PegaWorld | 33:43
PegaWorld 2026: Lightning‑Fast Self‑Service: Accelerate Go Live with Workflow‑Driven, Agentic Service
Speed wins, but most self‑service deployments still take 6–12 months. This session shows how to go live faster by using workflow as the foundation for agentic self‑service.
Learn how Pega Blueprint accelerates journey and workflow design while enabling both Pega and third‑party AI agents to operate consistently across channels. We’ll show how shared workflow supports seamless transitions between self‑service and assisted service, ensuring predictable outcomes at speed. Walk away with practical architectures and templates to answer one question: How fast can we deploy this?
Really looking forward to speaking to you today. So my name is Timothy Harfield. I'm senior director of Partner Product Marketing and strategy at Pega. And I'm joined today with Henry Hale. He's group CIO with lab. Today we're going to talk to you about. This lightning fast self-service. So I'm going to set this set the tone, but I'm going to quickly get myself out of the way. Actually I'm not. I'm going to start with a joke. It's the only joke I know. And it has nothing to do with anything that we've been talking about here. So it's kind of like, let's think of it as like a palate cleanser. You go for it. Why are horses good at jumping? That's not a joke, Liz. Come on guys, a punchline. Horses are good at jumping. I do this for the groan. Because they have frogs in their feet.
Okay, who got who got that? You got it. Okay. Explain the joke. Oh. That's right. It's a. Yeah. The anatomy of the horse foot. There's a thing called a frog. So there you go. You learn something, and I've distracted you sufficiently. Now we can go back to lightning fast self-service. So again, I'm Timothy, this is Henry. This agenda. Today we're going to talk about the promise versus reality of self-service. Talk about the three ingredients for successful go live. We're going to meet Kara. Have any of you seen Kara met her already. Liz has a few people have. That's great. That's great. And then we're going to provide some very specific, um um, action items 60 and 90 day action plan. So let's talk about promise versus reality. So the promise when people talk about investing in a genetic service, what they hear is speed. Go live in weeks, automate from day one with no dialog scripting required.
Now all of a sudden we're freed from having to do all that dialog scripting, right? They're looking for predictability. They want defined outcomes, governed AI, no surprises, no costly failures. Coverage. They want every call type automated from launch and not just simple FAQs. It needs to cover every call, type in every scenario and ROI at scale, because we know that even 10% containment delivers substantial return on investment. Gartner predicts we have Gartner in the room here. Gartner predicts agent aka AI will autonomously resolve 80% of common customer service issues without human intervention by 2029, leading to a 30% reduction in operational costs. This all sounds great. Doesn't that sound great, Henry? It does sound fantastic. Sounds fantastic. Where do we where do we sign up? Right. The problem is, reality isn't quite there in a lot of cases. In reality, most implementations take 6 to 12 months.
They're designed around channels rather than outcomes. What's really interesting when you have multiple channels for self-service is because you need to now architect around those individual channels, you lose sight very quickly of the business outcome. You now start architecting for the channel and not architecting for the experience. Brittle bots. We all know about this Legacy chatbots, they require extensive dialog scripting, um, and this kind of thing. Large language model solutions fail when it comes to handling complexity. And more than that, a lot of times when we see these things in the field, there's no proven path, there's no reference architecture. It's, uh, we have agents, we have a promise and we have confidence in our ability to deliver. But where is the credibility? So all of this has to do with, do you have the credibility around what needs to be true in order to deliver on the promise?
Right? You've got the promise. What needs to be true in order to deliver on that promise? It's these things. And more often than not, these things are lacking. So again, we have a situation where we have new technology, which is amazing, but it's the same old problem. And what is the problem? Orchestration. That orchestration problem keeps coming up over and over and over again. And Peg has been in this business of orchestration for the last 45 years. Plus agents, it's another thing that needs to be orchestrated. And unfortunately, that orchestration conversation in the most robust possible way ends up being lacking in a lot of respects. And so let's talk about not what the promise is, what the current state is. Let's talk about the future with that. Henry. Goodness me. You got there pretty quick. I'm up. I wanted to give you time. I've got plenty of time, so I'm going to try and get through a few of these topics pretty quickly at the start here, because the best one in the world you're going to you've heard a lot about pretty much all three of these things.
