C-suite leaders talk generative AI and applications

Please welcome Vice President Applications AWS D Kumar, Kareem. Hope you're having a good re:Invent. My name is Philip Kumar. I'm Vice President and I lead our AWS Applications group. We have an exciting session for you today. I'm going to be interviewing a group of our customers to see how they're thinking about generative AI for their own customers. And I'll also be announcing a bunch of AWS application launches all geared towards workforce productivity.

You know, customers have always loved building and using AWS infrastructure services for the applications that they have built. But increasingly customers are also coming to us and asking us to build end to end solutions that bundle together sophisticated AI and ML to be able to solve their hardest business problems. These up to stack solutions have the same characteristics that you would come to expect from any AWS service, the same reliability, the same scalability, security pay as you go pricing. But today we're going to focus on a subset of these applications. But there's one thing in common that all of these applications have, they're all geared towards figuring out how to improve and make the lives of the folks who are working, be it a contact center employee, a supply chain inventory planner or a knowledge worker more productive in their day to day jobs.

Take a look at this video of one possible future of work scenario in today's world. You can feel like complexity is the new standard operating process. Imagine if technology could remove some of your employees' daily road, how could your business evolve and grow? Welcome to the future of work where every step forward brings us an experience that's simply better for employees and customers rather than relying on badges or pass codes for access. With Amazon One Enterprise employees can gain access across physical and digital resources with their unique palm signature that can't be shared lost or replicated employees can simply hover their palm over the Amazon One Enterprise device to access their buildings, secure areas, enterprise assets and software resources.

With Amazon One Enterprise businesses can boost security and streamline operations with more convenient contactless employee experience. Now, what if that same simplicity could extend to your employees computing environment? Imagine a world where their computer is available over the internet wherever they need it. Virtual desktops make that possible. And their new Amazon WorkSpaces thin client device makes it easier than ever to access those virtual desktops all with a device that costs a fraction of the price of a traditional laptop or desktop computer.

IT administrators don't have to worry about provisioning or shipping devices and end users can set WorkSpaces thin client up in less than five minutes. Amazon ships directly to end users simplifying logistics and provides a simple setup guide to help end users attach monitors, keyboard mouse headset and webcam and connect to the virtual desktop surface with no IT intervention needed. And now that it's simpler for employees to connect to virtual desktop environments. What if we could also simplify what it takes to do their job with generative AI? That's a reality.

Amazon Q gives employees like contact center agents using Amazon Connect guidance in real time using generative AI agents chatting with customers can rely on Amazon Q to give them recommended responses and resolutions in Amazon Connect like helping a customer understand the perks of your credit card reward options, upgrading their current card to a traveler rewards card and resolving an open case and happier better equipped agents equal happier customers with AWS. You can step into a future where technology empowers your business with simple solutions that simply work.

Join us as we re invent a simpler, more secure way to work. As you may have noticed, there were a few product placements in there which brings me to the first of our announcements for today, we are launching, it's generally available. The Amazon Workspace is 10 client. It is a low cost hardware device which costs a fraction of what a laptop costs. It comes preloaded with all the software necessary to be able to connect to your company's applications. The IT administrator don't have to worry about provisioning. They don't have to worry about ordering a lot of inventory. And it's very simple. In just a few minutes, your employees can set it up and be accessing your applications. I'm super happy for this. This has been a long time coming and I can be thrilled to see what customers think of this device.

There's another one that we launched, which is now in preview, it's the Amazon One Enterprise edition about a year and a half ago, we launched Amazon One which is a simple, secure, effective way to enter, identify and pay and we use this in several of our retail facilities. We're bringing the same power of using to be able to use your palm to be able to identify yourself to an enterprise. Now, employees can access corporate workspaces, shared digital resources, data centers or even for time and attendance using exactly the same sets of tools that were available to administrators. So your administrators don't have to worry about figuring out how to set up employees or worry about different ways of accessing different facilities, whether it be hardware or software.

Amazon is one of these long line of investments that Amazon has made over a period of time in AI. But from the earliest years of collaborative filtering, which made itself into our personalization algorithms or pricing and supply chain which powers the core of our consumer business to more recent inventions, like just walk out on Amazon One, we've always found a way to bring very practical users of AI to the workforce. We are again at the cusp of another revolution, you know, generative AI has captured the imagination of everyone here. And as Adam said, we in this keynote this morning, we have a very unique proposition and a unique way of how we are thinking about it. But we will have one thing in common, the ability for people to have choice on how they use it as well as the ability to find the right business model that works for your customers.

