What’s next in contact centers with Amazon Connect and generative AI

Ok, great. Well, they, that sounds good. I'm loud. Um usually I'm loud about it on my own. This is even better. Hi. Um I'm Pascal de Mayo. I'm the vice president for Amazon Connect here at AWS.

I want to start by thanking so many of you for coming out here. I know some of you probably came from the Venetian. It's a bit of a bit of a walk. Uh it's great to have you here. I'm really excited to talk about what we're gonna do today. I also have some tremendous um, customers here from Capital One to talk about some of their journey.

I'm going to dive in pretty quickly here to talk about what's happening in the world of contact centers and generative AI and what we're doing in Amazon Connect to enable you to achieve incredible results for your customers.

I'll start where I always like to, which is to step back and think about the challenges folks face. And well, I've seen amazing things that customers have done to deliver customer service. Um the thing that I do find regularly is I still see so much of the same challenges that have been there where folks are still hamstrung by the technology they have that prevents them from delivering the level of customer service. they want to not, not just to say, be ready for answering questions, but to proactively engage and change that experience.

Instead, I see agents who are frustrated. I see supervisors who are challenged to know what their agents are doing and why they're frustrated. And then I see customers who are seeing the negative results of that. And I will ask you, well, I believe you've likely done some incredible stuff even in your context, centers today. Are you truly satisfied? And do you really feel like you're where you need to be moving forward or do you feel like you have a lot more you could be doing and should be doing and your customers are going to expect and, and, and really want you to do?

At Amazon, we believe deeply that customer service is at the heart of almost everything we do. We strive to be the earth's most customer centric company. I've heard people joke that we're a cu a customer service company that happens to sell products. Um I think I don't mind that honestly, I think it's a wonderful way to think about the problem at our heart though of that customer obsession is, is driving innovation to enable those experiences for our customers.

Will that be delivering packages over, you know, in two days and then eventually same day deliveries for many locations. Whether that be Alexa bringing you the power of AI to have wonderful experiences with your family in your home. Anybody out there have an Alexa device listening to some music on it, asking it uh at a time or maybe getting your news. Um the uh I love mine. I, I use it almost every day. Uh just walk out where you can literally go into a store, pick up what you want and leave without having to go to a checkout line or wait, it just automatically deducts it from you. You don't even have to make sure you count up what you have. Um we'll take care of it all for you. It, it's wonderful.

Um you've probably seen some of those in the airports and other places, stadiums, it's amazingly fast to get something in a stadium that way instead of waiting in those horrible lines and finally AWS which really revolutionized the way people think about technology, removing so much undifferentiated, heavy lifting where you would go stack and rack boxes and you were spending all this time and getting no advantage in from over your competitors and driving no innovation for your customers because what you were doing was so basic to what you need to do instead of taking that off your hands, so you can spend your energy, your effort and your money delivering results for your business and most importantly for your customers, we take the same approach in the contact center.

And that's why we developed Amazon Connect. Amazon Connect started as an internal tool years ago. Um in fact, 16 years ago, and the person who wrote that very first line of code is still a engineering leader in the Amazon Connect team. We're incredibly passionate about what we do around this. We think this technology really changes people's lives, changes the operators, it changes the and it changes the supervisor, changes, the agents and all of that comes together to change the lives of our customers.

Um we saw what we had done, there was really exciting to cut what we had built. There was really exciting to have customers of AWS as well. Um and we said, we asked them, do you want this, would this be something you'd be interested in almost universally? The answer was yes, we'd love to be able to take that same technology and apply it. The things that you're that we were using to service to service.

Um uh more than 10 million contacts, interactions a day uh today. And the thing that we use for more than 100 100,000 agents was something that many other big companies, even if they didn't have a contact center as large as that were very interested in. And so we decided to bring that to market and very quickly, it became one of the fastest growing, um one of the fastest growing services in AWS history.

And the result of that has been a large range of customers on here. We can't even come close to fitting all of our customer references up here. But you can see that there's a broad range of who's using this. It's not one type of customer or one industry or one vertical. Certainly places where there's very large contact centers like financial institutions, where people need value security really heavily and care very deeply about measuring and outcomes and are using it. But they also have small start ups, people like sliced pizza or insomnia cookies using it. And you couldn't think of two more different businesses potentially, but they both care very much about both excellence and accuracy in customer service.

And as you can see now, we are absolutely delivering a tremendous volume of cons service across the entire globe. What makes Amazon Connect different? What are the differentiators of Amazon Connect? Why would I pick this over something else?

The very first thing about it is it's truly a cloud contact center and it takes true advantage of what the cloud can offer, whether you have one agent or 100,000 agents, we can scale up and down for that. So if you think about COVID, you know, many people in the travel industry had a massive rush of people canceling flights, canceling hotels, they had to double triple, quadruple their capacity overnight, we simply let them use Connect to do so it went up with them.

Then two or three weeks later, they had almost no traffic and their contact center went down to almost zero calls. Same time we, they, we, they just lowered it down to zero. They didn't have to do anything. They just, the calls just stopped coming in. They weren't paying for a contact center that they had a, had to plan for all this capacity for. They didn't have their customers listening to busy tones because they couldn't have scaled up in the first place that just isn't possible with legacy contact centers.

And truly, we are a leader in the ability to do this. No one else. Uh I haven't seen anyone scale to what we can do here. Uh and we, and because our pay as you go, models match to it, the business drivers, your business drivers are matched to it as well. And well, certainly, I don't expect us to have another event like COVID-19. And certainly, I hope we don't, you're almost certainly going to have spikes in different changes and you probably have cyclical business cycles like we do.

When we do Prime Day, we have to double, you know, we have double our traffic in the middle of summer. Um and then also Christmas and things like that. Uh it, it's really important to be able to be elastic to your needs, not, and it takes so much heavy lifting after all that planning that you would have to do and to to get upfront on capacity in telephony trunks, we take all of that away and make it. So you really are focused on delivering great customer service.

The second thing we want to do is to really accelerate your ability to innovate and we'll talk a lot about this today. It's going to be a main focus of my talk. But the first step to that is giving you true self-service of being able to configure your contact center.

