PegaWorld iNspire 2023: AWS + Pega: Unleashing the power of intelligent automation across your Cloud strategy

Enterprises are digitally transforming their systems and processes to improve revenue growth, operational efficiency, and customer experience, while dealing with further decentralized data and applications. Through our partnership with Pega, AWS is helping clients accelerate their cloud journey globally, empowering citizen developers with rapid development capabilities to deliver a better customer experience.

Regardless of your industry, join us as we discuss combining Pega capabilities with key AWS services, such as Amazon Connect, Textract, Lex, Sagemaker, and others, to create greater value between your Pega and AWS investments. We will also explore using Pega and AWS’s Artificial Intelligence and Machine Learning to improve productivity and reduce errors.

Transcript:

- I am Dan Bell. I am the partner manager for AWS here at Pega, and it's my pleasure to introduce AWS. I've been working with them for some time now, and I think you'll all agree that they bring a lot to the table when it comes to partnership, when it comes to technology, and especially for our customers. So please join me in welcoming Romil Khansaheb, Daya Thakur, and Sharjeel Noor from AWS.

- Good afternoon, everyone. Thank you for joining us after the lunch slot. It's always a tough act to follow. We wanted to get started by getting a sense of who's in the audience today. So I have a few questions. Just raise your hand if this is applicable to you. So first, please tell us if you are a current Pega customer. Excellent. Thank you, thank you for being Pega customers. Hopefully some of you are also running on AWS. How many of you are prospective Pega customers? Okay, so this is a solidly an audience of folks who are currently using Pega. How many builders, how many folks would identify as someone who actually builds solutions? Excellent, thank you. And then how many folks are in a business function, either you decide on solutions to be procured or are consumers of the applications that are built? Okay, so we have a couple of you, excellent. And then finally, just help us understand which industries you represent. How many folks are from healthcare, okay? Financial services and insurance? Excellent, thank you. Public sector? Very good. So it sounds like we have a pretty even mix of folks in the audience, mostly builders, and mostly Pega customers, welcome. My name is Sharjeel Noor. I lead worldwide alliances at AWS for our customer experience partners. These are typically industry-leading partners in their respective Gartner categories like Pega. And today, what we want to do is to talk to you a little bit about what our partnership has been over the past decade, plus give you an insight into how we're working together on generative AI use cases to help you, our customers, be successful and derive maximum value. We'll also go a little deeper into the use cases and talk about the specific technologies, et cetera, that you can utilize from the AWS stack or from the Pega stack to implement solutions. And then hopefully, we'll have enough time in the end for questions. So if you do have questions, please hold them until the end. We're happy to take them then, and we'll also be available after the session in the hallways if you'd like to talk more. Excellent. Our goal today is that when you leave this session, whether you are a builder or a buyer, you have a clear understanding of how to work with both Pega and AWS to build customer use cases that delight your customers, and also help you drive efficiency and ROI in the business, and do that quickly. So a quick overview of the partnership. Pega and AWS have been partners for 11 years. That's a really long time in the tech space. And through this time, we've been building compelling use cases for our customers in a variety of different verticals and segments. I think the key thing to take away from our partnership is that regardless of how you buy, how you implement, and how you use software, Pega and AWS are going to be with you through the journey, right? Whether you work with GSIs to implement solutions, whether you require cross-stack synergies between AWS and the Pega stack, our partnership truly enables a lot of these capabilities for our customers. The other thing to think about is that AWS and Pega are both respective leaders in their space. If we asked when we started, you know, to understand representation from healthcare, financial services and insurance and public sector. These are priority verticals for both of our companies, and we have very deep connections and execution ability in these segments. And so if you are a customer who represents these sections of the market and is envisioning to build new applications with generative AI or intelligent capabilities to automate workflows because there are worker shortages today, just know that both of these companies will stand behind you and help you get to value quickly. So with that, what I want to do is invite Romil Khansaheb, who is the go-to-market and alliance leader for Pega, to talk more about our generative AI strategy, and tell you a little bit more about what the two companies are working on in the next couple of months. Romil, welcome to the stage.

