PegaWorld 2026: From Vision to Reality: Citi Wealth Delivers Client Servicing Platform in Record Time

Good afternoon everyone. Thank you for joining us today. We are very excited to welcome Citi here today to speak about the transformational wealth workstation that they created and built to help service their clients in wealth at Citigroup. And with that, I'm going to turn it over to Niket. Ready to go. Thank you Kelly. Um, you know, it's really interesting to be on a stage with my two colleagues. Um, two years ago, I came to my first PegaWorld and I realized I want to be on stage at some point. And, uh, I really wanted to build something that's really meaningful for our clients. Right. So really quick, um, my name is Nikit Kothari. I run wealth operations platform strategy for Citi globally. Pass it to Tomek. Hi, my name is Tomasz Rechtzygiel. I work for Nike. I'm product owner for the workstation that we are going to present today.

Hi everyone. Nice to see you. All of you here. I'm Gopal. I'm the engineering lead of workstation. Work closely with product business and change management with Nike and Tarmac. Perfect. So look, um, Citibank, we're a global bank. Um, you probably heard about it a lot, but I really wanted to dive into why Citi Wealth is really important, but Citi as a brand is probably why we're the biggest, right? Like I always hyping Citi because we're 150 countries. And within those countries, we have so many jurisdictions to even deal with. So when you're thinking about a client perspective, it's really hard to navigate that type of uncertainty all the time. And then when you get into Citi Wealth, and I don't know if you guys really understand how the market works for Citi Wealth. But in December of 2025, we incorporated the US retail bank into the wealth business.

Andy SIG is the head of wealth at Citi, and he's really driven, dynamic changes and has really brought in the thought about clients. Right. How are our clients experience important for us to drive revenue? You know, net new investment assets is something that we really want to hold ourselves to is how do we bring that in? And so that's what this slide really wants to talk to. It's more about when, when you think about wealth. Citibank had a way to think about wealth in terms of ultra high net worth, like Alan. Right. And then you bring it in and we decided, you know, how do we bring in that journey, right, for the client? How do you get our US retail presence, two legged individuals all the way up to our private bank. And those are how we classify our new business, right? So when you think about it, it's two legged individuals all the way up to 25 million net worth or more.

But the point of this platform isn't about who you are. It's about how you transact. Right? That's the key. Um, one more thing is, you know, I say it all the time to Gopal, our engineering lead is think about the client first. You can't make them change the way they do things. Their the way they are, right? And so how do we make it so that's easier for them to interact. I think that's the best. Citibank has 11.3 billion in revenue, 1.3 trillion in balances. But we also have one third of the world's billionaires as our client. Think about that. It's huge. And how do we service those clients? That's the most interesting perspective. Um, we built the wealth workstation. When did we start this journey? In March last year. And we realized that you need a way to connect. And we didn't have that because everything was swivel chair email.

It's like really hard to transact. And then there's no transparency if you think about it. Right? Hey, I want to update my address. Hey, I want to wire my wife a million bucks. Nobody knows where that is in that in that entire transaction. And you laugh. You guys are all laughing because it's true. I don't do that. My wife's like, I need $25 for Zelle. But it's different when you think about the ultra high net worth clientele. And that's why we built the wealth workstation. So with that, I wanted to play a video because I wanted to show what we're solving with Pega. Efficient and high quality client service is very important driver of the success of wealth business. Our teams are doing their best supporting our clients. However, they are constrained by the fragmented legacy infrastructure they need to rely on providing their service.

A typical interaction with a client requires client service officer to search for critical information regarding the client authentication products and accounts. Status of requests empowered to instruct juggling across six different applications, all while talking to the client on the phone. On top of that, currently there is no single place for tracking client's requests. So client service officers need to rely on their own notes to service our clients. While many cases follow the happy path until resolution, clients often come back to us for updates or provide additional information with every new email or call coming in, client service officers need to repeat the process of identifying the customer, finding the request and its status, and providing meaningful update to the client. There are no system notifications, which means officers need to manually track the ongoing requests until they are resolved to inform the clients about their completion.