As customers are thinking about generative AI, as I said, there's been a ton of excitement. There's been a ton of interest in it, but there's also a lot of questions that have arisen first. Is it enterprise ready? Is it private? Is it safe? Is it secure? How will my data be treated, foreshadowing all of those things were thought through in a lot of these things? But don't take my word for it here to talk about some of this as to how they are thinking about generative AI for their customers is our esteem panel. Come on out guys.

What? Yeah, thank you, Alex. Do you want to introduce yourself? Sure. Hi, I'm Alex Hood. I'm the chief product officer at Asana. Been at Asana for about six years Asana is the leading enterprise work management platform that makes it easier for work to get done across organizations.

Tracy, thanks. Yeah. Hi, everybody. I'm Tracy Simpson. I'm the vice president of supply chain at Woodside Energy. We're a global energy company headquartered in Perth, Western Australia. And our company vision is to provide the low cost, lower carbon energy the world needs. Um I'm a bit biased but I think I've got the best job in Woodside. Um my, my team looks after the end to end internal or inbound supply chain for Woodside. So that's everything from contracting, procurement, warehousing, inventory, logistics, corporate property, and travel. And the reason it's so exciting and so interesting is we get to see end to end across the three pillars of, of Woodside. So oil gas and new energy.

Hi. Um I'm Atish, I'm CIO Dish Network. Um been there for five years. I think a lot of, you know, us um from a direct broadcast satellite perspective um where we have multiple lines of business. So um pay tv is definitely one of them. We also have Sling, which is our live over the top streaming television. Um we're also in the wireless business. So as as many of you know, the the brands of Boost, Mobile, Boost, Infinite Gen Mobile, those are all our brands that we support. Plus we are the nation's only five G cloud native network in the United States. So I'm covered by AWS powered by AWS. That's what I like to hear.

Well, let me, let me sort of ask a question that might be on all of your minds. But Alex, I'll start with you. There's this little thing called JA I, that's back loading around. It's already mentioned once or twice during this conference and months leading up to it. First question to you, hyper reality.

Well, I'll tell you what, I'm the biggest skeptic of tech trends out there. And, um, this is the real deal. It's the real deal. I feel personally privileged as a product builder to be in the prime of my career at this time because what is ahead of us is so powerful. The new set of tools, the new capabilities to deliver on all the missions that we have is pretty amazing. And what's also interesting is we, we all now sit on piles and piles and piles of data that allows us to capitalize on this new technology. So when Chat GPT really was products, it just the lights switched on for all of us and now we're all trying to make sense of it. So I think, um, it is hype, but the hype is real.

So, Tracy, how are you thinking about generative AI in your business?

Um so I think de ever be thinking about it. Um I guess from a supply chain perspective specifically in Woodside, we're still in the process of really trying to understand um, what we're going to do with it. Um so what we are doing to prepare for the fact that we expect to use general AI in the future is we're working a lot on the foundation of data models, the ways of working in the enabler, the key enablers that will enable us to go after it. Um when, when we're ready to go to go for it and we know that we have lots of partners starting to introduce things.

One thing we're blessed with mentioning, working off what Alex said. So supply chain, we're best with an abundance of data. Um the part of the problem we have is with the resources that we have, we don't always have the ability to access it, interpret it and act on it. So what I'm really excited about with general AI is the ability for us to actually start to synthesize and harmonize that data in a way that we can actually start to generate insights and actions from that. And that's gonna enable us to do much, much more interesting and exciting things with our business.

I think it's a really important point like in almost all our customer conversations, every conversation of generative AI results in a conversation about how data is stored where it's stored. What types of access do people have and how do you sort of combine and have generative AI models work across a set of data? So I think having a data advantage is having a generative AI advantage. So it is the foundation of it.

I tell that like for you, how are you thinking about how it's going to improve your customer experiences?

Yeah, that's a good question. I mean, I think is I think about gen AI and kind of moving out from sort of the exploring and tinkering. The obvious use case for me is probably from a customer experience standpoint is in our contact center. Yeah. And and I think the most obvious place is chat, right? So we have chat bots today. They they probably handle about 70% of a customer's call and then 30% require a human handoff. But I think we have the opportunity to turn that 70% into a really rich customer experience going forward. And I think we have the opportunity to chip away at that 30% and make that customer hand off 20% or 10%.

The reason why I think we can make it a really rich experience is we can really turn that chat experience into something that's more highly personalized. It's definitely going to learn from the millions of other chats. So the response and and solutions we provide our customers are going to be well informed. So i think, i think we're in a good position to start to lean into chat in that regard.