- All right, thanks Sharjeel. So as Sharjeel talked a little bit about the partnership with Pega and AWS, has really seen a lot of momentum in the last, I wanna say, 18 months to two years. And of course, there's a lot happening, there's a lot happening today, than there was happening a year ago, and a lot more happening in the last month that it has reached a really a nice level of partnership, between us and Pega that we are working on new use cases, new applications all the time. So one of the things that I wanna talk to you about is a unique way we are working with Pega for go-to market, and I have another example of how we are working with Pega along the co-innovation side, around the building side. So this is a unique program that we're building with Pega, and as you know, AWS has one of the largest SIGSI ecosystems and so does Pega. So when we are looking at, hey, there is Pega, who is our technology partner, we have an SIGSI ecosystem, and Pega has their own ecosystem, how can we marry the three together for a solution that our customers can really benefit from? So we started on this journey with an initial set of partners, and there's a lot more coming. The thought process here is that when we can have a solution that's fairly unique or customized to an industry that we can take to market with AWS solutions, with the Pega solutions tailored for a certain industry, it would have a lot more value for our customers. And also, we can utilize the go-to market for all three of the partnerships. So one example is the part of it is the PPO solution that we have for the insurance industry with Accenture. This is one of our first solutions out the gate. You should definitely, if you have not already visit the Accenture booth, and you'll be able to see some content around that, I highly encourage that, right, especially, if you're in insurance. The other solution that's out the gate at this point is with TechM. And I see several of the TechM folks in the audience as well, so cheering for you. So AftEAZE is another solution that has been launched and we are continuing to work on go-to market for that. There are two more coming in the pipeline, ones with Cognizant around the financial services industry. And then there is one more around Capgemini with end-to-end customer service. Now, all of these should be very familiar to you from the industry perspective in terms of what we are trying to achieve. The list doesn't stop here. We want to do a lot more with a lot more partners, but this is our first batch of solutions that is coming out the gate does not mean that we won't be doing other solutions. And we'd love to hear from you if you have partners that you're working with that are not on the list today. Okay, there you go. Now these two words are like, in everybody's presentation, not in everybody's talk for the last, I don't know, two days. And you know, when we put some of this content together, I promise you, we did not cheat and had Kareem's speech or Kareem's presentation lined up here. But going back to AWS and the launch of generative AI that we did about, I wanna say a month ago, it was in May. So this was a big launch for AWS, and you know, I'm hoping you have noticed it. If you're living under a rock and you haven't noticed it, it's called Bedrock. You might be under Bedrock. So anyway, you'll get to appreciate my humor. So with Bedrock, we had this announcement and we were working towards our launch partners. And one of the things that's super important to note here is we had a very short list of partners that we worked with around the launch, and of course, one of them was Pega. The reason Pega was part of this launch is because of the market leading position that Pega has from an automation perspective. So this is just an example of how we work together on the launch, and there's a lot more coming around generative AI. And actually, I would love for the voice of somebody at Pega to come and tell us how this would actually materialize for our end customers. So I'm going to call on the stage Christian Goodman from the Pega team. He is the vice president of AI engineering, and he can talk a little bit more about the generative AI solutions together. Come on up.

- Does this work? Yes. So yes, my name is Christian Goodman, I'm vice president of AI and engineering decisioning. I started a couple of months back, very excited to be at Pega, and pretty impressed as to how quick we as a company have moved into this general AI space. I myself, I've been working with AI for good 25 years, I've done my PhD in this area, and it's just fantastic to see all this now becoming real. So as was pointed out, we have been one of two or a few ISVs that had early access to Bedrock, and that's of course, now, my engineering teams are essentially working and evaluating how we're using this technology best. And we are, of course, encouraging our clients and our customers to use and to be open as to which generative AI partner they want to use in the future. And in many cases, it will be AWS. It's also worth noting that due to this good partnership and the relationship we are having, we have, of course, also access too many of these new features and AI features that are coming out. Like, one of the things that we will discuss, I will be in a session tomorrow with Peter Fonda putting a show at 2:00 PM. So if you're interested to learn more, we will be talking about the architecture that's behind basically what Kareem talked about today. And one aspect of that architecture was that model gateway that maybe some of you have seen. And that is basically the ability for our platform or for the customer to choose which sort of models you would like to engage in order to use generative AI. So, and that's of course, you know, Bedrock seems to be really one of those, woo, there's a backdrop that I see that, yeah, seems to be really one of those technologies that is really powerful in this respect. So, hey, that was a very quick overview from my perspective. And I'll leave it there perhaps. Yeah, thank you.