Um we offer a very powerful and broad reaching UI to do that. We also offer APIs if you want them, most customers do, but many don't. And in fact, you can drive all of the changes and Connect typically without any interaction with Amazon, whether that be turning on something like Conversational Analytics or maybe looking at changing the way you want to look at your metrics and all these things, you don't have to come ask us for it. You just do what you want, we have lots of help for you, but it's all in your control.

One of the things people really love to do and configure themselves, which historically has been really hard is to deliver conversational experiences, natural dynamic experience for a customer. The example I always give is um instead of saying press one for sales, press two for support or now we see more commonly how can I help you, which is better? But really when you say, how can I help you. How often does the person say you can talk, let me talk to a person, operator, operator, that's how you can help me, right?

The answer to that question can be very different if you really are thoughtful on how you design these experiences. And one example I love to give is if you call, if you've been at the airport and your flight's been canceled a good chance, you've tried to call to figure out what's going on and get that fixed. And so when you call up and the thing says, how can I help you? You're almost certainly going to say agent, agent, you're, you're freaking out, right? You want to get that immediately solved and you don't want to and you want to be sure it happened.

Well, what if, instead of saying how can I help you? It said I see your flight was canceled. Would you like me to book you on the next one? Suddenly the stress of being on the phone has been evaporated and now you are actually feeling very confident that this thing is going to fix your problem because it's so aware of what's happening while the airline certainly knows the flight was canceled. It certainly knows what, what you were on. There's no reason why we shouldn't be delivering these types of experiences and i bet you have a lot of data about your customers that you could be bringing those dynamic experiences as well.

Um that is one example of AI is an incredibly powerful thing, but it also is something that's bandied about as a, as um you know, as a, a lot of jargon I think, and a lot of promises that aren't often delivered in the context in our world. And in Connect specifically, we believe that we have to make every use of AI count. We want AI to be something that has real world results that we will help you measure and it will be easy to measure and you can see them and demonstrate them to yourself and demonstrate them to your teams and your managers to show why you're making this investment and how you're benefiting from them.

I'll talk a lot more about AI throughout this entire conversation and has anyone here heard of generative AI? Ok

A few of you. So we'll definitely cover some of that today too. The thing that you'll keep seeing me coming back to is I'm not talking about like, oh, I have a scratch, let's rub some AI on it. It's really, hey, you know, when you take that and you really point it at real outcomes for customers, it can be game changing. And we will talk a lot about that today.

And the last thing about it is in the contact center, people don't buy a contact center because it connects to point in points. You all probably have a cell phone in your pocket. Anyone here have two cell phones in their pocket right now? So are you having a trouble connecting to someone? I don't think that's the problem. The problem is what do you do to make that incredibly efficient? How do you improve that over time? So that what you were doing yesterday is not nearly as good. What you're doing tomorrow and what you're doing in, in two years from now will be transformed from where you were today.

And the big way you do that is through having great data and great tools for testing and making changes based on real world outcomes. And so that's a big part of what we do in Connect is give you full access to your data. Uh so you can make those choices and I'll have a couple of announcements today about what we're doing there to make that even better for you.

So in Amazon Connect, as I mentioned, we launched back in 2017, we were relatively sparse offering. We did, we focused very much on voice. That was the challenge our customers were telling us they cared most about. And believe me, we build Connect based on your feedback. 95% of these decisions are made based on customer requests for us and what we decide to ship.

You can see how quickly we've accelerated our, our delivery here and it's moved us. Well, we think we still are disruptive offering. We do, do things no one else can do. We also, we also now have moved into a space where we're a leader in the industry. We have a broad offering that offers the capabilities you need to do from end to end, deliver a contact center. We built out the tools to make that easy. So it's not a lego kit, but instead something that is seamless to use.

And these investments have been um have been very high on our side of us. I'm incredibly proud of the team we built and Connect and how hard we work to make sure we're delivering these things on your behalf so that you can move quickly. And that is the main point of this. The goal of everything we do here is we want to be able to deliver a great customer service, but we know the court of doing that well, is the ability for you to innovate quickly.

You need to be able to get things into production with a click of a button. You need to be able to test those things and see the results the day you launch them, you need to be able to turn them off, perhaps, hopefully not, but sometimes things don't work the way you want them. You want to be able to know what's happening and then do that. And if you see something that worked well here, let's replicate it three more places and see if it works well, there, we want to make that incredibly easy for you to do.

And so if you look at some of the launches we've had here, whether that be something like tasks where you can then track all the stuff that happens throughout the process of handling, handling all the things that go on in your contact center, even the back office, things that are outside your contact center, but still impact the business. That's a great example where we're bringing the technology we have to bear. So you can really innovate broadly across the context that are not just in one little area.

Another great example would be our case management tools which now allow you to track some issues from end to end. Let me dive in a little bit though when we talk about why you can actually move quickly because anyone will say you can move quickly. But where does the rubber hit the road for that?

And so the very first place is the one i mentioned, self-service. Again, you have access to control, control, almost every aspect of Connect on your own uh without having to have any help from Amazon to do so. Of course, we're happy to help you and we have great partners who can help you too, but you really own this and you can get set up in five minutes, take that first phone call and that same contact center you set up in five minutes can scale to 50,000 agents, 100,000 agents. It's not a different concept. It's not upgrade, it's not new hardware, it's not having to go change a sku you have access to that capability right away and you can configure and adapt to your customer needs over time. 100% as you need to.

Nearly all of this is nearly everything also is low code and most of it's no code as well. So it's not something where you have to have a legion of developers or a legion of operators to do it. It's really simple. Most things are a few clicks. For example, if you want to turn on conversational analytics, just a few clicks to turn that on and you can start actually listening to every call in your contact center and making decisions and learning about what your customers are doing on every single contact.

The next one is omni channel. Everyone will tell you they have omni channel. Um I guess I'm here telling you the same thing, but let me tell you a little bit more about why, why you can call it, why we have omni channel? Why? What that means to us, what it means to us is that the whole system was built from the ground up as a single solution. So we have one engine for routing, we have one engine experience, we have one analytics engine that's tracking all things across all the different ways you interact in the contact center and we have one flow engine. I'll talk about that in a minute, but that flow engine is an incredibly powerful process and the process engine that happens to also be great at communicating with people.

