PegaWorld iNspire 2024: Top 10 Hottest Use Cases for Boosting Your ROI with AI from Pega

My name is Andy Bover. I'm senior director of product management for the Pega platform team. I'm joined here by Phil Knudsen, who is a senior manager on our solutions consulting team. For the folks in the back, I think we need to have our slides up on the, uh, on the big screens. Thank you. And, uh, thank you. First, thanks for being PegaWorld. Thanks for making us your first, uh, session after the plenary session. Hopefully we'll deliver the goods.

Now, I will tell you, we're, uh, our our charter here is to cover the ten hottest use cases regarding artificial intelligence. We've got 45 minutes. Obviously, that's not going to afford us a lot of time to dive in as deeply as we would like to, but we think this will serve as a good tasting menu for you to figure out what you want to see while I'm here at PegaWorld. What sessions should I go to? And the good news is there's a lot of choices. So without further ado, I'm going to turn it over to Phil. We're going to talk first. Just kind of provide a brief overview of everything that Pega does from an AI perspective. This is Process AI Voice AI all the cool new generative AI stuff, including the blueprints that we just heard about.

And then we'll dive into specific use cases and to kind of keep things structured, we're going to do this by persona. So we're going to go quick. Please stay with us. If there's any questions you please ask. And without further ado, Phil. Thank you very much Andy. And welcome everyone. Thank you. Like Andy said, thank you for choosing us as your first PegaWorld breakout session.

Uh, and as Andy said, there's a lot of content for us to get through, so I'm going to get straight into it. And as you saw on the keynote, there is a huge amount of focus at the moment on AI and automation. And with that, Expectations are rising. You have your customers who expect immediate resolution to all of their queries. You have your employees. They want to understand their role and be empowered to do their job effectively. And then you have the leadership teams. They of course want you to continuously optimize. And those dreaded four words do more with less.

So everyone is expecting more. And then we have AI. And if there's one thing that's for sure, AI is changing everything. It's rapidly moving. With the rise of generative AI and the commercial release of ChatGPT. Towards the beginning of last year. Right. And as we can see, Andy and I get a lot of questions around AI. But one of those questions that we get is where will AI add the most value?

And it's quite a fair question because when it comes to deploying and using AI responsibly. There's a lot of challenges with that. And you can see some of those challenges listed there. I'll just call it a few lack of resources. You need resources to prepare your data, train your AI models, integrate your AI models, and operationalize those models into your applications. And then we all work in a complex and ever changing environment. Your internal policies are constantly changing. So not only do you need to understand why you need AI as an organization, you also need to understand how do I deploy and make use of this strategically and transparently. And then once you've deployed AI and you start using it, that's not the end of the road.

You have to continuously maintain the efficacy of your AI. And this takes a lot of skills and a lot of time. But luckily Pega is not new to the AI game as you just saw in the keynote stage. We've been doing AI for decades now, all the way back with Customer Decision Hub, where we used it to improve sales, retention, and marketing. We used AI in the customer service space the early days to drive proactive customer service through the use of intelligent virtual assistants. And then of course, Process AI. And it just mentioned that where we used AI to to drive the the efficiency and the effectiveness of your organizations and allow you to make better decisions in real time by driving AI into your business workflows. And then the latest one, Pega, is rapid adoption of generative AI. And of course, generative AI was really good or still really good at generating code.

But it's fair to say that there is a lot of tools in our toolkit, but the way I like to think of it, and Andy likes to think of it, is kind of left brain, right brain, left brain being your analytical brain. Okay, so this is where you have AI capabilities like event processing, predictive and adaptive analytics, machine learning, process mining, natural language processing, all of those analytical AI capabilities to help sense, predict, and make better decisions in real time. And then on the right brain. Right brain. Bring that creative brain. This is where generative AI comes into the mix. Generative AI is really good at summarizing, translating, researching, and generally generating and creating content. Like for example, in the early days, the generation of code. But at Pega we don't actually use things like generative AI to generate code.