Blueprint you've been told about probably in every session you've been to. Anything involved in Pega the last 2 or 3 years. Blueprint would have been would have been sort of front and center, all those conversations. But it is a key part of being able to accelerate the delivery in this area. You know, that ability to, to discover opportunity, that ability to really rapidly ideate on that opportunity. And then the ability now to to rapidly implement that by importing that Blueprint into, into the Pega Platform, whether that be native Pega Platform or be that customer service, that entire piece natively now is really, really accelerated. As Timothy said, many times, the workflow is the center of this and we've been really good here today. We haven't got that Center-out slide that you that you'll see in a lot of Pega presentations, but that that concept is really, really key to be able to deliver this functionality very, very quickly, right?
It allows us to greatly simplify our wider architecture. Again, alluding to Timothy's earlier point, we're not building in the channel anymore. We've got that single definition of a workflow imagined by Blueprint, delivered in Pega that is then available to be executed by any number of channels, be that via the call center that we're all that we're all used to be that via face to face in a branch, a genetically via via email, a genetically via voice, which you're going to see coming soon in in 26, but also via good old fashioned digital self-service, you know, an app on my phone, a digital banking website as an example. But what we can now also do squeaky floor here, what I'm what I can also, we also now can do is infuse AI within the process as well, right? To deliver even more value, you know, fix look at things like document management, document introspection, look at summarization, all those, all those great functionalities that you would have seen many, many times this week.
So here, just going through a little bit of an example of kind of what we would do to, to bring this to life. And here we're talking about this in the context of a customer, a customer service implementation. Our Blueprint work would be a precursor to this. You would have heard many times about a high fidelity Blueprint. The more information we can get into that Blueprint as early as possible, the better. And that's going to accelerate that delivery, import that into Pega straight away. We've got that that executable workflow ideally ready in a matter of days. Clearly, we have complications around things like integrations that always add a bit of complexity, add a bit of time to the build. But again, we can get something stood up very, very quickly from a customer service perspective as a platform. Now, the ability to get something up and running quickly is is a more real concept than ever before.
The common data model, as we've talked to in point three here, that allows us to very rapidly integrate, bring integrations into play, to connect to your systems of record, to be able to actually fulfill the journeys that we're looking to expose. And then finally, now the, the setup of the channels is now much quicker than ever before. Very wizard driven. Right. If we're looking to to deploy a voice channel, you set that up exactly the way as you would have done before for a, for a traditional digital messaging channel. All wizard driven, all very, very predictable, very, very quick to deploy, even when we're defining our true genetic channels. Now, again, there's guardrails that help you. There's wizards that help you, guardrails that govern it, but wizards that help you deploy that very, very quickly. And on the on the right hand side here, we see what a what an example timeline could look like aspirational.
And it might look. I think these are really these are real examples. You know, we have taken a number of examples to production even faster than this, you know, looking at a couple of weeks to get that blueprint ready to go get getting that first version of that imported into the system a couple of weeks of foundational customer service work to get the bones of customer service up and running, the ability to, to be able to build that basic customer 360, to be able to integrate the minimum number of data sources to support the journeys that you want to then offer your customers through those channels. We're looking here at a number of weeks of configuration of those journeys. So we're taking those blueprints, getting those to a point where they're absolutely ready to go in all of the channels. We want to expose them through a couple of weeks of UAT.
Then from there on in, we've got hopefully repeated iterations where we start to deliver more and more of those journeys infused where it makes sense with AI exposed via all these different channels, where it makes sense to. So kind of summarization there is we're building it once we're deploying it everywhere. That's what Center-out really gives us is that ability to define a workflow end to end, but without having to put channel specifics in it, without having to build complexity into the channel to be able to surface that process. It's not about it's not just about running it in the channel. It's about being able to run it and access it across multiple channels. I might want to start in one channel, continue in another, maybe resolve in another, or throughout that journey, traverse those channels seamlessly. I think that's a key point there is that whilst we're now in 26, going to have the ability to execute a workflow, we're going to be able to talk to the process that isn't going to make sense for every every single journey that you expose.