- So as Christian said, you know, this is just the beginning. We are going to work together a lot more around this space. And when we worked with Pega around the launch, we were conceptualizing the vision of use cases, right? And this is, of course, some of these use cases you've been hearing throughout the conference. This is a theme, but when we were putting this together, and when the launch was being worked on, one of the use cases we talked about, for example, is customer engagement. It's a prime use case for generative AI. Now, I tend to break this down in a very different level, but like, for customer engagement, how many times have you called a contact center, and you get passed to somebody that basically ask your name or identity, and then they pass it on to somebody else? You have to do that all over again. You're three times and you're basically done. I'm like, "I've explained to you three times my name, and you still don't know what I'm calling about." So in the context of that, like, how can you think about a customized approach to customer engagement. As an example, right, I was calling a retail bank several weeks ago, and you know, I was just upset that there was some extra charges on my account. So I called them, and I basically was like, "You know, I want you to take these charges off my bank account." And of course the agent was pleasant enough and did work with me, but at the same time, if they had better intelligence, contextual knowledge about me and what business I had at the bank with, they would've been able to say, "Hey, by the way, you have this money lying in the account, and I'd love for you to use that towards a new financial product that we have in the market, whether it's a CD, or whether it's a money market thing or whatever it is, right?" So those are the kind of examples that I think of when we think about customer engagement, generation of content for the agent, for the customer as well, but let's start with the agent to start with. You know, I think Kareem talked yesterday about the automatic generation of development of low-code applications. Yes, I think this is super neat when I see the vision of hey, like, you know, having a supply chain application be generated, based on data sets, based on previous applications, the workflows that have already been used, 80% of the work has been done already in terms of pulling the solution together. However, you know, you still need humans, and you know, it does help to have the humans complete the loop and the process. You know, I talked a little bit about the past interaction summarization, you'll get into an example really quickly, but it also, this is about the contextual information we talk about in the context of the agent, and in instant insight from operational data, like when you have an application, wouldn't it be need for your application to auto-generate reports and be able to tell you that, "Hey, by the way, this particular process is not working, and it needs to be fixed here, and it's to some extent it's been taken care of already?" So those are just some of the use cases that can be enabled between our joint technologies. We wanted to take you for a little bit of a visionary ride here with us, and play along, right? So this is a scenario that, I think, a lot of us can actually relate to. I'll even go out there and ask, how many of you have been in a car accident recently? Yeah, okay. I was in a car accident, like a month and a half ago. I was with a carpool full of kids trying to drop them off in the morning, and you know, a car comes and hit me and I'm stranded there with five kids in the back, fairly noisy. So more on that later. But let's talk about this person, Susan. Susan's an imaginary person, but she's had a full day, and she is driving back home. Let's say she has picked up her child already, and she's on her way back home. This is when she happens to get into a car accident. She's fine, but let's say her car, has IoT-enabled embedded devices on the car, right? And this automatically, triggers a signal over two AWS IoT core services, right? So if the car's enabled with IoT, sends over the signal to AWS IoT core, and that registers that, okay, there's something wrong with the car. We go on, and the insurer has a Pega application that is connected further to the AWS IoT services. Now, this automatically triggers in the application for the insurer that there is an issue with the car, and it's rendered non-operational. Now, this is where something like Pega's decisioning engine, would kick in and say, "Okay, well, there's an auto-generated text that went to Susan that says, 'Hey Susan, are you okay? We see that you have been in a car accident.' Susan says, 'Yes, I'm okay, however, I really wanna get home. Can you help me talk to somebody? I really need to talk to somebody.'" So this is where another set of technologies comes in where the contact center there at the insurer is powered by Amazon Connect, and the seamless integration between Connect and Pega through the unified desktop agent comes into play, and the agent takes over. The case is able to talk, call Susan and ask her, "Hey Susan, I understand you wanna talk to us, you have a problem with the car, how can we help you?" Susan says, "Well, I just need to get home." So they dispatch roadside assistance and roadside assistance is on the way. Meanwhile, there still has 15 minutes left. So the agent asks Susan, "Hey, would you wanna file a claim while you're here waiting for the roadside assistance?" And of course, Susan's like, "That's a great idea." So she goes ahead and says, "Yes," and the Pega application sends over a text that says, "Please send us three pictures of your car that has been damaged, and of course, your driver's license." Another set of technologies comes into play where with the help of intelligent document processing with AWS, the claim is created now in Pega with the information that Susan sent in via the text. Now, after this claim is created, most of the job is done in terms of what needs to be happening that day. So Susan gets picked up, drives home, and is back to where she needs to be. At the tail end of this, the entire conversation, the summary of the interactions, what happened has been summarized with the help of GenAI, and she gets a nice little report when she goes home. So that's a nice little use case where you could see a multiple sets of technologies come into play when you think about both AWS technologies as well as Pega. Going into each of these further, I'm going to basically call on Daya. So Daya is going to walk through each of these sort of use cases and technologies in a little bit deeper scenario. Come on over.

