make the technology recede? How do we get the technology, everything but what you need to go away, and you bring them the experience need to help that human being focus on that customer? And I see that really driving a big difference. And it does it does result in shorter calls, but it's not shorter calls because you know shortest calls just hang up on the person when they call in. It it results in shorter calls because the human being is actually focused on the person on the other end of the call and they can actually address the issue. We also speed up all the integration points so that you can move more quickly as a human being, so you don't get rid of that alt tab. And I'm sure you've been on the phone with someone, and even if you couldn't hear the button presses, you heard the alt tabs. Right. So and so how do you how do you change that experience fundamentally? But the the thing that allows you to take that the next step forward is if one system is driving it from beginning to end, then the beauty of that is the system is aware of every aspect of it. So when I think about customer AI making a difference, I think about when like before the call even comes in, are you predictively reaching out to that human being to tell them, you know, hey, I saw your plane was delayed. Can I do you want me to book in another one? Then when you call in, I can have a good idea what you're talking about. And then I can bring all the information I have forward to whatever is the right way to best solve that problem. And it might be sending them a link to go do something. I mean, people think of that as being cold, and and and that might be the wrong choice if someone's trying to do something that is personal. But there are plenty of times where I'd much rather fill in some information on a on a web page than I would trying to talk it over the phone to a human being or to a or to an assistant, right? But if you'd have everything separable from different vendors, what happens is all these things become these little local maxima where one vendor is promising you that they can take care of all your problems and make them go away. Another vendor is saying, well, we'll do all this with a different set of challenge solves for the same problems, but the two things aren't talking to each other, so you end up having something that sure maybe gets a little bit better every year, but it never actually gets to a full end-end view of how you help a customer. And it kind of reminds me of taking that time machine back to the 90s when you know you'd call up and you'd get a bot, and the bot clearly had no connection to the person on the phone by the time you got to the end of it. And it was very obvious that a bunch of engineers had built this thing out. Sometimes you'd have two different even bot voices, right? You'd have the old. And a lot of folks are pushing people back down that road right now, and they're saying, Well, we'll come and solve your problem, and you know, it'll be a dollar for every customer we deflect. And it's like, deflecting your customers is crazy. Deflection is dead. Like if you try and take customers and get rid of them, I guarantee you your competitors would love to have you deflect them right over to them and help them. And meanwhile, by the way, Google and ChatGPT and a bunch of other folks are building AI services that once again are happy to take on that interaction with your customer and make you not even a excuse me, make you not even a party to that, right? And I'm not saying what they're doing is is a is a bad thing for the end customer. It's only a bad thing for that end customer if you as a company weren't doing the right thing for them in the first place. But I, as a company, I want to make sure that I'm maintaining that brand and helping that customer understand the value of our interactions together, because I think one, I'm uniquely positioned to do it if I really have an end-to-end solution that understands the customer. And two, I think it's the thing, one thing that separates me from my competitors that a customer can feel and see, which is if our interactions are fundamentally different, then I am a fundamentally different company to them.

SPEAKER_00 18:41

I want to zoom way, way out there because um if if I'm I may be putting words in your mouth and I'm gonna I'm gonna rely on you to tell me if I'm right or wrong here, but it seems like your perspective, uh, perspective of your group is less about the technology and much more about the solution, which sounds silly when you put it like that. But um I talked to technology providers all day long who would nod to that uh idea, but in reality, they really got to sell their technology. It you know, it's really the bottom line is about the technology. So um but what I think I'm hearing is you're much more outcome-oriented, uh certainly more uh I would say outwardly so than a lot of people I talk to. It would that be fair? And is that is that too facile to actually mention? I mean, is there something to that? I guess is what I'm asking.