So I think you hear a lot about we want to use AI. We want to use agentic. We don't necessarily always know why it has to add value. It has to be a better way of doing it. If Agentic is a if exposing it genetically is adding value, then do it. If it's not, then don't. Right. So use it where you need to use it. Don't use it everywhere. That's I think is a key point to to make. And again, that's where we come to in point two. There you when to use it when not to. I think also, as we say there, you know, the, the deterministic, predictable power of the workflow is something you shouldn't shy away from. The cost of trying to do everything identically, as we've heard many times this week, just isn't worth it. The value you get there has to be business value behind using a genetic functionality.
And then again, we know from a Pega perspective, we have that audit control. We can trust what we're deploying. We have the governance in place to trust it. And finally here we've just got some one statistic here regarding one of Pega customers. Overall costs reduced by 4 million a year, looking at resolving 77% of interactions through through a self-service channel. And this is these numbers are based on before we had the ability to offer a genetic resolution. Right. So you look now at how many more, how that percentage could rise yet further with the ability to to rapidly resolve functionality, sorry, resolve customers needs via a telephone call via digital messaging as an example. So just to reiterate here really of what that of what that really means to us at runtime, right? That single core capability giving us all of that case management that Pega has been been doing so well for so long.
The Workflow Automation aspect of it, any, any AI decisioning, any AI reasoning within that process, we define that all in a single place. Historically, we've always been to always been able to expose that via the CSR desktop and into the back office. As we've seen yesterday in, in some of the keynotes, the ability now to expose that via third party AI. So if you're your users have got Claude, for example, deployed to their desktop, there's the ability now to execute that functionality from there. Also the ability to call out from workflow to there. But now, and obviously digital messaging is something that's been there before. That's that's evolving. We'll see in 26 this the voice agent, the ability to expose that functionality via voice. So to connect that to your telephony platforms. Again, that is a very, very quick thing to do. And then Agentic email again.
So if we still want to service customer's requests via email. Again, we can do that. We can use the functionality of the platform to remove a lot of the heavy lifting from that, from that processing. And I'd like to think that if we've delivered the digital messaging, the voice agent, and actually one thing we haven't got on here is the digital self-service capability. We can start to move people away from those older channels that are less predictable. Emails are going to be inherently tough to handle still, but if we can push them through the other channels, then there's going to be efficiencies to be gained. So I want to talk a little bit about color. I can be selfish here and talk about what we're doing from a lab perspective to hopefully accelerate these these offerings yet further. So we mentioned that in in 26, there's going to be the ability to expose functionality via voice.
What we've brought together with Kara is the ability to, to have that functionality. But to have that in, in pretty much any architecture that you can think of a lot, a lot of clients have committed spends with, with the likes of Google to be able to to use their large language models as an example. Kara is a is a concept, a digital voice assistant. Voice and video assistant that allows you to. To expose Pega functionality connected to your telephony systems of choice. Utilizing your existing large language models, your speech to text models. Your your um, text to speech, uh say your existing telephony Platform you may have an existing agent desktop. We can plug that all in together through Kara. What we also have in there is that seamless escalation to a, to a human agent. So if at any point you've had enough of the authentic conversation and you're not getting the answer you want, you have the ability to escalate to a, to an agent.
But what we've also got in that is the is, is a digital self-service offering as well. That is part of Kara. So what we can't do today, I think, in my view, is that whilst we can expose things via a genetic voice, the next best thing must be a close experience or a similar experience to the customer. You can't say to a customer that you can phone me up to change your address, but everything else, you've got to come into the branch to do or everything else. You've got to speak to a human agent to do. You have to offer a good digital self-service experience underpinning that. That's going to really help you drive those efficiencies in the workflow. So there's a video is going to start playing now. I think I get about a minute or so before before we start hearing any audio of the video.
But what we've put together from a, from a lab perspective is ask Kara. This is our experiential platform that's going to showcase a number of scenarios but allow you to interact with the platform. So this is going to allow us to simulate a customer. Actually it's not it's showing you a real system actually on the screen. We're going to see this person here. Tarek. He's had a car crash. He's going to phone up. He wants to register a claim with his insurance company. And this, this platform is allowing us To show what that phone call looks like, to show what that experience looks like to the user. So here we go. We've got a phone here. He's going to make a phone call. Indeed. He's going to make a phone call. I'm going to be quiet for a second and we'll listen. Hi. And welcome to be.