And so that allows you to automate customer experiences, bring natural language into that experience. It also allows you to automate agent experiences and it allows you to automate tasks that you can move the agent altogether with from so that they can't make mistakes or that you can help improve your reliability or speed them up and get those things that don't really contribute to what agents are great at, which is in engaging with customers.

The next thing I want to talk about is the data and AI. Because so much of this is built to Connect, you don't have to go do a bunch of work to integrate this in. It just is there from the beginning. So you're going to see so much AI in this thing, you're gonna get bored sick of AI. Um but it, it really is something where we bring it into the system, make it seamless to use and then you see benefits from it. I'll talk about how we do this for conversational analytics. I'll talk about how we do it for forecasting and ceding of agents. I'll talk about how we do it for um caller authentication. There's so many different things you can do with AI here that can have meaningful results. And again, because they're built in and there's no cost for you to do them. You can test them and make sure they really work for you. And if you decide they don't, you don't have to keep using them. It's a wonderful place to be in power and control of your destiny here and be able to measure the results and make decisions based on data and outcomes, not based on guesses and promises.

The next one I'll say is that we are one vendor. AWS is a large vendor and we offer a ton of different services and ways you can build, we offer a tremendous place to store your data. We offer a tremendous place for you to build your apps and within Connect, we offer all the stuff in a a really tightly bound solution for you that is seamless to operate from end to end.

And this, this solution, as I mentioned, contains all of these things that will empower you to move from whether that's the front end piece where you want to manage telephony. We've got that all covered for you all the way to the back end where we are exposing your data to you in any way you want to.

And finally flow. As I said, I would talk a little bit about these flows are a core part of Connect, you know, um value proposition for you. You're in control of a very powerful drag and drop technology to create an experience for a chat bot or create experience for a voice voice bot or to, or to automate tasks. And now with recent launches, you can actually do build step by step guides in the same way for your agents.

What's a step by step guide? Well, the idea is that most of your agents are doing a lot of repetitive stuff, but they're probably having a lot of pain doing it. They're moving between different applications. You probably hips ones, you know, you've watched them work. It's amazing to blindfold them and they could all have at the exact right amount of speed at which point you would see them actually, at which point you would actually, you know, see them do things where they know how long a page takes to load every time and they won't start typing until it actually comes live. Right. That's amazing. It's pretty cool. I don't think it's what you want to be paying your agents to get good at. You want them having empathy with customers, you want them turning customers sentiment around. You want them solving problems thinking through things that are complex, not fighting technology.

Well, these step by step guides let agents be effective on day one because they offer the ability to show that agent what to do and then they can actually integrate directly to your back end system. So if you want to change an address, if you want to be able to, if you want to be able to take a credit card, for example, um these things can all be integrated there.

One of the things that's great about this is as I mentioned, these guides are the same flow engine. It's literally one new block to create a visualization. And now, so if you have an expert in chat bots, then you have an expert in voice bots. And if you have an expert in voice bots, you have an expert in task automation.

If you have an expert in task automation, you have an expert in guides. So these people are fungible across your enterprise. If you want to have separate teams doing it, they can talk together and learn, you can also build modules within this. So let's just take that credit card example, you probably have an integration with a provider there, right? That integration is very important to you that it's secure. If you find out that there's a problem there in a security issue, you want to change it and you have to change it everywhere. Well, with a module, if you make that change, once you submit it, that's going to change across your entire, your entire domain of every aspect you have this, whether that is inside of agent guide or whether that's in the IVR or whether that's in the chat bot.

So this is an incredible accelerant to your ability to provide these things and to eliminate human error to automate the things that you want to be there. And with that A/B testing to improve every day at everything you're doing. So, a real key aspect of the product.

Now, all of this, as I mentioned has helped deliver that set of customers that you saw earlier on which um is a, is what we're really, really excited about when every day we wake up, we think about enabling people like you to have great contact centers. Uh it's been very nice for us to see recognition for that I will say, and one of the recognition that we've had wonderful is that we were anointed a leader in the Forrester Wave, which is great, a great thing for us to um great thing for us to have and hopefully gives you confidence that what we're saying up here is something that you can deliver and use.

Um similarly, uh Gartner has moved us into leader quadrant. Um this is only six years after launching the very first, you know, the the product and um and after only a couple of years, few years of being uh actually taking part in these evaluations. So um i, i don't, i don't want to say that we're, we're proud because we're always humbled and you know, we have so much more to do, but it's incredibly, it's an incredibly wonderful feeling to know that we have delivered something that, that not just our customers are telling us works. but also that the, the analysts and experts in the industry are telling us works well too.

At this point though, I think it's a good time to actually hand this off to some folks who hopefully will be even more believable in that statement because they're using the product to deliver amazing outcomes for their customers. And that's some folks um, here, Chris and Ram from Capital One, if you guys don't mind coming up, they've been here with us from um really nearly day one and they have been innovators on top of Connect and they have also benefited from the innovations of Connect over this time to change what they're doing on a continual basis. And so with that, I'll hand off to you two to tell a little bit about your story. Thank you so much.

Thank you, Pa Rewan. Uh my name is Chris Short. I'm a senior product. Rum will be back later. He's going to do more and then just I promise we'll get to hear from Rum. Um I'm a senior product director at Capital One. Brahms, my tech counterpart, we're really excited to be here to talk to you about our journey with Amazon Connect over the last few years. Our focus at Capital One is in credit card servicing. So we're really focused on solving customer needs, that plan mix of AI and humans, assistive technologies, specifically, we own voice and chat platforms and infrastructure. That's largely powered by Amazon Connect.

So let's let's jump in. I'm looking forward to sharing some insights on how we're partnering with Connect. So first, like just real briefly, some context on Capital One founder l company started with the mission of making financial products accessible and simple for customers, you know, variety of financial services, third largest credit card issuer in the US top 10 bank by deposits. A lot of auto loans. My wife suggested able to point around like volume of commercials that we run. I'm sure you've also, but there's truth to that acquisitions business. But when you get past, you know, Taylor Swift and Charles Barkley, you're handed to the real talent Ram and myself who help with servicing your accounts. We care for over 100 million customers at all. We have 50,000 associates and the key unifying thread as we've added these products over the years has been differentiated, differentiated investment. We really think a lot around how we build experiences for the customers to pull this all together.