- Thanks, Romil. Hello everyone, I hope all of you're having great time at PegaWorld learning about all the innovations that are being shared in different forums. Now, you heard Susan's story from Romil, and some of the key enabling solutions here, are actually creating a customer experience that all of us want to create for our customers. So my name is Daya Thakur, and I'm a senior partner solutions architect at AWS, and I will walk you through some of the solutions and use cases that you can actually embed in your own enterprise architectures to create delightful customer experiences for your customers. So let's dive into intelligent document processing. So today, many companies use manual processes to extract data from their scan documents, or at the best, they might be using simple OCR technologies, which often requires manual configuration. And if you have worked on OCR, you might know that even if you move a field from top of the form to bottom of the form, you have to redo a lot of work in order to reconfigure the pipeline. While working with Pega, we have created a joint solution, it is called intelligent document processing on Pega, which basically takes care of all these inefficiencies in our document pipeline, and uses machine learning technology to accurately extract text, handwriting forms, or even like some of the insights from your documents. And in order to use this industry-leading machine learning technology, you don't have to have any machine learning experience. You can just start using this solution, and you will be able to process or documents at a big scale. And as you start using this solution, you will notice that more of your document processing transactions, go straight through process. And in certain scenarios, like even ML cannot give you 100% accuracy, right? So in those scenarios, you can use Pega's human-in-the-loop capability to verify the data and send it to your downstream systems. Overall, you will be able to serve your customers faster because now, you're not dependent on the bottleneck that was created by your document processing workflow. So let's take a look at what a typical intelligent document processing pipeline looks like. In the beginning of the pipeline, you have the first step is ingestion. So you can use Pega's interface to upload data into the pipeline, and once you upload your documents, it goes to an S3 bucket, and from there, Pega kicks in the classification step. And basically, what you're doing there is, you are classifying your document into different types. So imagine a use case where you're working on a legal workflow, and there are requirements to maybe attach certain rules or compliances that you need to follow for certain documents, right? And in Pega, you generally attach documents into the cases. And what I have seen in past is like, an analyst goes, and they will select the attachment category saying, it's a claim document or something, right? But using this technology, you will be able to automatically recognize the type of the document, and you can kick in the workflows, which will take care of all the compliance needs for that particular document type. Once you have categorized the document, the next step is to, you can identify which type of information you're looking for. So in order to query a document or if you're trying to extract data from the forms or tables or identification documents, you can use a service called Amazon Textract. Then we also have a service called Amazon Comprehend, which is a natural language processing service, and it lets you find insights in your unstructured data. And in majority of the industries like healthcare or financial services, PII or PHI, it's something that you need to take care of. And this service can actually look at the document and tell you like, these are the PII elements in it or PHI elements in it. And these services also return confidence scores with each extractions. So if you're getting an extraction, which is low confidence, you can always use human-in-the-loop capability to verify those extractions. And once that step has finished, you can move the data over to your downstream business processes. So this was one of the solutions that enabled Susan's delightful customer experience that Romil was talking about. So now let's talk about the next set of solutions. So we also have a joint solution with Pega, which combines Pega's intelligent AI-powered customer service with Amazon Connect. And it focuses on primarily three pillars. The first one is agent productivity. The solution provides 360 degree view of the customer interactions to the agents so that they can focus on building more personalized customer service and customer experience. The second thing that the solution provides is the context to the agent. So at the time of the call, they have the right context, and that reduces the call-handling time. And solution also provides the real-time suggestions to the agents. So when they're talking to the customers, they are being suggested to take the right actions at right times. Now, the second pillar of the solution is delightful interactions for the customers. With the multichannel support, you are able to provide multiple ways for your customers to interact with your brands, and you meet them where your customers are, rather than asking them to come to only a certain fixed number of channels. Your customers can also use proactive self-service features, and the solution also provides digital channels so that if customers are waiting and lot of customers are calling after a certain say, event, they they can also be routed to the digital channel so that they don't have to wait for an agent to become available. The third pillar here in this solution is operational efficiency. Now, with this solution, you have a lot of data available for your agents, and they can look at the unified view of the interactions, and with the unified agent desktop, you're able to see what is happening across that interaction. Then you can also reuse some of the automations that have been built in one channel, and apply them in the other ones. And finally, the supervisors can look at the sentiment of the call and provide proactive coaching or real-time coaching to your agents. So let's take a look at the architecture that enables it. So if you look here, I will not go in details, but one of the core components of this architecture