Agentic AI With Deterministic Workflows

SPEAKER_01 19:42

I I think there is, and and and the reason why I say that is a lot of it has to do with the birth of how we even became a product. We really were built in as an internal uh solution to a problem, which was how do you deliver great customer service? And at we literally at Amazon say that we strive to be the Earth's most customer-centric company. Now, I'm you know, we aren't perfect, we learn every day on how to do that, but we've been trying for a long time to deliver incredible outcomes for customers. And when you start from that perspective, my my internal customers weren't saying to me, how do you sell your product or how do you give me more technology? They were saying, How do you make a better outcome for our now at this point shared customers? Because I really look at it as a partnership. And so I went external and particularly with those early customers, it was like we were, it was the partnership was was very, very close because we were trying to learn from how they were doing things differently than Amazon was and other companies were, so we could deliver a great great outcomes for them too. And that that has come through even today. We still, you know, 95% of our features or more are coming directly from what our customers are telling us they need to solve these problems. But that mental model, you know, was was an amazing, it was an amazing affordance to us to have the ability to think that way from day one and to bring that forward every day since has fundamentally shaped the way I think about the problem. And and it's funny, but I've had, you know, I've had a conversation with Andy Jassy when he was CEO and Mac Garmin, now that he's CEO of um AWS, and they've asked me at times, you know, hey, I see this customer's usage is going down. And then they're like, and I'm like, yeah. And they're like, is that bad? You know, they're like, are you really losing the customer? And I said, no, actually, what's happened is they've learned how to use the product better. And as I said, great customer service is often faster. And so they are they are shrinking the amount of time they use it to handle the same amount of contacts, they're handling the same amount of engagements, they're having the same amount of conversations, but they're doing it better. And at that point, I'm like, well, I might, you know, I get a little nervous because I'm like, well, are they gonna tell me I have to go to the nicer? You know, yeah. Is that okay? And and and and I just remember Andy going, oh, okay. And it was no like, hey, you have to go get that revenue back, or what are you gonna do to go make more money off that customer? Um, the answer was, well, sounds good. If the customer's happy, then then that's good. You've done a good job. And and that was kind of a very freeing, you know, set of experiences as those sort of happening over and over again. And you know, and as a as a young leader, um, that was something that was formative also in my in my view of looking at how we should go go go into production, go into driving solutions with customers versus selling, just selling them technology. Right. Yeah.

SPEAKER_00 22:15

I'd be remiss, sorely remiss, if I didn't bring up the topic of agentic AI. It's it's the it's the buzzword du jour. Everybody's got a take on it. Uh half of them are fear-based, half of them are pie in the sky, and somewhere in the middle, there's a small slice of people who I think have a view to the reality of what this is now and what and what it will mean to us shortly. So I'd love to hear how you're approaching the the idea of agentic AI uh and uh agents in specifically where you see it going short term. And then I'd love to have a another conversation related to that about some of the hurdles inherent in deploying agents that uh kind of operate within their own guidelines. And how you manage that.