Um and as you can imagine, there's a real diverse platform and capability ecosystem required there that we view as like a mixed in-house partnerships with Amazon. Um so we're kind of always thinking about how we can make that better on behalf of the customer. Perhaps no better example of our commitment to tech investment has been our cloud journey to date capital. One's managed to build really a technology company in a highly regulated financial services industry. We began our journey with Amazon almost 10 years ago. At this point, recognizing the power that cloud could provide with respect to scale within a couple of years, we quickly declared we're all in on cloud and worked to fully exit our data centers. And along the way that modernization journey, built a lot of new tools and capabilities which we've since have started to externalize for others through our capital and software offering.

I highlight all this just the the put some key themes out there which really a nonstop building culture, constantly hunting for new value and really the balance and need within the financial services industry to build well managed and build well managed and take care of risk, but also finding ways to innovate on behalf of the customer. So really our focus is how do we build it right and nail the basics to create space for some of the shinier objects that we all get excited about with. Connect.

Specifically, we embarked when they're in disruptor mode, I guess about five years ago. But the goal of connecting our end experiences and creating a flexible network infrastructure, you know, our primary stakeholder set customers, agents, supervisors, administrators, consistent with what Pasquale showed before and what he hears from customers from a customer perspective, disjointed experiences, you know, who's tried something in an app and then called and have to restart and get transferred to the wrong place. Like we've all experienced some element of that. Unfortunately, we really want to get to a point where across the scale of a million calls and chats per day, we at least need some consistency for how we're handling these experiences.

From an agent perspective, they bear the brunt of that experience. I'm certain no one here has ever given an agent a hard time for the frustration we cause in our systems. But from an agent perspective, a lot of juggling of apps and windows and you know, difficult to onboard or upskill new agents and really just taking away the time from servicing the customer. So it really had a goal of simplifying the agent experience for the thousands of agents that Capital One has across the world.

From a supervisor or operational supervisor perspective, really long lead time for changes, honestly, like limited levers to even optimize the network in real time, a lot of manual reporting. So we really need to get to a point where we had more of the real-time monitoring and ability to scale in near real time.

Finally, from an administrator and technical perspective, it was a complex, you know, more complex architecture in the sense of a variety of different DCs partners tools and just the tech that required to run that day to day, like cut it away at the time spent innovating or thinking about the customer. And I'd add, like even not explicitly mentioned here, but organizationally, there's rarely perfect alignment on the relative order of precedents of these personas which at a larger financial services firm becomes a huge task tax on the team.

So what initial Connect migration enable us to do was unify the tech platform but also start to transform the operating model of how these different personas interact with each other or how we think and build with respect to the customer. So the initial Connect migration, we successfully migrated our contact center from to cloud on Connect, you know, we moved to cloud, we enabled all the network configurations, real time data, millions of cost savings, everyone who built that had a good performance review that year for sure, from a technology perspective. So we laid out the technical foundation which was awesome. But I think like the thing we we even underestimated at the time was the repeatability of value in terms of ok, our foundation is safe. We're not reactive in how we think about solving problems. We now have the space to for deep inquiry and thinking big around what are the step changes or transform transformational things we could do on behalf of the customer.

So since 2020 range, we've kind of been in this mode of rapid innovation which has been a mix of Connect, releasing new features for certain but because Connect created the stability with our infrastructure space for our tech to recoup our tech teams or product cycles to think about building new things on behalf of the customers. It's been a really amazing time I mentioned, you know, active, passive resiliency here. The point there is just we are proactive, making proactive investments in resiliency versus kind of like waiting for something to go wrong. So again, investing and innovating from a position of strength because the basics have been solved for us working on Connect and then from a self-service perspective, our ops partners now have the ability to, you know, enter a no-code portal and make adjustments to holiday hours or turn on, you know, some sort of emergency messaging or high call volume messaging to notify the customer without having to work through the actual technical themes themselves. So really simple, then forward-looking it really becomes around how do we expend, extend these benefits outside of card proper.

You know, we we envision a world where, you know, customers don't care how we're organized as a company, right? It's like i have, i have a bank account and i have a credit card account like just service me and solve my problem, right? So we think a lot around how we share the innovation across capital. One, create a shared resilient contributive platform, remove a lot of like those pockets of arbitrary tech uniqueness that we build and really just kind of consolidate investment and thought and really start with the customer versus kind of like the local needs.

So this side is not exhausted but is meant to highlight the benefits of Connect from a set of things that are kind of like always on in the financial services industry. Like for us, the price of admission for continued innovation is nailing the basics and Connect has really helped transform and simplify some of these key always on needs.

You know, rapid from a rapid configurations perspective, we're coming from a world that was hardcoded manual deployments to on Connects. We're able to standardize and create some dynamic configurations, reduce the volume of flows that we're carrying for. And generally just simplifying our our setup from an operational flexibility. I mentioned kind of like the no code UI but you know, just like increased levers and ways of making operational changes on the fly.

I highlight here. Q callback. It's very specific like when COVID hit a lot of uncertainty meant more than forecasted calls, you have agents that are moving towards home. So we had kind of supply demand issues with Connect because we already had the foundation in place. We're able to scramble acute callback feature in less than three weeks. So you know, if you call in and there's a longer than anticipated wait, we we just notify the customer, we will keep their place in line and give them a call back when the agent's ready.

The last one I call out here being in the financial services industry, just proactive risk management. It's just way easier to create like a standard across card and bank and financial services it Connects really helped in that regard as well. So again, not exhaustive but highlight some of the secret sauce of our operating model that we're now getting amplified value in everything we build and as we invest in more innovation, cool.

So with Connect as the centerpiece of our infrastructure, we're now kind of able to focus from, you know, modernizing the tech stack at the lower level to kind of thinking about how we want to build seamless and crossan experiences for the customer. Like I often refer to kind of like the older paradigm of, you know, a customer fumbling around in a mobile app and having the safety net of an agent, human agent to assist them. Our focus now is how do we best pull in the mix of AI and human to solve a customer problem, you know, in a more collaborative and assistive way.