Why? We've been generating code for many, many years now, and if you generate code, what you're doing is you're creating a reliance on developers to go and understand that code and actually make it work. Instead, what we do is we use AI, as you can see there, to unlock business transforming outcomes. And we do that by generating visual models that is easily understood, adapted and used by any one of these personas you can see on the screen now. So that's exactly what Andy and I are going to do. We're going to dive through some of the use cases that you can use for each in one of these personas, starting with probably the most important persona customers. All right. Thanks, Bill. Well, just to kind of hit on a few things that Phil shared, you know, first, this whole notion of generating code with AI is something that has really kept a lot of organizations from adopting it.

adopting it, right? The big concern out there has been, hey, I don't want to have a large language model go out there and put me at risk for taking somebody's code that's on on the World Wide Web. And then I'm at risk for being sued. We're not doing that. We are taking your ideas, feeding them into Pega, and generating Pega rules and Pega applications. So it takes that risk away from you. Also, just to kind of highlight that left brain, right brain, we've invested very heavily to make sure that the two sides of those brains work together in unison in a very, very easy to use fashion. So they both have their roles, they both have their places. It's really important that you have all those together in one system.

So to kind of set this up, we're going to talk you out, kind of going back to the keynote with Alan in the plenary session, he talked about how you can use Pega's Center-out model to share, to keep all the logic in one place and have a consistent experience. And that's exactly what we're doing here. Now typically in the back office in the service center, things are going to be ground down by having different silos of data, legacy systems, different processes, kind of a lack of transparency. And by using Pega and our Center-out approach, we can move much closer to self-service. We can start to have those same experiences being shared or being done in some cases by our customers. And ultimately we can end up with having optimized self-service scenario where instead of having a agent or a back office person doing work, we can offload a lot of that to the customer. And typically that can be a win win, right? Because customers can do things on their own without having to wait on a phone, for example. And it offloads work from our back office systems.

Sounds good right? Well, I think we all know what a great customer experience feels like, but we also all know what a really bad customer experience feels like. How many people here show of hands? Have you had to wade through FAQs, PDFs, documentation? Uh, you know, press 0000A gazillion times on an IVR to try to find an answer to a question that should be simple to answer. I mean, we've all we've all experienced that too. So this is where Knowledge Buddy comes in. Knowledge Buddy extends what Pega does with the Center-out approach, and lets us consolidate all the information that's relevant to a given scenario. This could be your instruction manuals, this could be operating procedures, this could be priceless.

This could be any information that might help a customer reset their phone. Uh, fix something else. You know, that eliminates that last step, that last mile, if you will, of action that's required for a person to, uh, complete the work that they're trying to do. So we've coined the phrase somebody else has coined it, but we're using it for the sake of this presentation, but we've called coined the phrase self solve. We take self service one step further by providing that very focused, well curated set of information that customers can use directly. Now, as you're thinking about this, you clearly you're probably thinking, well, gee, this isn't just for customers, right? I want to have these the same, you know, search and synthesis and, you know, instant answers be made available to everybody within my company. And that's exactly how it works. The cool thing is you can use you can have different buddies, or you can kind of have derivatives of the same buddy targeted towards different personas.

A good example of this, I might, you know, in more of a sales or oriented scenario, I might not want to let the buddy when it's talking to, uh, you know, my customers have knowledge of my price list. I might want to make sure that that's something that I want to protect, and I want somebody to call in for have a different channel of communication for that. But it's still using the same corpus of information and still using the same documentation. It's still as informed and still as integrated with the Pega system as everything else. So it's a really, really powerful solution, and it makes self-service that much more meaningful, much more, uh, much easier for for everybody involved. Right. So it's a real win win, not just for, uh, for us as, uh, business process owners. It's a real win for our customers as well. And with that, I'm going to turn it back over to Phil to talk to us about the customer service rep side.