I think the other place that, that it could play in a contact center as well is really a gen A agent assist. Meaning, you know, before, before we actually put any of our agents on the phone with our customers, they go through a lot of training so they have to learn our products, they have to learn our systems, our knowledge bases our flows. I think with gen AI as a sort of a copilot for them, they have the opportunity to leverage gen AI at their fingertips. And then really maybe some of that training is actually going to transform into prompt engineering as opposed to system training going forward. So I think that will get our agents out taking calls with our customers sooner but also a value-added experience for our customers because we're going to be able to answer questions and provide solutions in, in a much more expedited fashion than we do today.

I have a follow up questions. You said something about chip re about 30%. Are you seeing, do you envision a future without agents or do you envision a future with where the agents are augmented to do different things?

Yeah. Yeah, I don't see the agents going away tomorrow. You know, that conversation might not be used away in terms of, of a virtual system to be able to handle something holistically. I i do think gen AI still needs some guardrails associated with it. I i do think a target rich environment though that it's in that vein is, is an IVR. Um, if you think about IVR process, I think we've all lived through it. It's, it's a little static, it's impersonal. Um it can be frustrating. I think we've all kind of been there. I think with gen AI kind of like the chat bots. Um we can actually personalize it to a higher degree. We can um have it be more of a human to human experience even though it's a, it's a gen AI agent on the other side of it. So, so i think that's a starting point. And then I think what we'll do is we'll probably measure and we got to keep measuring. I think the experience would probably be stronger than what the IVR is today, but probably not quite where an agent is.

Yeah, i think this is very similar to the, the earlier days of when machine learning and deep learning came in and people felt that this was going to be a massive shift towards automation. And while it was, it was also a shift towards agents and employees doing more value-added tasks relative to the things that they were doing. So there were things that the machines were incredibly capable of and there's things that the agents and the humans were incredibly capable of and they were, it was a good mesh move on to productivity benefits of generative AI the, the question that is on, you know, the audience's mind is definitely on that I hear from customers is this, is this enterprise ready? And as I'm thinking about this, you know, what is the ideal state and the expectations I should have? What are some of the concerns that I, when I as a customer employing generative AI for my business, what are some of the concerns and the ideal state that I should be thinking of? Let's start with you, Alex.

Sure. Yeah, let me, let me start with the promise. I mean, the, the what, what customers are talking a lot about right now are chat bots

And I interpret that as a product builder as back to the Henry Ford quote of "If you ask my customers what they'd want, they'd say a faster horse." And I think right now the faster horse is Chet bots. However, I'm excited about just really deeply embedded AI driven experiences in supply chain. Or can you just imagine the world's customer experience is redefined where you're not trying to deflect from a call to a call center, but getting that help right there at the moment that a customer wants it.

Well, we're, we're excited about copilots at Asana as well, but we're really excited about for us in productivity software is being air traffic control. So that means um you've got a lot of things in flight, but you can optimize how all those projects and tasks are, are going along um in real time, fully optimized and when the weather changes or there's drama, then you're able to quickly pivot people and teams to, to get to different outcomes without causing collisions and doing it in a way that's very, very safe. So that's the promise. But what we're hearing as well is, um, gosh, we did this survey just to get ourselves smart on what customers are saying about the space and nine out of 10 folks say, hey, I'm concerned about the ethical use of AI.

So before we started product road mapping this cycle, we thought hard about what are our AI principles or how we build and we want to share a few of a few of them with you. The first is that we build AI to help the world's teams achieve their goals. So AI is in service of us doing great things. Um it, it helps us be more human at the things that we're passionate and great. At number two, we promote transparency and what that means for us is AI should and needs to unpack its thinking. Show its work as you were saying, work. Yeah. Yeah. And just as humans should too because often when there's differences of opinion, it's because we don't share assumptions.

The third one is people are accountable for decisions. So I think AI will help us along the way. But you can't imagine a world where um individuals are throwing AI under the bus for bad decision making. So we are going to stay in the driver's seat and that's how we're thinking about building software.

Uh Tracy, the the productivity benefits, you said that you are early in this journey. Uh you know, a lot of, a lot of customers have this two pronged belief where they feel like they're either too late or they're too early and everybody is in the same boat. And I think one of the things that you were talking about earlier, which I think the folks here would be very interested in is in this early stage. How are you thinking about setting the right vision for your organization for what you want out of AI, and what are all the things that you're concerned about relative to any other type of project that you undertake?

Yeah, it's a great, it's a great question and look, um I'm clearly not the expert in digital. Um I rely heavily on my digital colleagues at Woodside for that and, and they do a great job of educating us on some of the things that are key for, for our business. One of the things I'm going to borrow from a colleague of mine is a quote where he says "just because we can do it doesn't mean we should." Um and I really like that ethos because it really sets up the structured thinking, I think that we need to consider before we enter these projects.