I see you are calling from a recognized number and device. Tarik, what can I help you with today? Hi. I crashed my car, so I wanted to start a claim. I'm so sorry to hear that, Tarik, I can help with that. Just to be clear, before we proceed, I'll raise a formal motor insurance claim for you, is that correct? Yes, please. Okay, that's all set up for you. Your reference number is Two two. Before we proceed. I just need to authenticate you with the code that was sent to your phone. Do you have that code ready? Yes, that's 9621. Great. Thank you. Now, I just need a few more details about the accident. Can you tell me when and where it happened? Also, if you have it. Do you happen to have a brief description of what happened? Yes, it happened June 2nd, around 9 a.m. in London, and I was driving on the highway at high speed and had to all of a sudden swerve and hit my car into the guardrails.
Okay. Thank you. Regarding the accident, do you happen to have any details of other people or vehicles involved, like their name, insurance provider, car details or email? Yes, there was a third party. His name is Mike Smith. Um, his insurance provider is general insurance, and he was driving a BMW. Got it. Just to finalize these details, do you know if a police report was filed or if there were any witnesses? Yes. I have filed a police report and I did not speak to any witnesses. Thank you. We've now processed your information and will be in touch shortly with the next steps. Your reference number is F-18022. Thanks again for calling and have a good day. So what we can see in the background now is this. This workflow is continued. We're now going to switch to a digital self-service channel. An app embedded in the.
Or an app available on the phone. The process is dictated that actually, what the customer now needs to do is upload some evidence. No, no human agent has had to touch this process yet. This is this has been fully genetically executed up to this point in the mobile app here, which is utilizing Pega API to be able to expose that process in a in a react application. You can see here that Tarek had a pretty hefty hefty car crash. Must have been a pretty solid BMW he drove into. He's going to upload that image, and this is then going to be an example of where we're using AI within the process to analyze that image, understand the damage. So what we're going to see on the left hand side now is the employee reviewing that. You can see here we've got a custom component from a Constellation perspective that is mapping the the damage that has been assessed from that image onto a splat diagram.
Any one of us have ever hired a car. You'll be familiar when you hand it back of that of that diagrammatic view of the car. We've used AI to do to summarize the damage. So that's what you see in that description field there. And at this point, we're keeping the human in the loop to be able to review the findings of the AI. Agree or disagree. And then and then continue the process, all the while keeping keeping our customer in the loop as to what is what is going on. I think the customer, the loop piece is a key piece. I know diverging a little bit from the main theme here today, but the ability to to gain trust in the AI at this point, we're not, we're not continuing the process and assuming that the AI is made the exact, the right choice and made the right assessment, we're allowing at this point the human to, to make that final sort of arbitration, if you will, on, on, on what's going to happen next.
And we can see here that Tarek is uncommonly happy that his claim has been has been approved, and he is super happy. Maybe he needs to get out a little, a little bit more. But what we find with this, with this platform here, is it's a really good way of bringing kind of Center-out to life to show what that really means, to show that single process existing across multiple channels or multiple channels at the same time, the ability to inspect or introspect that process from multiple channels. And we believe with Kara, we have the ability to deploy this stuff quicker than ever before. And so taking advantage of your existing infrastructure, this will work with any telephony provider you can think of. And again, any pretty much any AI model can be plugged into this to support the speech, to text and text to speech, text to speech elements.
Um, so, so that's, that's where we are with Kara. And again, in 2016, you're going to see some native functionality coming out of Pega to allow you to do this. Uh. Time and time again. Sorry. So we've seen kind of what the functionality can do for us. I think we've spoken a little bit about how you can, the important steps you need to take to be able to deliver that functionality. I'm mostly done. What we're going to talk a little bit about now is just a little bit of an action plan as to how you can start to bring this to life yourselves. What considerations do you need to have to bring this to life? I think Timothy's got some microphone problems. I'll carry on. Right. So what can you achieve sensibly in in 60 to 90 days? The first piece here I think is for me is absolutely key.