Our solution to those challenges is Wealth Workstation, a scalable and user friendly platform. It will allow us to centralize data from multiple applications into a single convenient view, move communication off the emails to make it trackable. Decommission many legacy workflow platforms to optimize our cost base and automate manual processes, increasing service efficiency. Let us walk you through the components of workstation and explain how they will improve client servicing. Firstly, workstation offers a single screen with complete client profile required to service clients. It includes products and investments, overview, security data, client demographics, as well as history of requests and transactions. Workstation users will have access to a landing page where all requests are tracked with automatic sourcing of status from downstream applications. All active requests are being monitored. And service officers receive notification when their action is needed. Integrated email and call ingestion automatically recognizes calling client and pulls up authentication form and customer 360 screen for the service officer.

This is followed by automatic categorization and generation of draft requests to operations. Interactive Search Bar allows to easily search for clients, related requests and key information through the workstation. AI assistant offers quickly answers to any process questions. Workstation also intelligently generates email response suggestions in the context of particular requests, enabling Service Officer to save time writing emails to the client. This platform is a game changer for Citi Wealth operations. Users can't wait to start using it. Applause. Come on. All right. At the end of the day, that's really what the platform is really doing. It's really for our clients. Right. So with that, I'd like to pass it over to Gopal, who is my engineering lead and, you know, building the whole thing himself. Thank you Nick. So what you just witnessed in the last three minutes, right, is the true product excellence. And we transform the journey in in record time in six months.

So who don't love the high touch experience? Like let's talk about the example, what Nick presented earlier. We have two main line segments in Citi private banking, ultra high net worth, massive affluent client base. And we have wealth net worth clients. So we want to have a high touch experience elevating the client experience. That's what the journey started about, right. How do I remove the frictions and increase the personalization and enhance the client experience, whatever the transaction that comes all the way in omnichannel, whether it's email or phone or web service. If you want to streamline that end to end orchestration with end to end visibility for any given point of time for our front office and back office teams. So our bankers, associate bankers and operations team prior to this wealth workstation, let's take a step back how they process the transactions in a Bau, in a, in a, in a old state, right, with the client emails coming with the instructions for address maintenance or for payment wires or for alloy forms.

Everything is processed through email, outlook, email, spreadsheets, spreadsheets, and then they do contacts, civil chat between n number of applications for a given Transactions. You have an initial email that ended up with three transactions, for example, the account maintenance or the payment wire and the form journey. They have to do in three four different applications for each transaction. So look at the turnaround time, right. And the number of frictions they have. So how do we solve that inefficiencies by engaging them more and provide the context aware and reduce the turnaround time and increase operational efficiency and thereby elevate the customer experience. So this technology stack evolved in a in our Citi ecosystems. It took from the inspiration of our Pega customer service framework. So Niket and tarmac, we were talking about how do we provide rich experience by taking everything in one place, one consolidated system for all transaction processing, where they can see the end to end visibility at any time.

So we have bankers, we have associate bankers, we have client service officers. Those are the front office team, and they seamlessly have the intake channel and then move to the back end office with operations team for. And then you have a data collection, data remediation with maker checker, and then you have a fulfillment done on the product processor. So we, we improved this with the wealth workstation journey and reduced the swivel chair. And the technology is also part of the old stack, creating a high resiliency scalability issues. So how do we streamline that into one region, one technology ecosystem. That's what the wealth workstation is all about. So with that, you see that wealth workstation. The demo showed us the overall experience on the portal where when user logs in, they can see the client. 363 60 degree view of the financial summary transactions, their assets, their balance, and they would be able to see that seamlessly.

What transactions are pending actions, past due transactions. Everything would be in a single screen with the queue page and queue summary which which took the concept of our work work work queues in dashboards Pega dashboards with a Customer Service. When you log in, you want to see that 360 degree view. The same thing we did with our wealth workstation and with our Citi ecosystem. So the front end is the angular framework. We have a rich set of angular framework, which basically seamlessly have a reusable assets across different client segments. So today we went live with client servicing and operations platform for our CSOs and bankers and for the operations team. Tomorrow it will be like other set of teams and client segments. So this particular platform is evolving and is, is having leaps and bounds of opportunity to integrate seamlessly with lending platform or investment platform or payment tracker, wire transfer platform.

So this is a game changer for us, and this is the state of the art, what we call a state of the art, unified intelligence, next generation platform that we come up with. So the auto ingestion is one key example. When you want the you want to harness the engineering process. So the the transactions, any transaction moving to the system when it is. It is repeatable. If it is repeatable, you want to get a sustainable and predicted outcome at any point of time. So whether the transaction is coming from omnichannel, from email, inbound email or through phone. We want to harness that engineering with repeatable, scalable, sustainable and predictable outcome. And that's what we did with the intake ingestion. We have a next generation NLP which talks to our Citi AI through our Citi gateway that does the NLP with entity analysis, topic analysis, sentiment analysis that auto generate that AI insights for our user.