So we we think a lot around meeting customers in the channel of their choice. And that was that's been only been possible by investing in the foundation and then thinking through, ok, there's going to be a mix of what a customer explicitly tells you they want to do and then kind of the things that we can apply through modeling or internal data on our end. And the power of Connect is in a very consistent way, we can apply rules and logic for what the next best servicing action to take based on your context or what you tried or where you fit or what worked for you last time you attempted to contact us

And what's been cool is how the tech platform marries up with the needs of how you actually optimize the network itself in terms of how you distribute to, to the best agent for that service they need and how you upskill them and how you forecast and run all the data cuts based on that. So Connect's been super powerful for kind of going from like the old paradigm of just, you know, a static telephony channel that that kind of like flows from picking up the phone to a human agent to one that we're really starting to think in a true omnichannel way.

I think Ram's going to come up now and kind of speak more specifically on how we've kind of brought that to life for the outbound servicing channel.

Yeah. Thank you, Chris. Good morning everyone. My name is Ramla. I'm the Director of Software Engineering in the CAR Technology space at Capital One. Just giving you a broader context about our contact centers. We have contact centers across the globe. We want to reach out to our customers or let our customers reach out to us at any given point of time throughout the year. So our contact centers are open 24 by 7, 365 days.

Um, so obviously, you know, if something goes wrong, I get a midnight call. So not a good thing to have, but thanks to, you know, the partnership with Amazon when we launched the Amazon Connect, the whole paradigm shifted where everybody is getting good night sleep.

Now, in my team, um, so getting into our, um, the dialer outbound dialer use case, what we have done is we want to reach our customers, we want to let our customers reach us at any given point of time, but also help our customers at any given point of time based on their need.

Dialer, outbound dialer use case is unique in a way where these customers are looking for opportunities to get back into a servicing mall where on their own terms, and we want to make sure we are empathizing with their situation, giving them options to get back into the servicing mall. But at the same time, we also have other personas like the agents, the supervisors and the administrators who do all the brunt of work behind the scenes.

We want to make sure our agents are not doing a lot of manual tasks of reaching out to the customers and getting the busy tone or just the dial tone or incorrect numbers or voicemails for that matter. We want to automate as much as possible so that the dialing and connecting to the customer is completely automated and the agent is connected only when there is a successful customer contact.

From a supervisor standpoint we want to give them the real time access to the data because what we are, we are in the business of financial services being, uh you know, our top priority is risk management and the regulatory and complaints requirements. We want to adhere to all the TCP regulations.

We don't want to reach a customer at midnight. We want to make sure if they are residing in California and having a New Jersey number, we are not calling them at 9pm Eastern time or midnight Eastern time or early in the morning. We want to make sure we are adhering to the state laws and the federal laws, right?

So having that real time access to the data ensures we have the real time suppressions and all of that. And not, but not the least, the last but not the least the administrators where who are working on creating this outbound dialer campaigns, they need to have access to this data to make sure they have the right customer information at any given point of time.

Going into our legacy platform last year, we had a hosted solution where the solution was not on cloud, it was on prem and hybrid, it was on prem and on cloud. So we had our own set of challenges with lack of, you know, access to the real time data, which was a big problem, manual and complex reporting.

Given the the the solution was not on cloud or you know, it was on prem and they had their own issues with their infrastructure, having the real time information of the reports was a big problem and given it's a hosted solution and we didn't have it, it was a black box to Capital One.

And the last thing you don't want to have is calls are happening and there is an infrastructure issue and there is a maintenance happening in the vendor, in infrastructure and our agents are unable to reach the customers, right?

And if you look at the entire ecosystem, we had the inbound calls, call routing and everything on Amazon, which was we had access to our data, we had no control of our destination. But the dialer, the outbound dialer solution was feeling like it was standing on a remote island which with no connectivity to the main line, that was a big problem for us.

So what we did was in late Q4 last year, we started doing an analysis on all the other opportunities out there in the marketplace to see what is, what is the alternative? Is there a way we can consolidate all the inbound and outbound infrastructure into one space so that, you know, we can have better control of our destination.

And Amazon Connect was a clear winner when it came to meeting the technology requirements, but also the nonfunctional requirements when it came to, you know, the resiliency standards, meeting the TCP regulations, giving access to real time data for that matter.

You know, we wanted to also ensure we had the risk management aspect, which is the resiliency, right? We we had a clear winner with Amazon Connect so that where we had the entire inbound and outbound capabilities in one P stack with clear transparency on what's happening in our ecosystem when it came to the real implementation, right?

It took us less than three months to deliver this. We leveraged Amazon's Pinpoint and High Volume Out Communication Channel services to integrate with our internal systems. The real task of setting up the Pinpoint and HVOC was five hours sitting with Amazon Connect team, engineering team and Capital One, software engineers.

We were able to configure everything but the complexity of the system was it's a pretty distributed ecosystem within Capital One because we have our own CRM platforms. We have our own, you know, data warehouse, Snowflake. And we also have our own system of record for our customer resiliency where the real collections data is coming from, the customer data is coming from the CR platform.

So what we have to do is build Lambda functions and host the data or the attributes in a Dynamo or store the files in S3. So it was pretty distributed architecture where we had to build these integrations across our ecosystem and all of that including the change management, the risk, the compliance, the legal reviews, all of that was done within 2.5 months and kudos to the partnership we had with, with Amazon Connecting.

We during that 2.5 months period, during the implementation, we were pretty much having a live Zoom with, with the Amazon team and we were able to get the issues resolved. And the cool thing about it is the issues were not getting resolved in days. It was less than a day.

We had, we had our observation shared with the Amazon Connecting, they would go back and fix it. By the time we come back online, the fix is already there scaled at uh you know, in production across the region so that we could continue with our roll out process.

And oh, not to forget the key metrics here. When we launched, we launched with 50,000 calls per day. But within a couple of days, we were able to hit 600,000 calls per day. And within a couple of months, we hit right now, we are hitting a million dials per day, which is not no easy for a contact center of our scale.