Yes, indeed. So customer service reps, uh, if there is one area where AI is going to have a massive impact, especially generative AI is in the area of customer service reps. And we've listed out three high level use cases. And we'll go dig into some of those now. Now just before we do, Andi just touched on Knowledge Buddy there and the way that you can now chat with your documents. But we all know that scaling, knowledge, expertise, productivity it is hard. Let's not kid ourselves. That is hard. But AI can help make every one of your agents perform like your best agent.

And one of our analysts actually went and looked at the effect that large language models and generative AI would have on contact centers. And what they found after looking at about 5000 agents in a fortune 500 company, they saw a 34% overall improvement in call resolution times for your kind of low and novice skilled users, but also for your more tenured folk and your highly skilled agents. They also saw a 14% overall improvement in call resolution time. So AI has a definite role to play in the world of customer service. Now bringing that back to Pega, this first use case is the use of conversational AI, and how you can use that to guide and automate every interaction that your agents are having with your customers. And for us, that comes in the form of voice AI and messaging AI. Now the way to think of this is it's every agent having this always on copilot sitting there, constantly listening in real time to conversations. And as the conversation is happening, whether that's through voice channels or digital messaging channels, it can suggest content. It can kick off tasks.

And the other side of it is, as it's listening to these conversations, it can identify key information that's being mentioned in the call or the digital messages, and pull that information out and use it to automatically fill in data collection forms. You saw an example of that on the keynote stage. As that call was happening, the agent just had to click and say, yes, this is the right information. So your average handle time on calls are massively reduced using generative AI. And then we all know generative AI has a big power in the power of summarization. So we can use that to start summarizing everything that has happened on a customer account. So when an agent gets a call, rather than digging around multiple applications or even in the same application, fishing around multiple tabs, trying to build up a mental picture of everything that has happened with this customer. Generative AI can do that for you, right? And it can produce a summary of everything that has happened and the most important things to be aware of as that agent gets the call, saving them a huge amount of time.

But it's just not a start up. Wrap up time is pretty much eliminated as I'm doing things with the customer, interacting with the customer, having conversations, filling in forms, resolving cases. All of those interactions can be summarized as well. So if this call is handed off to another agent, that dreaded wait time that you get for another agent to start building up a picture in their head that goes away. Using generative AI. But we all know that customer interactions goes far beyond that initial interaction. A lot of the times. Okay, so how can we now use generative AI to help you keep some of those promises that you might have made on that call? So again, generative AI listens to the conversation and it can pick out where we've said we will do something for the customer, or the customer might send an email for us and then automatically go highlight those follow up tasks, schedule them, and then the agents and the customers can go and follow up on that.

And similarly to suggesting follow on tasks, of course we can use the ability for it to go and generate content, especially in the digital messaging space. So again, listening to the conversation in real time suggest content replies, maybe infuse that with knowledge articles that it's gone out and retrieved. So rather than your agents spending a huge amount of time crafting elaborate replies to digital messages or hunting around for the various knowledge articles, Generative AI can produce that in real time. So I think it's fair to say that using summarization or generative AI power to summarize information is of really, really high value. And Andy and I, just off the back of a napkin, did some number crunching. And we very, very quickly came to some significant savings that you could have. Think about the second the minutes that you could save across all of those activities you see on screen. We used 1000 agents, about $30 per hour, fully loaded rate per month. And we very quickly came to $1 million.

You can scale that up and down according to your organizations, but I'm sure you'll see the same value there for generative AI. So that's generative AI's power to summarize information. There's one more use case that is getting a lot of people excited in the world of customer service, and that is using generative AI to deliver operational insights. So allowing your users to chat with your data using natural language. Slice and dice your data using just natural language. Asking it a question. So this massively reduces those data literacy gaps you see in a lot of organizations. But sometimes asking the right question is the hardest thing to do, right? So why not use generative AI to suggest the questions that you should ask your data.