Trust, Security, And Reliability First

SPEAKER_01 23:04

Yeah, from I I from day one, Agentic AI, it was clear that there was something, you know, pretty, pretty amazing happen. Um, but like almost all transformative technologies, where you think it's going to go is is really where it starts, right? And you think there'll be and there'll be fits and starts along the way. Uh again, I'm blessed to have a team of folks working on agentic AI throughout, you know, different parts of Amazon, and including my own team, who are, you know, who are true believers in the technology and its transformation of capabilities, transformative capabilities. On the other hand, I as a leader spend every day thinking about how to solve problems for my customers. And I get the I get the ability to say, well, how would I want, if I was the person who was going to use this product, how would I want this to change day-to-day the way I operate? And the answer isn't replace everything with agentic AI. Instead, it's allow me to move into a future where agentic AI is doing what it's best at and human beings are doing what they're best at. And in the middle, I want to bring what I'll call deterministic AI and deterministic workflows to bear on every aspect of that interaction where that is a better outcome. And I'll use a very simple example, which is, you know, agentic AI in theory could write code to solve every problem at every instant. And, you know, it could write a new payment processing solution every time someone asks to give you a credit card number. Or you could do something a little more intelligent, which is to say I'd like to process credit cards every time the exact same way. And if I want to make a change to that because of a new security concern or because I want to change providers, then I literally can go make one change and connect, and that can take on every workflow, will then automatically be changed throughout my entire, my entire um company, my entire state for how I process credit cards. That is a huge win for something where a Gentec AI is working with deterministic workflows and human beings, by the way, who may want to also take a credit card the same way. They don't want to sit there and type in a credit card, and you probably just want them to hear your credit card. And so by allowing you to have each one of those steps improved by the right technology, and sometimes the right technology is a person. Um, by having each one of those things enabled that way, it it's fundamentally changed it. Now, there are some things right out the gate where large language models, um, and I'll treat that a little separately from a gentic AI, but large language models were clearly better. And I gave the example of note-taking before. Very few human beings take better notes than a large language model at this point. There are a few that do. I'm not one of them. And I don't like to take notes. I want to, if I'm sitting there, I I still sit um in our contact centers and listen to calls, and the whole time I'm there, I'm not thinking to myself, I can't get wait to get to the note-taking part. I want to get back on another call and help another person, right? As a human being. So I want to do the things that both are more useful as a human and also are more enjoyable and really have that connection. And so many of these human agents, that's the same mental model they have. Uh, so we bring to bear and take out some of the annoying things like wrap-up codes where they're just grabbing the first one on the list and things like that. Just make that go away. That's not helpful to anybody. And but on the other hand, again, I can take each step of this thing and say, well, if the person has a very vague meandering explanation of what they're trying to do, a gentic AI can probably listen to that and actually just say, hey, here are the three things. That person said, you know, they want to change their address because they need to update when they get when they get their shipment of the next the next shipment they're getting, and they also want to change their payment thing. We can break that all down and then take each step of that and turn it into something to solve for that customer and then help them. And when again, when it's uh when that's something that's the you know, changing taking a credit card payment, it's probably something you want to make very formulaic and have repeated over and over again with that deterministic workflow. When it comes to talking through the options of something, that might be a thing where a human being may want to be a better person for it. And then if it's something as simple as just saying, hey, I need to have a new credit card sent out to me, there's no reason not to just automate that fully at the beginning. So each of these steps, we want to make sure that we're doing the right thing for the end customer and really bringing the human to the fore in all the cases. Uh that's so that's that's my mental model about how I think about agentic. If I think about where agentic is going, I think I think a lot of these, a lot of the things that have been happening now are kind of turning agentic back into a little bit more of the deterministic aspects that are really powerful. And I see a lot of things like tools from MCP and different things like A to A, where you're actually hardening these solutions. And I think that's just going to be incredibly important for customers. And so when we build Connect, we make it easy to bring in MCP technologies and A to A and also these workflows as AI tools, meaning the, you know, the sort of formal use of that in the agentic world. I'm probably a little bit in the technology. I am, I am at some level uh still an engineer. And so I do like to talk about that stuff. But when we think about that, we're saying, well, how do we let people bring those things forward and make agentic safe? Not just by saying, you know, trying to use guardrails, which I think are interesting, but also by saying, you know, the actual tools you have can enforce a lot of a lot of that safety. The uh, the the the workflows, the terministic workflows can enforce a lot of that safety in a really powerful way while still getting a lot of the benefits of the agentic capabilities. It has that natural communication and can help unpack a complex problem and then hand it off to the right tools, the right people to solve it at that point.

SPEAKER_00 27:58

There there have been in the recent past, and I I uh uh quite a number of news reports and I will tell you lots and lots of anecdotal conversations about the uh potential for, I'm just gonna call it broken trust as agentic AI uh flows start to become a little more prevalent in in everyday business life and you know, life life. And um I've only spoken to a couple of companies that seem to be thinking about how do you manage for that trust. I mean, look, it was only what whatever it was three years ago where the public became aware of AI and its and its consumer potential, but there were there were uh enormous trust barriers at first. I think we've overcome them largely by now. But uh it's a whole different world if you're thinking B2B, for instance, when a company's got its brand reputation on the line, um you know, there is no room for uh for error, for trust error. And I'd love to hear how you and AWS Connect are thinking about um if it's a trust preservation or if it's uh you know trust repair uh before it happens, or you know, is that even an issue? Or is it just uh, you know, kind of the news hype cycle trying to find something to be interested in?