So the first thing for me is what are we trying to solve for? Um making sure we know what our use case is. The second would be um making sure that we understand the role we think JAI can play. And um and you know how can we do it because it does the technology is it is it up to speed the third for me? And it's essential is path to value. So these are not shiny new toys, we want to give our teams, these are things that have to drive real business value and benefit for us.

And so once we've kind of evaluated, can we do it, then we get into the should. And so for us at Woodside Energy, there's a key principle which is we have five company values and we have to make sure that anything we do is absolutely compliant and consistent with how we want to behave as a company, the employee experience that we want to have and what our customers expect of us.

The final one for me is timing actually. So it's a critical one we're blessed in supply chain. We've got there's people in this room, whether that be farmer or retail or any sorts of many industries that are not energy related, but we can learn from some of those people and we will be able to see innovations that, that may generate from other industries um that are gonna be really beneficial for us in terms of how we adapt to adopt those.

So we may not want to be on the front edge of, of all of these things. And, and in some cases, we may deliberately choose to wait to see um use cases play out in other industries. So for me, that's really exciting because we get to choose the bits we want to go early on and we really get to stagger the use cases and really understand collectively across the supply chain, it sort of domain. What, what will work?

I think this is a, this is a super important takeaway. There's this notion I've met customers on both sides of the spectrum where they feel that they are, they are not sure of what's the best way to move forward. But at the same time, they feel that if they don't move forward, they'd be left behind and being thoughtful on what it is that you're trying to achieve and going through this process is always very useful at, until I'm sure you've been in a meeting or two in the past few months on generative AI the generative AI approach for dishes.

One of the questions that tends to come up in this is that in addition to concerns and questions like who's making the decisions, is the decision-making process very different. Is it very similar? What are the expectations for a genAI driven project relative to any other project that you undertake a dish?

Yeah, I'd say from an expectation expectation standpoint, genAI probably comes in two flavors. One is, you know, folks like you who are embedding it in your, your platforms, right? And that's probably our fastest path towards adoption, at least in terms of our core processes. Um I mean, we're, we're going to see it in, in productivity tools like office 365 or google workspace. My HR department is gonna be super excited about being able to do a job description in minutes and, and my team building slidess in in hours not days.

Um they're certainly gonna be excited on that front. But when you look at our core processes, I think the fast path for us with gen I gen AI is probably going to be things like, you know, Amazon Connect from you guys um that will potentially can unlock some of the capabilities I was just talking about or it could be a platform like ServiceNow, you know, if we have tickets that are going into ServiceNow today, um it would be fantastic. We also use that to isolate the issue to resolve the issue and, and draft a fantastic root cause analysis out of it.

Um Adobe is another one I think, you know, they presented at this conference as well about Firefly. I think there can be a lot of creative there that we'll be able to leverage and put into our digital assets. So I think it's, it's we've been spending a lot of time with partners like you and, and really exploring it. Um but with that said, we, we'd be negligent or I'd be negligent if, if we weren't doing some POs on our own.

But the second way it can propagate which is us learning and applying and doing POCs with foundation models and LMs. So we've done some early POCs with things like Falcon and LAMA where we've taken order data and see what, you know, information we could glean or take our CI/CD tools and draft release notes using gen AI. But it does kind of segue into a little bit of the concerns that you were talking about that.

I want to make sure we have in place. So one big one for me is governance. You know, we really need to govern it on a lot of different facets, meaning first of all, just in terms of the work that's being done, we don't want lots of science projects, we do want to learn, but we don't want to just kind of generate science projects without really focusing a value creation. So the use cases really matter to us.

The other concerns we have, which I think everyone has is really around IP infringement as well as protecting our own. IP I think a lot of us have seen things in the news where where people have had a little bit of a footfall. We don't want to recreate that we want to learn from what others have done on that front. And then Alex kind of touched on this as well. Another big piece for me is also ethical AI, you know, there's a lot of things that these models can do that can influence a bias. We don't want to introduce unintended biases.

So it's going to be really critical for us to understand the data says that we are training these models with as well as testing and inspecting the outcomes and the outputs that's coming from gen AI. So, um a lot of opportunity but a lot of things we've got to be watchful and mindful for as well.

I think that that's a super critical point is that models can be biased in certain degrees, but the data set that's used to train them can have their own natural biases. So like if you're not vigilant about watching out for that, it can lead to unintended consequences.

Um last question for all of you, you know, as several of our teams are working together on many things, whether it be AWS App Fabric Supply Chain or Connect as a technology partner and as an application provider. How can AWS help your customers?