is Pega's AI-powered customer service application, which enables the knowledge management, next best action unified agent desktop. On the call center side, we have services like Amazon Connect, which you can configure in minutes, and it can scale up to serve millions of your customer interactions. Then we also have Amazon Lex. So if you're looking to include conversational capabilities in your customer service applications, you can use Amazon Lex for that. And it's all very easy to get started with because Pega's customer solution application, provides native integration with these services. Let's talk about the third solution that enabled Susan's delightful experience. So if you remember, in the beginning, when Romil was talking about the car breaking down and it communicating with the AWS, this is the solution that is enabling this, right? So with Pega and AWS joint solution, you can bring data from various connected devices, sensors, and other systems and use real-time decisioning and workflow orchestration to take actions on them. You can also monitor asset conditions and predict failures and down times, which can be really expensive at times. Now, let's see some of the services that are enabling this solution. So on the left, we have the services which enable ingestion of the data from the industrial environment, right? So AWS IoT Greengrass, which is open source runtime for the Edge. And then we also have device gateways which manage the interaction with the AWS. And then on the processing side, our main service is AWS IoT Core, which is capable of connecting to billions of devices and processing trillions of transactions without you to manage any of the infrastructure. So you're able to just go there, set up the service, and you're good to go. So for this example, we are moving the data over to Industrial Data Lake from where we are running something called Amazon Lookout for Equipment, which is machine learning service that can look at your industrial or equipment data and find out the anomalies in it. And once you have found out that okay your machine has broken down or is a about to break down, you can move that insight into Pega where you can decide whether it's good to fix the machine right now or should you wait till it breaks so that you can take an action at that point. You can also orchestrate field service and claims workflow with Pega. So, and since we have been talking a lot about GenAI, I thought maybe I should also put my part there and talk a little bit about it. So you heard Romil and Christian talk about some of the work that we are doing with Pega. So the key service that is enabling all those GenAI features is Amazon Bedrock, which we announced last month. Now, Amazon Bedrock is the easiest way to build and scale generative AI applications with foundation models. It is a fully-managed service, so you don't do not have any infrastructure to manage, and you can choose from foundation models or large language models from multiple startups or even from Amazon, and it is accessible via API. So you don't have to set up notebooks and deploy them, and then then access it. You can just call an API, and it's actually just single API that you can call and consume any of these models. So the models that are available right now in Bedrock are AI21's Jurassic-2, then we have cloud A from anthropic, then stable diffusion from stability AI, and Amazon Titan family of models, which includes both text and embeddings model. One of the concerns when when we talk to customers about AIML is that you're going to use my data and maybe train your own models and do something with it, right? So there's a lot of concern around privacy. So with Amazon Bedrock, we are addressing that issue. You can actually privately train or customize the foundation models, using your own organization's data. And when you train foundation models on Bedrock, none of that data leaves your AWS account. It stays there, it is within your VPC, and it's not shared with any of the model providers or with AWS. You can also encrypt your data in transit or at rest using encryption capabilities that are available within AWS. And finally, you can use a lot of familiar tools. If you have been using AWS, you can use your familiar tooling to deploy your generative AI applications in a scalable and reliable fashion and securely. So let's take a look at how we are working together with Pega. So if you're using Pega's GenAI application, which is built with Amazon Bedrock, you will be able to use foundation models from Bedrock just by setting up your credentials. There's nothing else needed in order to access the models. Pega will take care of all the heavy lifting of connecting to the APIs and translating the data back to Pega models. You can also fine-tune your models as I said, talked about in the earlier slide. And Pega also provides human-in-the-loop capabilities, and both of these features are enablers for enterprise use cases that we are seeing in the market. Pega will also provide prompt engineering features that will let you use the foundation models to the fullest extent because most of the time, like, you may have to decide, do you want to really retrain some of the foundation models or can you do that work just by using the intelligent prompting, right? And Pega provides you that capability. And of course, we have built checks and balances across the spectrum of the AI here, and so that the use of the AI is responsible in a way. And so what's next? We have been working with Pega to enable full spectrum of AIML services, and for example, you can use a Amazon SageMaker to build custom models, or you can use pre-trained AI services that we talked about like Textract or Comprehend, if you want to just focus on your key business outcomes. We are also working with them on enabling more pre-trained AI services, and you will see those announcements coming in future. And finally, I would like you to maybe visit Virtusa booth for a demo of a combined Pega AWS Virtusa IDP solution as well as Accenture booth for learning more about the PPO insurance solution. And thank you very much for your time, and I'm very excited to hear how you use some of the innovations that we have shared here, and you build your own delightful customer experiences. So I think we will now take questions if you want to know anything else here. Romil. Okay, so we have a mic here or-