And one of the big concerns we had was ability to hit the or detect the voicemails when the customers. I mean, not everybody leaves a standard voicemail. Someone says leave a voicemail, someone says, thanks for reaching them or someone says they don't say anything.

So how do we differentiate and identify the voicemail detection? And we were pleasantly surprised and we were very happy to see that Amazon Connect's voicemail detection was hitting at 95% which was at least 15% better than what we had earlier.

And when, when it comes to the connection, you know, success rate actually reaching out to the customer at the right time at the right place, we had a 30% increase in the connection rates, which means we, it's, it's a better satisfaction, job satisfaction for our agents because they are able to talk to the real customers and not just leave voicemails, they are talking to the customers, helping them with their situation and getting the value for both the customer and for Capital.

Now, this is what we did is for Capital One. The next thing in our journey is how do we scale this at an enterprise level as part of our imperative and one of the biggest imperative we have at Capital One is to become a platform company and remove the arbitrary uniqueness like across Capital One.

We have the same Amazon Connect instance, which is not the same, but in fact, several instances of Amazon Connect across Capital One trying to build the same experience for the customers. And what we want to do is to remove that duplication of effort and optimize the way we build and manage our infrastructure. But at the same time, provide the top class customer service to our customers across product segments.

What we also are focusing is how do we keep Amazon Connect in the, in the uh in the center point and build an ecosystem around Amazon Connect where we are able to provide top class customer experience, meet our regulatory needs. But at the same time, remove uh our uh the complexity of managing this infrastructure.

The beauty of the whole dialer implementation is we are not, I mean, it's 100% serverless. We, as I said, we had um Lambda functions which are serverless. Uh you know, we don't have any APIs is all Lambda functions. We have all our attributes in Dynamo DB, which is a managed read replica and our custom portal is on Fargate, right? Everything is uh several less and we don't want to be in the business of managing infrastructure.

Uh sorry. Uh yeah, managing infrastructure, but be in the business of creating product innovations for our customers. I'll wrap it up saying no, we, we want to be in the business where we want to provide our customers with the top of the line experience across product segments and align with our mission statement which is changing banking for good.

Thanks for having us today. I would like to invite you back on the last to continue with our next flight. Thank you so much. Uh Chris and Ram, I thank you for being such great partners and I love the fact that we use that term. We talk about our relationship with our customers. Thank you for um being early adopters of the outbound campaigns, our learnings together have made the product a much better product and it's um it's tremendous and we just want to keep, we just want to keep pushing forward with you on that. And thank you guys for being an amazing users of Amazon Connect overall.

Um I love that. I love that story. I love the way we work with our customers and I love the way we see our customers results together and celebrate them together. Um as we talked about before, we really do think about the contact center with a couple of different sets of users in mind. The first one, always being the customer, then the agents who's a direct contact to that customer. And it's really the front line of that thing, who is really the face of your business. In many cases, this may be the one time you actually talk to a customer uh face, the one time that customer actually has an interaction with your business, it's sort of more meaningful than just looking at a web page, which hopefully that is a meaningful experience too. But um but this is a real differentiator in that way, the person is now going to think about your business and you have that opportunity to disappoint them. You have the opportunity to maybe meet their expectations or you have an opportunity to potentially delight them. If you really are thoughtful in the way you deliver your customer service to them, supervisors, obviously co this.

Um and they are often the ones who have some of the toughest times in the job because the tools they have don't give them the insights, they need to do what they want to do. And then of course, your administrators who often have the thankless part of the job, which is most of the calls they get are because someone's unhappy with something or wants something. Um or in the worst case scenario, something's gone wrong.

Uh we want to change that to a place where all these folks are empowered to deliver the things they want to deliver. I've yet to meet a person who works in this industry who doesn't want to deliver great customer service. I have met a number of them who are hamstrung in their ability to do so by some of the challenges they face every day.

So as I think about this, from the customer's perspective, they want to meet you where they want to meet you, they have expectations of being able to meet you digitally, they want to be able to call you, they want to be able to talk to you.

Um, we continue to innovate there and delivering automated experiences there that can delight. Um, this is, uh, this was something we, as I mentioned, built from the ground up to be able to, to improve that experience and to make it so that everything you're doing across one channel really accrues to another one.

We have great customer outcomes here. Texas was able to understand much better what their customers were calling about, what their students were calling about. And then they were able to actually, um, and then they were actually able to move much, much quicker to be able to deliver our customer service. In their case, they moved from, um, 15 minutes down to, um, sub-2 minute experiences for their, for their callers and their contacts.

Uh, this is not an uncommon thing for us to see because you can move so quickly. A lot of those things that are bottlenecks often, you know what you want to do, but because it becomes so much easier for you to make these changes, you can deliver them very quickly across so many channels.

Channels are an interesting part of the business here and obviously, customers are having more and more ways they want to contact you. One of the things we launched this week is SMS integration directly inside of Connect. So no longer do you have to deal with, what is a pretty heavy lift of trying to integrate with an SMS provider then trying to work that through your broader contact center. We make this very, very simple to do now for you, you can claim a number and get going with SMS and have your customers communicating to you.

Um with that, with that, it is a very highly um a highly leveraged communication mechanism for people um even more exciting potentially than SMS, which I think is incredibly cool is we are announcing and, and bringing to you in GA form in app and web based calling using WebRTC.

So you may know that Connect uses WebRTC for years for the agents. With the changes we've seen in, in people's expectations with the changes in the way people interact and also the changes we've seen in the network capabilities, we believe the time is right now for folks to be able to bring the benefits of the security, the integration of the experience, the power of being in an app when it starts and pick the right channel to communicate with the customer to you in a way you can now do that in a web page or in an application.

It's great to be able to bring back the data, you know about the customer, the fact that they are authenticated quite, quite likely and use all of that to speed up the interaction. On the other side, whether that be automated or for an agent, bring that data right into that experience and bring it right to the agent's fingertips.

Um another thing we're, we've been working on is generative AI for self-service experiences. Um this is an area which is getting a lot of press. Um I suspect some of you have came to this conference specifically just to hear about this. Um it is very exciting. It's also an area where we want to be very thoughtful in how we do it because as you know, generative AI is incredibly good at understanding things and expressing and communicating with folks. But that doesn't make it honest all the time.