Generative AI understands the application, understands the data model. It can go in and pull the right dimensions of data out for you, and then look at some suggested replies or suggested questions actually, and you can see a list there of AI suggested questions that users can ask, saving them again a huge amount of time. So this saves, as I just said, a huge amount of time, but also that dependency on it to go and craft and manually develop these reports for the business people so that operations. All right. Thanks, Bill. So just to kind of reiterate a few things that Phil shared. First, it's still kind of the early days on the AI side in terms of return on investment. But what we have seen with all forms of AI is that the ROI has been huge, right? I think we kind of think, well, we're saving a minute here, a minute there.

That really adds up. And we've had a number of scenarios where our clients have seen such high results from doing POCs in their early use of generative AI and other forms of AI. That kind of set them aback. They were like, Holy cow, we need to think about this more because we don't believe the numbers. So, um, the cool thing is there are some surveys. McKinsey just came out with one recently that shows that the organizations are really getting in front of this, are seeing really big returns from it. So, you know, when we're done with PegaWorld, please reach out to your account teams and dig into it further. The second thing was chat with your data is something that's really cool. It's something we use a lot internally and sometimes AI doesn't get it right, but even when it doesn't get it quite right, the insights application is showing you how it does it.

So it's actually kind of a neat teaching tool. So if that's something that you're not, you know, it shows you the queries that it's generating, the types of filters that it's creating. And it makes it easy for you to, you know, see what it's doing and you make some small corrections if you want to. So on to operations. Uh, we're going to go back old school. So far we've been talking mostly about all this cool new stuff. Now we're going to talk about traditional predictive analytics, natural language processing and so forth. And we're going to look at how we can improve operations with AI powered decisioning. And the short answer to the question, what can I do with artificial intelligence to improve business processes?

The answer is yes. And what I mean by that, you can literally inject artificial intelligence into any element of an existing application business process within Pega. Some examples I can use natural language processing to read descriptions, and then do auto case writing, or assist a person that's in charge with that. To do that job better, I can use an out-of-the-box adaptive model Pega has. We're not going to go into details here, but out-of-the-box models that are self-learning, we automatically predict whether or not a case will make its SLA or not. I'm sure your organizations already have models in place to predict fraud, to predict churn, do things like that. Boom. It's really easy to just add that into the. We have a new field type called prediction field type.

You can just add those variables into your application and it's really very easy to do. So definitely check out some of the booths downstairs. A lot of this falls under Process AI, but please check those out as part of your time here. So we'll, you know, we'll kind of go through some of this quickly, but we'll kind of go down to real time. And the end user operator, I want to draw your attention to this little blue box here. And that is something that we call a prediction widget. It's part of Process AI just kind of a little it's a widget, but it's a really important one. And this is sitting on the operator's screen. It's something that is being updated, you know, automatically by default.

It gets updated every time a case changes the stage, but you can make that more or less frequent if you want. And it's showing that, hey, this prediction is showing that the likelihood of this case missing its SLA is about 70%. If it's a valuable case or there's penalties associated with it, I might want to do something different with it. Right? So it gives the agent, the operator that's using this information. Uh, you know, some, you know, better detail in terms of, hey, I might want to treat this one a little differently that they didn't have before. Now, on top of that, if you click on Learn More, it tells you what's behind this. So in this case it's an auto claims case. You can see that, uh, you know the driver.

It's a teenage driver. New license, recently licensed. And there's a pedestrian involved. So probably a little more complex claim scenario that is leading the prediction to say this is probably going to miss this SLA. This is valuable for a couple of reasons. One, it helps increase adoption, right. If you're, you know, as a user, if you're looking at a black box and you don't understand what's behind things, you might be you might have some trepidation about accepting it. Two. It also gives you confidence to make decisions that don't agree with it.

The agent might look at this and he or she could do their job more quickly because they say, oh, yep, that makes sense. That makes sense. I'm going to route this to a specialist team, or I'm going to draw this to a team that has more capacity and take care of this problem. Or they might say, you know, this doesn't represent the full picture. I have superior information and I'm going to override the recommendation, which also has a closed loop effect of making the model itself better over time because it's a self-learning model. So um, so we're going to click I'm sorry, we're going to go from end operator real time all the way back out to 30,000ft and kind of taking a look at your, uh, your business applications from very high altitude. And what we're using for this is Pega Process Mining application. So in an ideal world, everything that we do would follow a nice, neat business processes that we design and create. It doesn't work like that in the real world.