- [Man] I can talk loud.

- Okay.

- [Man] For each of those Amazon services you put out there, they're gonna use customer responsibilities, manage and set up?

- So there are couple of ways we are working on enabling those. So in some scenarios, we are letting you manage them yourselves, and in some cases, it'll be Pega which will be using them under the hood. So we are enabling both types of use cases there.

- [Man] Okay.

- Yep, please

- [Man] No, thank you for the good use cases. We have been AWS customer from last five years, and using a lot of service like Pinpoint and Connect from so on. We are very new to Pega, and we kind of exploring the Pega to Amazon Connect integration, but same time like we had from Pega to Pinpoint, which we're using a lot, there's no out of box integration yet. Is that something, I know we are moving a lot to AI, but you have something somewhere, Pinpoint can be used with Pega?

- So a lot of customers are using Pinpoint with Pega, but I think as you said, probably it'll make sense for us to work with Pega and maybe enable that direct integration. So thank you for sharing that and we will share that feedback with Pega and try to work on that.

- [Man] Thank you.

- Any other questions? Okay, please.

- Hello. Yeah, so today, I think it was well explained about the generative AI which is in the market now. So that's kind of a rule type in Pega. So is it something that has to work with AWS Petro or it's something that can be configured to connect to other cloud providers also?

- So are you talking about Kareem's keynote where he introduced Pega's generative AI or are you talking about some of the stuff that we talked about?

- [Man] Oh no, no, in general, I was talking about generative AI when we use it in Pega, so it's gonna be available, has a developer and everything. So is it expected to work along with AWS Bedrock or it's something can be configured to connect with any cloud platform?

- So we are enabling Pega to use AWS technologies-

- [Man] Okay.

- But they might have some other partners that they work with.

- [Man] Okay. Okay, sounds good.

- Better with an answer hopefully.

- Pega wants to give our customers, hello? Pega wants to give our customers a choice as to what generative AI vendor they wanna go with. So they're gonna be multiple generative AI vendors that you can connect Pega to. AWS being one of them, and hopefully the biggest.

- There you go.

- Any other questions? And we are here in the conference till tomorrow, so you can catch us like we have a meeting room in the innovation hub. So you can reach out to us, and we can definitely, if you want to talk something in detail, we can.

- Hello, my name is from Thank you for the presentation. So I do have two questions, not from the technical perspective, from the functional perspective.

- Okay.

- So let's assume that Susan is already met with an accident, and she couldn't do the texting, okay? Do you have any system that can actually convert speech to text models so that you can actually create a claim?

- Yes, we do have a service called Amazon Polly, which can, actually, you can convert your, like, it enables some of the IVR features as well. So you can use that service. We could have very, well, maybe in our presentation, could have said that Susan receives an automated call which is powered by Amazon Polly, and it talks about various scenarios, right?