Well, it's just interesting hearing our conversation here. We Tracy, I hear you say, oh, I, I don't want to be the bleeding edge on some certain I want to hear what AWS helps us, helps develop as a standard for Sana, we want to be on the bleeding edge with AWS and figure out with AWS, what is the standard like, how do you create AI that has the right privacy settings. How do you create, how do you get the best amount of data to pay off the promise of AI but do it in a way that is safe with, with and secure.

So we see our partnership with a with AWS really about co creation in particular. We've been proud and excited to be building AppFabric with AWS. And what AppFabric does is it connects SaaS applications to serve privacy, use cases and productivity use. So that that security is taken, taken out of consideration. So it makes it very, very easy to look across data to do great things with AI. And it's been fun and exciting and we've learned together along that journey with the AppFabric team in Asana Tracy.

Uh so for me, it's kind of an easy answer because the clues, the clues in my job title. So supply chain, if I think about AWS, the DNA of the company is, is really kind of quite centered in supply chain. So we've been really fortunate to be able to work with the AWS supply chain team um in, in the um order insights module and, and optimizing that in terms of how we would use that within our business.

I think one of the things the AWS supply chain teams brought to the table is really that domain experience as well as their shared experience from many of the other customers in this room. So they've really been able to bring that to the table and that's helped us um get sort of scale and speed. Um I think that we would not have been able to achieve on our own, which has been really exciting.

I think the other thing for me in terms of what can AWS bring to the table is, uh, my view is possibly a bit different from some of the other digital spaces within Woodside, but I manage a supply chain. So I want my best people doing what they're really good at. So my inventory plan is, for example, planning and doing the things around my inventory space that really make a difference and drive value for us as Woodside.

Um I don't really want to have to pull them out into some sort of large development project that's going to take them, you know, 18 months, two years to develop something highly customized and really difficult. So I'm really, really excited about how we might work with AWS supply chain to sort of um adapt and, and uh kind of really um deploy rather than sort of spending, having to spend a lot of time. So I think that's the definition of partnership is really relying on your strengths.

So we bring the knowledge of what we want to do and, and how we want to get there. And really the AWS supply chain team has helped us, can navigate that path and bring their expertise to the table to help us get a bit quicker.

It's great to hear. Attila. Yeah. Do. But I think if the question is, how can AWS help? Ii, I guess I'd really say how can you not help? It'd be a massive this opportunity. If you didn't, you didn't, weren't on this journey with us.

Um I mean, you power our storage and compute you power our data lake. Um you, you provide applications like Amazon Connect like Supply Chain, like Clean Rooms. Um and then you have platforms like Bedrock and SageMaker and others. I mean, the reality is that all the tools are there and you're deeply ingrained in our ecosystem.

I think the big thing that we probably need from you is, is really a deep understanding of our ecosystem and, and what operational excellence means to us because I think, I think we've used the word use case a couple of times. I think I've used the word use case a couple of times already. It's, you know, it's, it's understanding the use cases. We need to achieve that operational excellence. You, you guys are uniquely positioned to help us.

Um so we just kind of need to bring you on the journey as well. I think the other thing you can do to really help us is really coming back to the concerns i raised. I mean, I think there's things you can do to help us. When I mentioned IP infringement, you can help us with indemnification.

Um you know, we really want to make sure that we're on this journey. We're, we're also thinking about um not only just our assets but our customer's assets, our data and, and, and all the regulatory concerns as well. So, you know, i, I look at this journey and it's very similar and I might be dating myself, but it feels very similar to when public cloud first came out because there's a lot of preliminary concerns around risk, security cost privacy.

I think all of the things that you guys have already been through already and in your first journey with public cloud, we'll probably be discovering again with gen AI. So those are things that you definitely can help us with and, and coming back to the guard rails. I talked about whether it's ethical AI or um just making sure that we're actually using it in a way that doesn't create hallucinations and gives bad data to our employees or our customers. You guys know the best practices that we need to employ. And so we're going to definitely look to you to help in that capacity.

What challenge accepted. Ok. Awesome, awesome. All right. Well, I want, I want to thank, I want to thank all the few again. And thank the panel. A big round of applause for the panel.

Yeah, we just talked about some of the things that customers and what's on their mind regarding generative AI. And I want to get to three of our launches that have generative AI in a very practical way that you can consume.

The first, I want to start with this Amazon Connect. We launched this in 2017. It's our cloud based contact center. It's simple, easy to set up. It is a fully managed telephony network. It bundles sophisticated AI and machine learning and not surprisingly, it's one of our fastest growing services. And as Attila said, this area is ripe for invention when it comes to generative AI.