So what we're seeing here, presumably we're not going to go into a whole lot of detail here, but we have a depiction of what the Happy Path process is for, for this workflow. And it's not it's not pulling this in from the case design. It's pulling this in by analyzing all the transactions, all the log data that's coming, that's being generated by your organization, not just from Pega, but from any other system like that. So it's painting this picture from scratch with that kind of data. Now, the things that, uh, you know, in this case looks like most things are going pretty well because most, most of the cases are going through the happy path, but it's also showing you where things aren't going as well as well as you'd expect or as planned. So if there are bottlenecks, if there are reworks, if there are dead ends, if there are things that are happening that you don't expect Process Mining or this type of analysis will help you. I think the benefit for this is important, right? Because if you don't know, you have a problem, if you can't see a problem, you can't fix it. So internally we talk about Process Mining and Process AI as kind of the find it, fix it duo.

And Process Mining is great at finding this. Ma'am, do you have a question? Yes, sir. Taking any technology. It doesn't have to be Pega. It's looking at event logs and, you know, system logs. And then we do some normalization and then it does it uses machine learning to generate those. And it's really cool. You check it out downstairs.

The it provides a really nice financial decoration on the different types of analysis. And it's a very neat technology that's easy to use. So pretty cool stuff. Sure. And with that, I'm turning it back over to Phil to talk about marketing. Great. Thank you very much, Andy. Um, I really love Process AI because, Cause, I mean, we showed the prediction of an SLA breach, but you can actually use that to predict any outcome. Okay.

So think about your own organizations, the value that could have predicting an outcome and operationalizing that AI model so that you can start making better, more proactive decisions and things. Now let's stick with left brain a little bit longer. And I'm going to go with an oldie but a goodie. And that is using AI for hyper personalized customer experiences. You saw T-Mobile speak about this, and if you're not familiar with Customer Decision Hub, we've been doing this for many, many years. This was our first venture into AI for Pega, and we've seen a lot of ROI when it comes to this. Now again, if you're new to Pega and this is your first PegaWorld, what this is, is, well, it's not us doing segment based marketing where you target a group of your customers with some irrelevant offers, hoping they will click on it. This is using AI for targeted 1 to 1 customer engagement. And to put that in perspective.

Here's an example. We have a customer, Marian Sutton, and using AI, we can predict with a 86.3% accuracy that Marian is going to call into a contact center in 15 to 30 days to talk to us about a mortgage product. So we can use next best actions and next best offers to help nudge Marian to that conversation earlier, if need be. And the way we do that at Pega we like to talk about are always on brain. A single brain that sits in between all of your channels and all of your various applications. And it's constantly listening, constantly taking in data about your customer data. Like which emails have they opened up? Did they click on a link in that email? Where did they go on your website?

Did they tweet about something? And it takes all of that information in and it combines it with a little bit more of the customer context. What do we know about this customer? What is their lifetime value? What is the propensity for them to accept a certain offer over another. And then we also combine it with what is our organizational strategic objectives. Which products do we want to choose. And it does this constant balancing act to try and get the right offer or the right next best action. And it generates that in real time, often in less than 100 milliseconds.

And once a customer gets that and they either accept or decline that offer, that goes back into the decision hub. And it updates constantly learning so that the next time they might change channels or come in with a different conversation, they don't get the same offers. Now, as far as ROI is concerned, This is huge and it has been delivering massive value for a lot of customers. I won't go through all of these. You can take a screenshot or take a picture of that. I'll call out a couple NatWest Group. They saw a 10 to 1 return on investment using CDH. We have HSBC 265% increase in revenue per contact and then Vodafone. What's that 40% increase in revenue growth?