Right. We've got the Blueprint workshops, right. The key thing for me is understanding where you're going to drive business value by delivering a self-service capability. We spoke at the start of aspirations to deliver everything self-service. You've got to have everything to start. That really isn't the case. We've got we've got to identify what those priority journeys are, and I recommend that you do that with a consideration of, of a balance of, of true business value, but also balance that against complexity of delivery. Okay. You don't want to pick the most complicated journey first, but nor do you want to completely ignore any complexity that may, that may come along later. So balance that with with volume of calls, balance that with the value that the fulfillment of those journeys will bring. So in those first 60 days, we've, we've done that work, we've identified what those priority journeys are.
We've done that CS Foundation, we've got that basic foundation of customer service up and running. We've plugged in those minimal amount of integrations that we need to support a usable customer 360. If the if we are using Pega Customer Service desktop and our customers want to speak to an agent, they need an amount of information to be able to sensibly converse with that customer. We've got our data integrations, not just for that customer service piece. We've got our data integrations ready for the journeys that we're going to deliver. And again, that's where it's important that we pick the right journeys. We don't want journeys that need. I wouldn't recommend anyway for a first release, a journey that needs, you know, ten, 15, 20 integrations. Let's pick something, pick something reasonably simple. Let's group them together by domain. For example, if I'm in a financial services example, maybe I'll pick 2 or 3 processes related to cards.
You know, I might offer the ability to order a new debit card. I might offer the ability to cancel a card, maybe, you know, maybe change the limit of a card as an example, because you're going to have synergies there across the implementation. But as a customer, you're giving a coherent experience of what they can do in that first release. And they're going, ideally, they're in that 60 days, we'll have that first Agentic channel live and deployed or at least ready to go. Phase two, another 30 days later, we've got that all of that UAT testing done. We've got things into production. We've tested all the all the escalation paths across those processes. If it falls, if the process is going to fall out of the agentic kind of channel, as it were. Are we handling that correctly? Is the is the human able to pick that up?
Are they able to resume where the customer got stuck? Are they able to deliver on that promise, as it were? Hopefully by then we can look at additional channels. Right. So we may have started with just digital self-service. Maybe now is the time to activate that fully genetically. Maybe now is the time to offer digital messaging, maybe maybe email, whatever that second channel may be. After then, hopefully we've proven that architecture right. 60 days in, we've got a few journeys, sorry, 90 days in, we've got a few journeys live. We've proven that we can accept case. Case creation through those channels. We're pretty happy with where things are at from that point forward. We now can iteratively, iteratively deliver further journeys, expand those case types, look at any additional channels. Perhaps by this point we might have that appetite to do exactly as we saw in Kara.
There, the ability to expose effectively an avatar to, to, to be that kind of to be that almost human like integration into the system. Maybe our customers aren't quite ready for that yet, so maybe, maybe we won't. I think the important piece for me on the Beyond 90 Days piece is around what the implication is on this architecture for your wider delivery organization. I think from a Pega standpoint, we can always be pretty agile with the way we deliver these things. We don't need a cast of thousands to deliver this capability, but if we're putting Pega right at the center of all of this, then we can greatly simplify our engineering efforts elsewhere. We no longer need a front end team to build our our mobile app, for example, that need to know lots about the business process. They just need to worry about presentation.
The API allows us to surface this functionality through those multiple channels. They just need to care about what it looks like, not what it does. So no longer do you need that that business knowledge in those front end teams. Historically, we would have had APIs from the front end calling APIs. In Pega, you'd probably have an API team delivering a suite of APIs to to support that. Again, with Pega being at the center of all of this, using the APIs that come out of the box, you don't need that team anymore. So I think there's there are some considerations on the wider engineering teams that you can, that you can consider here as well that bring even further savings to the equation. And from our perspective, from a lab perspective, I just want to bring one of these examples to life here. Now this this example here with CMC markets is a client that has actually been in a long term user of Pega.
They've been using Pega for, goodness me, 15, 16 years now, maybe even more. Or so. This example. Here was an example that is about self service, but it's also about modernization. They had a typical or a very typical architecture. They had a a custom front end delivered in react. They have a cast of thousands delivering APIs to that front end. And they had Pega with a bunch of custom APIs being the workflow provider at the back end. They offer a huge suite of products across a heavily regulated industry across a large number of jurisdictions. They were maintaining onboarding workflows across multiple technologies to support these different products that they were looking to sell. We wanted to show them how, by putting Pega right at the center of that, they could reimagine their entire architecture. We were really fortunate that they were a client that was willing to take that step.