When you think about the kinds of things that you can do with generative AI, there's four launches that I want to talk to you about. The first is for administrators, self-service, chat bot like chat, self service, chatbot creation, which is available in preview today. Chat bot creation has always been a little finicky. It requires a lot of expertise

There's a bunch of customization necessary, bunch of edge cases to deal with. There's some operational management, but now with self-service, there's no need to sort of hire experts to be able to do this. You should be able to do a lot of this by specifying these things in natural language, easy to set up, easy to maintain, eases the burden on your administrator so you can get your contact center up and running quickly.

The second one, which is again for administrators is Connect is a very powerful application, but it gets even amplified when it connects to very different types of data sources. Now, these data sources all bring in very reasonably similar data elements that are sort of named differently in different parts of the organization. So one of the things that's necessary is the way to be able to create a customer profile data mapping, which can be highly accurate, which can be driven by. This is a, this is a great example of how generative AI can figure out and suss out the differences between the same field that is mentioned in the operations database, which is very different than the marketing database, which is very different than another customer service database and having a composite view of your customer is the starting point for any kind of outbound or marketing related activity.

So that's another thing that's generally available for admins. The third one that I want to talk about is this notion of post contact summarization. Now, for supervisors, this is available in preview today. What they have to do is they have to go through and pull through transcripts of what a call center agent is doing so that they can provide the right coaching and training and they can generally get to a small fraction of it with now regenerative AI, they can get the right summarization across the entire population of transcripts and build agent evaluation forms on this with very little effort.

And Adam mentioned this in the morning for agents Amazon Q and Connect a real time intelligent conversational assistant that allows that is listening for customer intent, the right phrases what you need to do to be able to get the agent, the answer that's necessary as well as the actions.

Let's dive into some of these features that look like in real life for agents and supervisors. Think about the day in the life of an agent. You know, customer call comes in. The agent has to understand the context of the call, understand everything about the customer. How often have they contacted the company? What is it that they are specifically calling about? Then they have to scramble to be able to find the right sets of information to be able to help the customer with their issue all the while, while being attentive, while being empathetic and being patient, speed matters here and an agent that's just one call. Now you have to do this tens and hundreds and hundreds and thousands of times.

Take a specific example. Let's say you're calling in to upgrade a credit card and you have and you have some questions because you went to the website and you didn't totally follow what you are eligible for or not when a call comes in and you want to know the benefits of the call. The agent has to know whether there is what your unique status is. What is it that you are eligible for? What are the benefits of that particular card? Are there any exceptions? Typically they are scrambling and to try to find this information through six or seven or eight different assisted applications that may be in very different parts of the business all the while while the customer is on hold.

Now with Q the customer can just as the call is coming in. This real time assistant is listening, it is able to pull out the right intent of the specific things that the customer is looking for and use your knowledge base, your FAQs your policy documents to be able to build the right and pull up the right information just in hand for the agent. But it goes a step further. It doesn't just provide them with the ability to be able to pull up this information. It also gives them the next most helpful suggestions on what they need to do in order to resolve the customer contact, being able to resolve these customer contacts quickly, keeps customers happy. It reduces your call handling time. So you're able to both lower your costs as well as improve your customer service levels. It's a huge benefit.

Flip to another user which is the supervisor and the supervisor has also a very interesting but a very difficult job. You know, they're, they're managing tens or hundreds of thousands of agent, agent attrition tends to happen on a regular basis. And in all of that, as I said, they can only listen to a very tiny fraction of the calls. But now the generative AI, they can, you can get the entire population of all the calls summarized in a very concise manner, Contact Lens automatically summarizes the call. It allows you to be able to figure out adherence to the script. It allows you to be able to provide the right coaching and the guidance that's necessary so that you agents can up level. So you have a new agent that starts their ability to be able to scale to become a superhuman agent is very easy. It's very useful for the supervisors too because now that they can actually do coaching at scale rather than extrapolate from a very small sample of calls that they could identify with.

From administrators when it takes to setting up this to agent assist to supervisors. Here's what customers are saying about using generative AI features in Amazon Connect Orbit Irrigation said that they would spend 10 to 15% time less on every contact that can be huge. Fujitsu said that for supervisors performance evaluation efficiency improves by 60%. These are staggering numbers, they add up all of this results in better customer service at a lower cost.

Contact centers is just one place where we are helping workers switching to the second application that I want to talk about. There's been a huge sprawl in SaaS applications. The reason that the SaaS applications have increased in popularity is because they're purpose-built. They so very particular use cases, everyone who uses it, I use it, you all used it. We all love these applications now and you know, most organizations use somewhere between 50 to 100 of these SaaS applications on a regular basis.

Now, while this has been super popular with employees, it is not necessarily as popular for IT administrators or security professionals working at companies because they lose the ability for observability, which is a critical part of their job by that. But I mean each of these applications has their own APIs, they have their own log formats, they're not necessarily interoperable. And while companies have built point to point integrations, these point to point integrations can be costly to maintain, they are hard to hard to replicate. And as new versions come out, they have to keep getting it upgraded.