These numbers are big. So if you haven't seen Customer Decision Hub in action, I urge you to go to the Innovation Hub and see the magic for real. Sales. Thanks, Phil. You know, actually, I, uh, you know, after after the, uh, the big sessions this morning, I think we could have said Customer Decision Hub gives you customer love and been done with it, but no, it is. You know, that's what they're using. And that that degree of personalization, uh, it really does make a, make a big difference. I mean, we all know when we're being talked to as a segment or, you know, a, you know, dear Mr. blah, blah, blah.

Uh, it's a different maker. So we're going to pick up the pace, we're going to talk about sellers pretty quickly, and we're going to dive right in into AI powered assistance. And this is using Pega GenAI Coach. Now, GenAI Coach. You probably, uh, you know, you might notice looks a little bit like the Knowledge Buddy we saw. And the technology behind the scenes is very similar. We're using a lot of the same techniques. But what's different here is not only are we looking at whatever corpus of data that you want, we're looking at all the active opportunities, all the information you have about your customers to help sellers. Uh, we're also we can also pull in any kind of guidance that you want to have.

So Phil talked earlier about how you can take, you know, take any customer service rep and make them your best service rep. You can do the same thing with marketing internally at Pega. We have guidance from Alan Trefler. We have we can pull in guidance from our better sellers, and we can take that same experience and apply that to you. Make that available to everybody in the organization. New salespeople, experienced ones, ones that have changed jobs, and it doesn't make their jobs a lot easier. And just as a simple example, it's kind of a two fold thing, right? You know, one, it has that ability to provide some synthesis. In this case, we see a sales rep asking Coach, hey, you know, I've got a trip next week.

I'm busy. What's the most important thing that I need to get done? And it responds with, well, there's one key stakeholder that you need to see. It's evaluated all the different opportunities that this salesperson has figured out, which is the most valuable or has the highest likelihood of winning. And it comes back with that response. So something that's very strategic, you know, provides a high level of synthesis. And then it shifts gears and does something very simple. Please send out a meeting invite. Please schedule a follow up call.

You take care of this mundane things that would take me time to do, typing up and figuring out the timing and everything. So it not only provides that expertise, but it also provides that, uh, that, that time savings as well. So, um, with that, like I said, we keep the sellers quick. Back to Phil to talk about developers. Thank you very much, Andy. Yeah. So finally, probably my most favorite persona, uh, often overlooked, um, is developers. Uh, and I'm really pleased to see AI starting to make a real effect for developers, especially for Pega developers. But before we go into the details of that, we all know that business transformation is really hard.

It takes a very, very long time. Lots of discussions and it's expensive. Okay. And if you look at some of the stats that the Project Management Institute has published, 44% of IT projects fail. And I don't think that's a surprise if I ask the same question to everyone in the room. I think I'll get the same number of hands. Probably come up, if not more. And the reason for that is a lack of alignment, of course, between business and objectives and generally business and it. So how can we use AI to help drive these numbers down a bit?

Blueprint GenAI Blueprint you saw the mainstage demo. This is going to be absolutely game changing and the roadmap for this. The early views that we've seen is just going to be unreal. What you can do with GenAI Blueprint. But again you'll see a lot of that if you haven't seen it yet or played around with it. I'm sure by the end of PegaWorld you'll know all about it. But you can basically just put in a simple prompt, tell it what problem is that you want to try and solve? GenAI can then go and pull out from internet best practices and then suggest workflows, Stages, steps, personas, data, models to help solve that problem. But as Kerim also pointed out, you can also use this.

Pega has been doing workflow solutions for about 40 years now, and we have a whole host of best practice templates of how to solve certain problems. You can feed that into Blueprint as well, and a lot of our partners have the same solutions and things. So you can feed that into Blueprint as well and give you that kickstart to help optimize your workflow design. And you can go from idea to prototype in hours, not weeks or months, literally hours. But that doesn't mean you don't need a developer. Developers are still important. There's more complex use cases. They still need to go and implement it. And of course you can just go and import this into App Studio to start that off really, really quickly.