Beyond Dashboards Toward Conversation

SPEAKER_01 29:31

I I think that trust is a core value of any interaction. So from that standpoint, I don't think that's going anywhere. And and I wouldn't ever want to have an interaction with anything, a human being, uh, a bot, uh a webpage, if I didn't feel like I could trust it. And and I think a lot about that day to day in my in my personal life, because if I someone sends me a link to a webpage, I think a lot about it before I click on it, right? And so I don't think that's going anywhere. Uh I think that it's our jobs as as technologists, but more importantly as folks trying to solve a card human problem to make it so that trust is something that's built from the ground up. And so we put security as our number one priority at all of AWS and certainly in Connect and reliability comes right after that. And then we think about innovation on top of that. If you have to build from the you have to build from the bottom with security, and then you have to give people the tools to make security easy and to make trust easy, because if you make it hard to do it, folks will still, you know, you can give the most trustworthy solution ever, but if it's too hard to use, no one will ever use it, right? And I'm sure you've seen situations like that. Uh uh, you know, I do remember the days when you could put your keys on your visor and your car, too, right? I mean, times have changed, unfortunately. It's and it's a it's a race for us to continually allow our customers to be able to avoid fraud and to help customers who they are engaging with, the real customers they're engaging with, be able to move forward quickly and solve their problems quickly and with the right outcomes. That race is something that is, again, the reason why I come to work every day. It's so exciting to be here using the forefront of technology. I will often say to people though, if you want to use Connect and be 100% agentic from day one, that's 100% possible right now. If you want to be 100% human and stay that way forever, that's also fully possible. The beauty of Connect is you can have anywhere along that continuum you want. And you can bring in agentic AI at each step of the way, you can bring determinist workflows, and of course you can bring in human beings and have amazing outcomes. And each step of the way, all I'm thinking about is how does this technology stop being technology and start being a solution to the problem that's getting out of the way of whatever you're trying to do and then just giving you the right answers to help you move forward. And I think it'll be a while before that's a fully agentic capability for most for most companies and most end customers' preference. Um, and I think, and and maybe never, and I think that's okay. I think the right answer is let's use the right the right technologies to make the technology go away and it just become a solution to the problem. Um, and I think AI can touch every single aspect of that to help that, whether it's helping improve uh a self-service interaction or whether it's helping a human being do a better job. I think those things are fully available today, which I think is just awesome.

SPEAKER_00 32:02

It is it is an amazing moment. So speaking of amazing moments, um, and this is where the conversation could really devolve into crazy places. Um AI has uh continues to transform every aspect of work, work life, business uh in ways that weekly become uh you know what was unthinkable last week is now possible this week. And uh it's you know, we're all aware of the shift it's creating in process. What I'm really interested in, I'd love to hear your perspective on this, is uh the uh how it's creating a shift in how we think about process from within a company like like AWS, uh like like Amazon. Because um when I talk to people who even companies who are selling and developing AI solutions, um they uh have approached, I'm making a broad statement, and you can certainly challenge this. A lot of them are great companies, really bright people, have approached uh the use of AI uh with a sense of like I can make this thing incrementally better, maybe even exponentially better, but it's still the same thing. And yet uh I think companies like Amazon have a unique kind of position in the on the pyramid of the pyramid of world technology, if you will, to tell the trends, to the opportunities, to the problems, even that the rest of the world probably never will, just by virtue of who Amazon is. So with that in mind, uh that kind of perspective, if you looked out, no, I'm gonna change the question. I don't want to know what it looks like two years uh from here in your perspective. What I'd really love to know is how you're thinking about AI. Because I because I'll bet you it's different from what other technology providers in this space are thinking.