So in June earlier this year, we announced AWS AppFabric and as what Alex was talking about, we this allows us to be able to connect multiple SaaS applications. We normalize all the log formats so that the individual SaaS applications don't have to do it, provide a single pane of glass, it connects to your security applications that you've used. So if you're a security professional in a company or if you're an IT administrator, you get the level of observability that you have always wanted. But now you can do this without any of the necessarily heavy lifting.

There's a reason why our AppFabric has become very popular very quickly with customers is because it specifically hit on what IT administrators and security professionals had been craving for. We also announced at the same time how applications can do more to be able to improve employee productivity. One of the things that tends to happen is that take a look at your own daily lives. On, on average, you use somewhere between 2 to 5 applications on a daily basis. You're also context switching quite a bit between these applications to do those tasks because they are slightly interrelated.

So for you to be able to do your job, you need to switch from one application to the other. The problem with generative AI always comes back to data is that these disconnected applications create data silos. So even if you have a generative perfect generative AI application built, it's only operating on a very small subset of your data when you want it to actually operate on a much larger subset of data.

Let me show you a little bit in this demo. As to what we're thinking about today, employees typically switch between 6 to 8 SaaS applications from different vendors to complete tasks, switching context constantly. Application developers see an opportunity to use generative AI to reduce this context switching and provide a richer end user experience but are often limited to data within their own software without including cross app data, data silos form and generative AI results miss important context from other applications that help end users make better decisions.

AWS AppFabric for productivity, reimagines productivity in SaaS applications by generating insights and actions with context from multiple applications. AppFabric fully manages the integrations with multiple SaaS applications and automatically normalizes data for use across apps. This means developers can enable AppFabric productivity features directly into their applications, user interface to maintain a consistent app experience for their users. Whilst surfacing relevant context from other applications with cross application context, developers provide a more personalized user experience that increases adoption and loyalty.

End users benefit from accessing insights they need without interrupting their workflow. Embed AWS App Fabric in your application's new or existing generative AI assistant to enable your user to be more productive and more connected across their preferred apps. As Alex was talking about this is now available. This is an Asana application and rather than just having aggregated insights from just only what Asana has access to. We're now pulling in information from the applications that you use and we're surfacing it in the applications that you're currently using.

That's super important point is that one of the tenets when we were designing this was that we didn't want what's the point of having people improve their productivity if they have to learn an entirely new way in order to do their daily jobs, that didn't seem very reasonable. So one of the things that we said was if you're using a set of SaaS applications, in order to be able to do your job, we wanted to be able to pull in the right level of insights from the other applications and surface it in the existing application so that you can, you don't have to context switch, you get the information that's necessary that you need to do. And now you're suddenly a lot more productive. And this is available in Asana today. Mirror Smartsheet Zooms coming next and soon we sort of want to build this on a series of applications because we feel that this is one way where you can get people to be able to do the productive task that they need to do without the context which that they're subjected to on a daily basis.

Another place where we're using generative AI is supply chain and this is the last one that I want to talk to you about today. We launched supply chain in April with three things in mind. One of the things that was very top of mind for us was companies had spent a lot of time and energy in their own supply chain applications and they come to us and they asked us for some very specific things. The first one was we wanted to be able to pull in all this disparate different supply chain applications that they had built over a period of time into a single common supply chain data lake.

On top of that, we felt that one of the areas that Amazon has had, you know, 30 years of experience is this notion of forecasting and demand planning. So we built a demand planning module that brings together Amazon's expertise to be able to attempt to allow the models that's necessary to be able to forecast demand. Once you have the supply chain data lake and you have demand planning, you can then layer on an ML powered insight thing that allows them to be able to figure out exactly where they have inventory shortages and what actions they need to take.

But as for those of you who have been, who are associated with supply chain, who have been in this field, this is a very small subset of what supply chain covers. So today we're announcing three new modules, supply planning, interior visibility, sustainability, as well as Amazon Q, an assistant that's going to be resident in the AWS supply chain. Let me talk about some more of these.

As I said, supply chain is not really about demand planning. You can forecast customer demand, but at the end of the day, you have to turn that customer demand. You have to order products from vendors, you have to be able to place them at the right warehouses and you have to get these products and locate them in the right locations. Now, when you have a few products, maybe this process is easy. But companies generally have, you know, hundreds of thousands of products, they're working across hundreds of vendors, geographically distributed, multiple different warehouses. And soon this becomes an optimization problem.