But when it comes to developers, there is a real shortage Shorted in developer skills. So we really need to look at AI and see how we can make our developers more productive. And just to put this in perspective, the US Bureau of Labor Statistics, back in 2021, they said there was a 40 million shortfall in global developer skills. And that will rise to about 85.2 million in 2030. So we definitely need to look at how we can use AI to start, for example, double developer productivity. And for Pega, this is where generative AI plays a big game, and especially GenAI Autopilot. And it comes in two flavors. First, there's the conversational guidance that you get. So when you're in App Studio as a developer, you can start having a conversation with AI, ask it about best practices, what is guardrails, how do I implement a case, etc.

GenAI knows the context of the application that you're working in. It understands and has been fed in all the information from various knowledge articles, documentation sites, training material so that developer doesn't have to go and hunt around for relevant information. They get that immediate productivity boost when starting to develop and starting out with Pega, especially for new developers. So that's one side of it. The other side of it, of course, is how can we use AI once we start developing to actually give him a bit of a productivity boost? And again, this is where we can use generative AI to generate those stages, those steps, data models, even complex data models and references to data objects, etc.. And my personal favorite, a recent one, is actually now generating user interfaces as well. Massive productivity gains to be had there. Again, generative AI knows the application you're in.

It knows which stage on which case you're Urine. It knows what data collection step you are on and it can figure out using the data model, what is the right information to capture in this step and automatically generate that UI for you. Now, of course it won't be perfect from the start, but it will very quickly give you that kickstart and you can just make those final tweaks and things. And if you want some numbers, Capgemini did a great report in 23, where they found Pega was about eight times faster than traditional development and some of the numbers there 50% reduction in workflow design, 60% UI development improvement. The kicker here, this was actually done before all of this new AI innovations that we've seen today or this year. So there's real value there for developers and the use of AI. All right. We're almost done. I can't believe we made it.

Uh, do we get through ten one, two. Three? Yes. We covered ten use cases. I say use cases with Air Force because I know we didn't really have a chance to dive deep dive into many of these, uh, you know, please spend. The good news is there's a ton of additional information that's available before we move on from Blueprint, though, you know, one. Yes, sir. A question. Do we need to be on Pega Cloud?

I mean, how can we generate the Blueprint? Yeah, the Blueprint you just you need a web browser. For the generation of what if you wanted to use. Uh. No, you don't have to know. Oh, I see, so to import the Blueprint and create an application. No. You do. So the question is, do you have to be on Pega Cloud to use Blueprint to import the Blueprint to create new application?

The answer is no. I think there might be a minimum version of 23.1, but we'd have to double check on that for you. One thing about Blueprint that I think is worth pointing out, um, you know, every time you have a new project in a company, there's politics involved. And we've had, you know, Kerim mentioned we've had over 30,000 blueprints created. It takes all the politics out of it, right? It's not it's not Phil's idea. It's not my idea. It's just a neutral thing that created a really good idea based upon all of our inputs. And everybody can contribute to refining that by interacting with it.

So it's a really phenomenal collaboration tool that does kind of neutralize a lot of, uh, the bad behaviors, the bad side of bad side of us as people. So, uh, the good news is, even though we didn't deep dive into everything today, uh, there are a ton of sessions here. I'll leave this up if it's on our app, of course. But, uh, there's a lot to choose from. Uh, both today. These are just the sessions that are AI related today, um, as well as tomorrow. So, uh, with that, of course, the materials for this will be made available after PegaWorld. Uh, if you'd like to get a copy of them before then. Just see Phil or I afterwards, and we'll take your contact information and send it to you.

Um, we have time for questions. Um. Yes, sir. Blueprint for existing applications, if there are any better for you. Good question. Yes. So you can use a Blueprint and layer that into existing applications. And, you know, rather than me kind of describe how that works when you're downstairs and the Innovation Hub. Check that out.