The Wake Up Call For CX

SPEAKER_01 34:10

Yeah, the um from my standpoint, the the things I'm focused on right now is is really allowing users to not have to be technology experts to get results. And so fundamentally, I want a business user to be able to change the way their interactions with their customers happen. And so that might be what you call consider a classic contact center experience, but it might be something that feels nothing like that. It might be something where I'm instead I'm redesigning my workflows and my understanding of how I want to process business tasks. And so we're actually turning Connect into a tool where you can come in and have a back and forth conversation with your quote-unquote contact center. As I said, we haven't called Connect a contact center for two or three years at this point. Um, I consider it just customer AI. And that customer AI is whole whole reason for existing is to allow the person running the business to get the outcomes for their customers that everybody wants. And so, how do I make that easier? And that becomes an interactive conversation, not a set of dashboards. Now, I still walk into some customers and they'll point to a dashboard that I know was probably built in the late 90s and say, this is what I need to do my job. And I'm like, well, I we can figure out how to get you that dashboard that looks exactly the same. But I bet by the end of this conversation, we're gonna be having a very different conversation. And how do I get that same dashboard? And even I would say some of the folks who've been the least excited about moving into the future are thinking about the fact that they can finally stop saying no or or being told no and start saying yes to being able to fundamentally change the way they engage. And and the flexibility at which we can present an experience around this for a customer, that's a little bigger than I like to say, but like if I want to say, hey, look at this dashboard, I can have that be part of any experience I want to, but I don't want to start with a dashboard and say you figure it out. I want to say, hey, here's what's happening in your contact center, here's what's happening with your op on engagement, here's what's happening with your customer interactions across every step of the way. So, you know, the idea of what you used to call a contact center is a teeny little bit of that. And instead, it's much more of a customer experience. And we're gonna go help you drive that. And by the way, we're gonna point to you when you come in in the morning and say, here are the four things that were interesting yesterday. And they're not the same four things that were interesting at the end of last week. And maybe it's a holiday, maybe it's very easy to explain why, but maybe it's not. Maybe a snowstorm last week actually turned out to have a delay in some supply chain thing, and right now you're now feeling the effects only multiple days later, and now you've got that, and now you've got to handle that. And now you can have a conversation with another part of the business to say, hey, here's what I'm seeing. And it's not just a list of the same old average handle time and first call resolution, which are fine things to measure at some level, but but not really what you want to be doing is your first order set of engagement. But instead, here's how customers are fundamentally interacting with me different right now than they were last week and a year ago. And believe me, in the next year, you're going to see a massive shift in that. And you know, right now, people are either being trained by the good companies that they can lean in and get great outcomes very quickly, or they're being trained that you want to avoid those experiences altogether. I want to make it so that those experiences, again, just feel like one single experience to the end customer, one continuum of goodness that travels through the optimal way to say, hey, let's do self-service for things where that's the right thing. And then not every human being is going to be enabled. But we started, I mean, we started the process by helping agents get better a lot of times, by giving them feedback and giving them the tools without having to alt-tab. We said, let's bring that to the end customer and deliver those same AI for self-service. And we had a revolution there. And now that revolution is moving to the business leader inside of the company, trying to get the outcomes they want there. And sometimes I say we kind of worked, you know, big to small, helping agents has very obvious great outcome, happy human beings, the human agents. And then enabling self-service has another set of very powerful outcomes. And now the next step is say, okay, now there's a lot fewer of these business users by by nature, typically, but they can have a disproportionate lever in how you think about every aspect of the way you engage with your customer. And empowering them is going to be, I think, uh the next round of really amazing, really amazing outcomes. Uh, each one of these areas still has plenty of room for us to innovate. I'm not, we're not slowing down any of them, but this is just a whole, like a whole nother door has been opened that you can now go explore, which is just so exciting.

SPEAKER_00 38:18

So you've just been given a megaphone to speak to the entire CX world and tell them something they don't know or need to be aware of, or watch out for or prepare themselves for. I'm I'm projecting because this is not that megaphone. We have a smaller, but much more directed megaphone. Um, but you have a chancer to speak to CX professionals uh and tell them something that they probably don't know that you do. What would that be?

SPEAKER_01 38:47

Well, I would I would say first off, I think your megaphone is extremely focused on a bunch of folks who really care deeply about customers. And that that's I think an awesome opportunity. The conversation I would want to have with them is say, like, you probably got into this business because you love human beings. That's probably a good reason why you started here. And and that was a good instinct in my mind. That's a big reason why I'm here too. And so now the technologies are going to allow you to do things to help other human beings in ways you could never have imagined two or three years ago. So let's go on that journey right now. Let's get let's not wait. Let's jump on that train to help make everybody better, everyone have a better outcome across the entire thing from the end customer to that agent, that human being who's trying to help them, and to the managers every day who are doing their hardest job to make both the end customers and the human beings who are helping them better. And then the business user trying to drive all that. We can do that right now. It's so exciting to be around. And if you wait, unfortunately, it will be too late before you know it. This is this is that moment in time or that inflection point where all the things you dreamed of were probably when you started this job and have been so hard for so long are finally becoming possible. This is a wicked time to be to be in this business and trying to help people.

SPEAKER_00 39:54

Completely, completely agree. Well, Pasquale, I can't thank you enough for the perspective, for uh the guidance and for your time here. Um, what's certainly worth the wait trying to juggle schedules, putting this together. Again, this is Pasquale DeMaio. He's VP of Amazon Connect at AWS. Um, I, for one, am going to work really hard with your team to try and schedule another one of these because I have this is not a visual medium, but I have four pages of notes, and we got through exactly one half of one of those pages. So there's so much more to talk about. And I think you've prompted another several pages worth. So with that, thank you very, very much. Really, really appreciate that.

SPEAKER_01 40:34

You too. Thank you so much for the time. It's been a great conversation.