- [Man] So it's again integrated with the unified desktop.

- It is not, but you easily integrate with Amazon Connect.

- Mm-hmm.

- So basically, you can embed it into the same architecture, yeah.

- [Man] Okay, the next question is, you guys are using intelligent document program, right, IDP. So within IDP, you have next best action for repairing and production, and then you have the claim center and all this stuff. So from compliance perspective, how you are actually handling the controls, so-

- Can you give me a little bit more context in that question?

- [Man] Okay.

- Sorry.

- [Man] All right. so if a customer is actually raising a claim, so that means he's actually involved with his personal identify information, that is PII. So with the PII information flowing through and you're trying to create a claim, how much controls are you maintaining that whether this person is accurate or not, whether the claim is, whatever he's requesting is accurate or not, and whether it is a fraud? So that kind of a controls, when you're actually dealing with IDP, how do you maintain it?

- Yeah, so with IDP, IDP is primarily focused on the document extraction, right? But there is a layer of controls customers will have to build in order to authenticate the customers, right? And even the transactions that are coming through, we have various ML services that you can use in order to identify your customers. So for like know your customer scenarios or things like that, you can identify fraud in the account onboarding or in the, like, for in case of Amazon Connect, you can even use voice-based authentications, right? And there are models or machine learning services that you can use for identifying frauds in the transactions itself, right? So there are multitudes of services available, but they have to be built in as a extra layer into the solution. What we discussed here was more like out of possible, a very straight through scenario, right?

- [Man] Right, right.

- It was not a actual solution.

- Right, so the thing is like when you deal with IoTs, so IoTs should be able to handle false positives.

- Yeah.

- Yeah, so if this is an accident scenario, okay, so it can be a false positive also.

- Yep.

- She would've hit a tree instead of hitting a car or some human being.

- Right, 100%

- So IoTs should be able to handle false positives as well.

- Yep, and that's why controls and human-in-the-loop and all those things become very important, when you're dealing with any AIML capability, right? Because you do not want to do things which might result or harm people, right? Ultimately, you need to design the safeguards, around all the issues.

- So, but these kind of OTs, when it comes into picture, like, there is a lot of scope for fraud. So the controls we actually develop in document management and also in creating the Pega case management, those are very important when it comes to compliance.

- So yeah, we can maybe, if you have time later, we can talk about some of the controls that can be enabled through AWS, and some that might have to be built in Pega, right?

- [Man] Thank you.

- Thank you. Right, thanks for coming, and thanks very much.

- [Man] Thanks much for your time.