Question over here. There's a lot of. Self-service models for different customers and also third party advisors or whatever. So will there be a point where I can actually read those user experiences also without having to actually go for a project? Yeah. So the question was, can we use generative AI to create the user interfaces as well as different user interfaces, as well as instead of having to do that by code. And the answer is yes. So you know, when you're downstairs, you'll see that today. You know, when you create a blueprint, you can preview a blueprint for an agent for, uh, end customer on the mobile channel, all those different things.

And then if it's missing a channel, you can just ask the blueprint to create that for you. Uh, and then when you create the application, it can auto generate those assets. Uh, there are certainly still some things that we're building out, but it's coming so fast that you're definitely worth checking out. Other questions? Yes, sir. The. Question is, can you use a blueprint to reverse engineer an existing application? And I'll tell you, I think the answer is yes. And I'll say it because if you have documentation about that existing application, whether it's BPMN documentation or just a description, you might typically when you invest that much money in, you know, like a core operations process, it's well documented.

Just copy and paste, put it, put it in a Blueprint and it will it will auto generate those recommended applications for you. And you can fine tune it. So yeah. Yes, sir. Okay. Okay. I didn't follow the whole question. Um, I was going to let Phil answer it, but then because he covers the developer section, but I can speak a little bit to Prediction Studio. So, you know, they're a little bit, uh, how to say this, um, you know, prediction studio.

Uh, let's talk afterwards. I think that's I want to make sure I understand the question and answer it correctly. And, uh, I just want to make sure if there's any other questions, we have time to cover those as well. Sorry about that, but we'll come back to you. Other questions. How do you how do you convince our senior management that it's not the best thing? Okay. Okay. So the question is how do you feed data into the AI?

And then how do you ensure that it doesn't leave your company. Is that the question? Okay. So we're using kind of an industry standard approach called retrieval augmented generation. So the data is going to stay within your system. Uh we pass um some information like the question out to the, to the large language model, Depending upon the application, it's going to return information that's used to generate different things. I think just to get a detailed understanding of that, stop by. There's at least half a dozen GenAI booths downstairs. Stop by and check it out.

The neat thing is, if you're a Pega coder, there's a GenAI rule, and it's really easy to use and you can just I mean, you get going right away with it. So yeah. And just to add a little bit more color there, I don't know if there's any other questions. No. Um, you can mask sensitive data with that connect general rule. So if you do send any data to a public large language model, you can mask any PII data automatically by just checking the box. Right. But I don't know if you saw in Kerim keynote, we have this new gateway service. So if you're really sensitive about sending data outside, you might have your own large language models within your organization.

So now you can start switching and you'll see more of that coming in the next couple of releases, where you can start switching and use your own models internally if need be. Okay, okay, we have time for one more question. Um, after your question, it gives you an opportunity to make different ones for like an hour or two before we do all of that. Yes. You can try to make it on time to get all of the different. I would say prompt engineering is becoming a very, very interesting topic. You can actually depending on how you write your prompts, you can get that outcome. We've played around with it. I've had exactly that where I'm like, okay, this could be one case type.

And then I had to go back to my prompt and just change my prompt a little bit to to tell it exactly what I wanted to do and things like that. Now our internal prompts are becoming a lot more better as well. As time progresses, so you'll start seeing that as well. The other thing I would say is a lot of our best practice templates, if it doesn't go to the internet to get suggestions. The best practice templates is already built in the right way. Okay, so if you have, I don't know, change of address. We've done that at Pega before, so we'll surface that up instead and say, hey, this is how we've done it for many customers. And that's done in a proper case design and things like that. So you can start there and then do the tweaks and things.

But prompts become very, very important. I don't know if you have anything else, Andy. No, I think that covers it. And it's just yeah, it's one of those things just like playing around with ChatGPT a little bit, you know, you got to say, I want one case type and you'd be as specific as you can and you can regenerate too. So if it doesn't do exactly what you want, you can play around with it. All right. Thank you, everybody, for your time and attention. I really appreciate it. Uh, we welcome any feedback.

If there's any questions, Phil and I will stay around afterwards. Uh, thanks for all the great questions. And enjoy the rest of PegaWorld. We'll see you around. Thank you. Thank you very much, everyone.