This is one of those areas where Amazon has spent a lot of time and energy in figuring out how to build the right kind of supply chain algorithms to be able to provide that level of visibility. So the first module in supply planning now it allows people to be able to help plan position and replenish and figure out exactly where inventory is. They need to. It gives them the suggestions on where the specific actions that they need to take in order to be able to make sure that they are adhering to the right efficiency levels and the right service levels for their business.

If you go back to the original car manufacturer, the example that I had out there, we were thinking like when we think about supply chain, we think about finished goods, right? You buy, you buy a product and you sell it. So lots of finished goods. But there's seven sets of companies, manufacturing companies in general who are very interested in figuring out how to be able to do this at a component level, the individual components that go into the product, not just the finished product.

Think about a car engine as an example. A car engine isn't a single product. A car engine is about, you know, 50 100 different products that are coming from multiple different vendors. Each of those products has their own tier of vendors that they are actually dealing with. And one of the things that happens that works in supply chain is that any blip in any part of the supply chain has a cascading impact that then causes you to not be able to get that finished product built to be able to service your customer demand. And another key tenet for supply chain is that the earlier, you can figure these things out, the more visibility you have and the more time you have to be able to react to the obvious fluctuations that are going to happen as part of any supply chain process.

So we built n tier visibility which now goes upstream and it extends visibility and insights much beyond the organization. So rather than just looking at your first tier of vendors, it can provide you this visibility for multiple tiers of vendors up the supply chain so that you can predict plan and react as efficiently as possible. It through the system, you can actually invite all your vendors and suppliers to be able to part of it. You can share all of this information, you can share product information, you can share inventory information, you can share, share purchase order and increasing visibility makes the entire process significantly more efficient.

As we started talking to customers about supply planning. One of the things quickly emerged that companies don't just share purchase plans or product information or inventory information. A constant and a very frequent requirement these days is being able to track sustainability information, be it, carbon emissions, material transportation, audit trails. And it's an increasingly burdensome task for companies to be able to figure out and chase and to find out how is it that all the products that they are being increasingly asked to report on? From a sustainability perspective of performing. It is a nontrivial task. When you think about it from a component level, it's also a pretty sizeable task for finished goods as well.

So that's why we built the sustainability module in AWS supply chain. And one of the things that it does is it provides a single, simple way of inviting the vendors that you have, that you're sort of interacting with. It allows you to be able to upload the data to be able to capture the request that you need. But it also alerts you to the kinds of things that you may or may not have or the you're missing so that it can fill in those blanks to get that auditorial necessary for just for the reporting requirements. This is proved to be a huge benefit to companies because they feel that they are spending up to 30% of a planner's time on chasing down supply chain requests when they could be spend time actually planning for the future for their customers.

We build this insights module when we launch supply chain. And when we think about the insights module, the insights module is all about trying to figure out and predict, you know, about 60 70% of can from a reporting perspective, the kinds of things that may happen. But if only life were that simple, generally speaking, when we talk to companies, 30% of the time, they are actually asking for very bespoke type of queries that change on a daily basis. If your if your demand plans were not going to be the same. If your supply plans are going to be fluctuating, there needed to be some flexibility that we offered. In addition to the insights that we're creating to allow customers to be able to ask the questions that they needed to do.

This is where Amazon Q in supply chain in supply chain comes in. Now, you can have a conversational system that you can ask specific questions. If you have your data in the data lake, you can query the data lake with very specific things that might change from week to week, or it might even change from day to day to be able to get both graphical representation and rich information that tackles what when and do what if simulations. The what if simulations are super important for companies because they are constantly being asked questions about the health and the status of their supply chain. And with Q this, you no longer have to wait for a report. You no longer have to wait for somebody to be able to run a query that might sort of finish in the next six hours. You now have access to all of that information available. And with just a natural language query interface, you should be able to get the supply chain information that you're craving for.

One of the things when we launched supply chain is that we felt that we had this route in retail and that there would be a lot of companies in retail that would be gravitating to it. We've been pleasantly surprised by companies across manufacturing energy and CPG all interested in this and we're just getting started here in all the applications that I shared today. As I said, one of the common threads of what we are going for is this notion of how do we improve the productivity of the folks who are on the front line or the knowledge workers or supply chain planners or inventory analysts or just making people productive without requiring any contact switch.

I feel like we're just getting started. I want to thank our great panel of leaders and for you all for coming, there's a lot more that we didn't touch today. There's about 50 plus sessions that are going on. And if you haven't caught it, that goes into a lot of depth into each one of these specific things. And you can always stop by the AWS application booth in the Amazon village. And of course, the most important part, there's a couple of parties that are also going on. So if you are, if you're looking for something, this is the place to be. Thank you all. I hope you have a great re invent and I'll see you again sometime soon.