Lots of clients want to talk about making that step. Very few will want to do that as quickly as CMC did. And the example here is that, say product onboarding. Historically, they had very long deployment cycles. They would do releases every 3 or 4 months on their Pega estate. So they were taking a long time to release value. The process was very fragmented. They had, as I said, multiple systems that were handling onboarding depending on the product, depending on the jurisdiction and depending on the channel you were accessing it through. Again, there was a different solution. So we drove them through a, through a Blueprint process to, to design onboarding. We were fortunate for this first example. It was a very simple onboarding process, but we really wanted to have the ability to prove what a new architecture could could deliver to them. So we delivered Pega delivered the Blueprint.
Sorry, we delivered this channel agnostic workflow right at the center because they wanted it to be exposed via digital messaging. We knew they wanted a contact center channel. And really importantly for them was the digital, the digital self-service piece. So from day one, they had that self-service capability. They had the ability to expose it via an assisted channel. So via the call center. This was done prior to to the Agentic voice channel being available. But because we've delivered it in Constellation, it is a genetic ready, right? It is. We have the ability now, and that will be the next step for this client to deploy the case creation fully, authentically. Within seven weeks, we were live with this. So for them, that was quicker than anything they'd ever delivered before. And that was but this, this seven weeks we've got to remember isn't just the work in Pega.
That was the ability for them in their front end to implement that custom design system as well. And what that's led to now is that we now have a foundation to put all of their product in product onboarding across all of their jurisdictions into Pega, across all of the channels and say there's a genetic channels will be the next ones. It's a brand new way of working for them, and it's a greatly simplified architecture. say a few a few numbers that just underpin that, right? 42 days from from concept to production. Right. Once they once they decided they want to do this and they wanted to see what a new architecture could look like, we had that live in seven weeks, three channels from one shared workflow. And you'd have to trust me that it is a genetic RUDI, a 30% reduction in onboarding and onboarding time.
And I think that's a very conservative number. And I know that as soon as we start to on board, the further, more complex journeys, then that number will, will only go, will only go one way. And then from a maintenance effort, we're not having to maintain any logic in a third party platform. We're not maintaining API's specific to the product. So again, we look there again, I think it's a fairly conservative 6,060% reduction in maintenance. If they want to make a change. Now, typically they make that change in one place only. So greatly simplifies testing, greatly simplifies deployment, accelerates the whole process. I think for now that's it. So thank. You. Question. But do you have a question? I have a question. I have a question for you. Henry you're working now. So at the outset we talked about, uh, part of the promise in order to fulfill that promise, it assumes that you have not promised to solve the problem, but a promise based on the fact that you've already solved the problem.
I think, I think, uh, Blueprint helps take us a long way because now you can actually see the final app before, before, um, building it, but you've also been a lighthouse partner of Pega's. And as part of that, you have developed a reference architecture. Can you talk a little bit about what that means to be a lighthouse partner and what that means for your ability to execute on this kind of work? I think being a lighthouse partner is a good position for us to be in. I think it's a it's a fortunate position, but it's a position we've earned. I think what it does allow us to do is, is allow us to work very closely with Pega product teams to kind of set the direction of travel for the product, to really poke and prod around the edges of the product whilst it's in development, to make sure that when it is ready to go to market, it is going to deliver on what we want it to do.
And that for us has been it's been a fantastic experience really, to get early sight, to be able to input into that functionality, and that's really what's going to allow us to deliver this. And that's also what's led us down the path of, of, of our product from, from a career perspective is to understand where Pega's direction is going, but also for us to show actually there's other ways as well, right, to be able to, to actually broaden the scope of where, of where we can take this functionality. That's great. Great. Thank you very much indeed.
Related Resource
Product
App design, revolutionizedOptimize workflow design, fast, with the power of Pega Blueprint®. Set your vision and see your workflow generated on the spot.
Reimagine your processes and turn any workflow into a build-ready application with confidence.
Visualize customer journeys and engagement strategies across all touchpoints and activate them.