Um you've probably seen some of the things where uh there was a lawyer who wrote, who wrote, asked the GPT-3 AI to write it a case study and provide a bunch of different examples and it wrote one and the examples are really good. Unfortunately, none of them were true. None of them actually existed. And he actually said that he sent it to another GPT-3 AI and said, hey, can you check and make sure these are real case studies? And it said yes, they were. And so, uh I don't think he has a, uh I think he don't have a license anymore.

But um uh well, probably, um there are some cases where that kind of error wouldn't be too bad in certain scenarios. If you answer a question wrong occasionally, maybe it's not a big deal. But I'm betting that in most of the cases, we're interacting with customers. It's pretty big issue if you get something wrong.

Um and doing things like math is something that they're not necessarily strong at. So what we've done is we've integrated this into our experience today and made our tools, the Lex NLU tools better for that. So your experience inside of Connect actually just improves it's much easier to design the bots.

Now, we keep the same guardrails though that we have with the Connect flows in the same observability. So you can control what's happening, the LLMs can improve the experience and then you can also measure what happened afterwards. So instead of saying, hey, there was a black box that did something, I'm not sure what, but this is the outcome, you can say no, we know exactly what happened and we know what we did with the customer and we can record that.

Um and that's that same flow language we talked about before, that's enabling you to get these benefits uh with our generative AI investment here. And that's just a really powerful thing for uh for you to be able to know you have the confidence you're doing the right thing for the customer. And while still staying up with the technologies that are going to make those experiences better every day.

And um I'll jump to the, I'll jump to the agents now. Agents of course, are the heart of so much of the operations here. And, and well, um everyone's told me agents are going to disappear within the next two years for the last 10 years. I, I haven't seen it yet.

Um and, and the reality is it's because not just because people can't automate stuff, there's so much, so many things, automating things is actually fairly straightforward. And with Amazon Connected, you know, you can automate many tasks in an afternoon for your customers. But that doesn't mean it's the right thing to do for your customers.

What the right thing to do here is actually often to provide the experience that transforms their interaction with you. The example I use is if you get overcharged by $5 on something, you probably not going to break the bank you for the year, maybe you would have to skip Starbucks. You'll probably figure out something, right?

But the feeling you have about what that company did to you is very real and very meaningful. And you want to know you want the satisfaction of knowing that they are going to correct this and they're not going to do it again. And so that's a place where talking to a human being in that scenario can really change your perception of what happened, correcting the issue, getting the $5 back is probably not the biggest thing for you there. It's probably more about knowing the relationship you have with that company feels respected and you feel like you are being treated right by them.

So doing this, doing this is that we want to doing this is something we want to do to, to remove all of the experiences agents have that are detracting from their ability to do that. So much of their time is spent, as I mentioned, all tabbing, it's fighting with tools. It's trying to figure out where to go, it's trying to remember where to go.

Um and so we built all of our, all of the heart of our system around this to be able to give agents what they need. So we have an agent desktop and Connect. That's something we didn't have a few years ago. Um but, and we were, and we obviously have some great partnerships with folks who deliver things like Salesforce to give a great experience for our integration. In fact, Salesforce U uses us for their own customer service.

But we also love the fact that we know that that some customers want to be able to configure something that's very specific and tailored to their needs. And so we started building out our agent desktop strategy. The first step was the Contact Control Panel, making that fully integrated all of your contacts in one place. So you can move seamlessly between them, whether they're a voice, whether there's some form of text or whether it's a task and then being able to now have a much broader experience about bringing customer profiles forward.

So we have a whole tool that can bring all of your customer information in one place. Pull it out a different system back end systems. We have um nearly 100 connectors to different back end systems and you can also connect to your custom ones there. And it uses AI to actually bring together that um and match those customer data records so that you can not have duplicates across it and know who's who.

So you can bring all that information from all the silos you have into one place to make decisions. You can give that information to the agent the way you want to. I talked about step by step guides that allow you to move through this stuff. We now offer cases to track issues for you and um and we also have knowledge and agent assist and I'm gonna talk a little bit more about that.

You have probably heard the announcement of Amazon Q yesterday. Did any of you see that in, in Swami's talk? Um yeah, the uh our, our work there is really exciting. We are generally available with this so you can get the benefit of generative AI today in Amazon, Connect for agent assistance. Uh, it's very, very exciting for us.

This is um this is something where we can pull in the information you have turned into an, an asset of information, sorry snippets and things that your agents can actually use directly to communicate to the customer instead of having to read through the knowledge, the article and try and convert it into English or whatever language you're choosing to speak. But trying to convert that into natural is something you could say again, take that cognitive load off the agent so that they're spending their time doing the right things with it.

So we use a generative AI to be able to provide next best actions to suggest the right things to say. We're continuing to iterate on this to improve it. We already have customers in production with it. Orbitz is one of them. They are reporting back that they've seen measurable significant improvements in agent handle time there and we are going to continue to work with customers like you to bring this forward. So you get the benefits right away of this.

The um here are some examples of what this looks like in the place. And in the middle there, you see one of the step by step guides, this is created very easily within the flows. And inside there, you see the experience where we indicate we've detected the issue. So we're actually able to listen in on the call and instead of the agent having to type in the what they want to understand there, we can actually search and create those answers for them on the fly and we do it in a way over there where it's very clear to the agents that it's happening. But we also try to keep it in a place where it doesn't interrupt the agent's flow of what they're, what they're doing at the time too.

So the agent is still in control, but they're getting the benefit of this in a subtle but powerful way that it helps them move more quickly and keeps them usually using the word flow but keeps them the flow of what they're doing to support that customer.

So supervisors, supervisors get a ton of benefit from the tools that make their agents better without them. But they also have a ton of work to do on their own to make agents successful. And so we have a very powerful metrics and analytics engine and Connect. It works broadly across all those channels and all this data. And it also works to support all the new features we bring out.

So you're getting these analytics from day one in a way you can consume them easily and directly. Your data is your data. We want you to be able to get access to it and we want you to be able to use it the way you want to.

We're continually innovating in this space. It's an incredibly exciting area for us. One of the things we brought out this week is new dashboards and these new dashboards make it much easier to see the data in time. You can see the powerful filters in there that allow you to look at things like week over week - show me what this, what this, how today compares to last week on the same day, what it was like last month on the same day, things like that, that allow you to have a much more powerful experience much more easily.

We still will have all of the powerful detailed data for every individual contact as well that you guys love about Connect. But this makes it much easier for a supervisor to get started and to get to the data they want quickly in a way they can consume it easily.

This is another one - for those of you who listen to Adam's keynote, you probably heard "zero ETL" used a few times. Folks may or may not know what that is. I know it's a bit of a jargony term, but it is actually a very simple value proposition which is instead of having to suck data out of one place and then try and transform it into the format you want it to be and then load it somewhere else to get access to it, instead we can deliver it in AWS for you to be able to point directly at and access it without doing any of that heavy lifting of a data pipeline.

This is an incredible lift for work that is just repeated by many, many customers over and over again. We were already a leader in giving you the data in a way you could consume it. Now, we've made it incredibly easy. I'll give you an example of this that would be easy to understand - for example, now you could simply point QuickSight or Tableau directly at this data to access it. This is in preview, it's something that our customers are incredibly excited about. It makes it really easy to join that data with other data points - for example, things that those tools have the ability to point to multiple datasets at the same time so you can see data from across your enterprise as well in one place.

So generative AI powered post-contact summarization - I don't know how many of you are using Contact Lens or if we have Contact Lens users out there. I see quite a few of you. So this is a tool that does a really good job of transcribing, understanding what's happening, alerting you, making it clear when you want it to detect things that have happened in the contact - did an agent say the right disclaimer for example.

One of the things it did is summarize these contacts, but the summaries were very long often because they were taking what was happening and essentially just making the transcription much nicer to read. Well, that's great, but what you really wanted was a true summarization which is show me a few lines of what actually happened on the call as opposed to making me try and figure this whole thing out.

So now with generative AI, which is exceptionally good at understanding those types of things, we can bring that to you directly in it. It is a great example where instead of having to go buy a generative AI product and try and bolt it on the side, we're just taking a product that's already really good and making it better with generative AI. So very exciting for us to deliver this to you - something I think you'll really appreciate if you've been trying to go through and understand what's happened in your contact center quickly.

That one is in preview but will be GA in the not too distant future.

Finally, I'll get to administrators - these are the folks, many of you probably have administrative tasks in your areas or certainly you're working with administrators to help manage your contact center.

One of the big things we mentioned about the deployments of Connect was that it can be so fast - Dow Jones did a full production rollout in 3 weeks. We've had a major bank in the UK that did their selection process in about 6 months. Then they did an implementation and evaluation process planning process for about 6 months and then they did their deployment in 6 months. So end to end from deciding they wanted a new contact center to them being fully deployed in Connect with 6000 agents live was 18 months, of which only really 6 months was the true deployment.

A lot of - there's a lot of factors here as the Cap One folks talked about - I'm sure your enterprise, if you're a large enterprise, has a lot of stakeholders, right? Managing all that was probably the biggest and hardest part of that whole deployment for them.

Another area I mentioned those customer profiles where we can bring all that data together. One of the most cumbersome parts of that has always been trying to figure out how to match the schema across the different datasets. It's a manual process to go in and say "Hey, this is the first name, this is the last name, this is some other ID."

So we said, okay, what we should do here is look at generative AI - if it's so good at recognizing things, could that actually recognize, could that actually help you recognize the data you need to import into your gen and speed up that really painful process? And the answer was yes.

So today, you can actually do this directly in Connect and it's generally available as part of our customer profiles platform. Folks are using those customer profiles to power agent experiences, to power automated experiences in Connect, and they're even using them to power experiences outside of Connect where they want to be able to point to something and change the behaviors of their app.

Choice gave a talk on this this week which was really compelling to talk about how they're using it that way. We love the fact that it's so flexible for people, very performant, so you can make sure those experiences are incredibly quick. You can choose to use it as a repository, as the main copy of this data, or you can also use it to join the data from other places that you consider to be the main copy and then also write it back to those places if you want to keep it up to date because it's really hard I know to keep your data across all these different things together. This makes this much easier.

I mentioned the no code UI builder for guides - I mentioned the step-by-step guide. So this is the way you actually just go in and use drag and drop tools to create those guides to make that experience for agents seamless and do it very quickly. A lot of fun, a really fun feature.

Third party applications - we love the idea that you build everything into a guide and integrate with your third party applications directly through those guides, but we also know that's not always practical or necessarily the best experience. We now offer third party integration directly into our desktop. Customers are loving this to be able to bring those parts in. This is in preview - we'd love to have you guys try it out if you want to. Really powerful. People are having a great time with it.

So what was the whole point of all this? I just threw a bunch of new features at you. Hopefully you followed most of them. We love shipping - we're shipping, we'll ship, I think more than 150 significant features this year too. And those features will improve your ability to innovate on top of Connect very quickly.

But really important is that it's one seamless solution for you to be able to do this with - so you don't have to go and buy and couple together stuff again, it's about your innovation. That's why we're doing this work.

Of course, you still may want some help. I'm guessing we talked about the partnership we have with our customers. We have a ton of offerings here that can help enable you to do that successfully. We will help you as much or as little as you want. I have customers I've never spoken to - they've never talked to us, they just showed up one day in my report of customers. I have other customers who I talk to on a very regular basis. We love the fact that we have both.

So we don't have time today for any Q&A but I can step outside in a minute and we can have that conversation if you guys have specific questions for me.

If you would like to join us tonight, we're having a little bit of a customer appreciation event. You can get signed up for it and do this quickly because I'm going to give you a couple more things to look at in a second. You can also talk to Ryan - Ryan will help you get signed up for that if you have any trouble.

Here are some of the breakout sessions with customers - again, it's all about the customer success here. These are some of the customers speaking about Connect on our behalf with our team here. We love that.

And here are some more. Thank you so much for today. As I hope it shows, I'm incredibly excited about what we're doing here with folks like you. We want to do more of it. So thank you so much for coming over here.