When they look at the transaction, they would be able to take a decision quickly with human in the loop with Agentic AI. And then we also have IDP Intelligent document processing, which seamlessly extract any key elements from your documents that's coming from the client instruction, the email, which auto populate that on the screens. And we also have a language translations. We do have a Spanish translation where some of our ultra high net worth clients sitting in different geographical regions. We have around 15 teams of client service and service ops teams, which are geographically distributed. And we want to give a high touch experience with the language translation. Either it's English or Spanish. We want to provide the same consistent experience wherever they come through. So with that, we onboarded six major service workflows, which includes the payment, which is the major outcome of our transactions, followed by the account maintenance and followed by the lending and investments.

And we do have exceptions human in the loop for any kind of investigations, fraud investigations or internal instructions inquiries. And we do also have a Salesforce referral process for cross-selling up selling the product and investment ideas for our clients. So those are the game changer. And we build this entirely with our ecosystem. Citi ecosystem, fronted by the angular framework powered by the microservice, moving from monolithic to microservice, modernizing the legacy platform to the next gen platform with the low code, no code of Pega Infinity X platform features. And then we also have the data API abstraction layer at the back end to seamlessly fulfill the transactions to the corresponding product processes, whether it's lending, deposit investment, it goes to the corresponding product processes for fulfillment after the intake journey of the maker and checker. So this provides the end to end visibility transparency for our clients from our client service client servicing journey.

So with that, I will talk about how we do this architecture with the scaled Platform of business Center-out architecture, which I just explained. Client would be coming through email, phone or web. We want to build once and reuse across the channels, right? That's a motive. I don't want to do silos. I don't want to do a repeatable. Every time is harnessing the engineering process. Prospect of having high volume transactions need to be processed with sustainable and predictable outcome at any point of time, and we build this with the headless API with a powerful API version two, which goes through from our presentation front end angular experience layer, and then the back end with all the data lake and the report and analytics platform that seamlessly integrated with this overall ecosystem. So we are in the, we are the first. This is the first application that went on X cloud with native, cloud native with latest and greatest tech, and with Pega Infinity with zero trust architecture.

And we have good observability of elk elastic logging, logging and stash with Kibana reports and Splunk alerts. So this is a game changer for us. And moving to the next, let's look at the birds view of the enterprise component architecture. Right? When we started this we were focusing on the client servicing and operations. Then we thought about how we come up with the model of workflow as a service platform, right? Tomorrow I can seamlessly put any application with Pega situation cake layer, whether it is lending, deposit investment, or client servicing journey. The backend workflow platform will have a Situational Layer Cake layer with built on application stack with seamlessly reuse the enterprise assets of our framework and then integrate the platform and onboard more client functions, more business functions in record time. Right. So this is the game changer. The upstream system, you see that we have an architecture enterprise grade with our RPG, which is our API gateway, the web user channel.

When coming into the system, they go through the RPG engine X, which is a reverse proxy. The request goes to your RPG for the front end with the entitlement service. Once the user is authenticated authorized, they would be able to log into the application. So the upstream will talk is talking about the request flow that started from the web channel going through the PSG, which is a platform secure gateway which is assisted channel. And then we do have nginx to reverse proxy that middleware there where the middleware is. Your orchestration journey with the Pega being the core. We just. With the Center-out architecture principle and you the angler in the angler widget, what do you call experience layer? We have a seamless features that we onboarded recently. As you witnessed in our demo, the page summary, the dashboards, the communication widgets. We recently onboarded the system alerts and notifications.

So when somebody is canceling the payment wire transfer for the ultra high net worth client, I need to be notified immediately. So this is an event driven notification that comes upfront. Whenever there is a critical business event happens outside the layer, they will have a transparency and end to end visibility of any transaction. They've been doing it. So that's one of the critical features we onboarded recently, and I want to name a few. The other features with Pega Pulse, which is a seamless communication channel that you basically interact with anybody in the team, similar to the chat feature, where you want to notify the team and send communication. So we have a rich set of communication features. And also we have a communication emails outlook emails, which is tracked with full traceability of the request and response, both inbound from the client as well as the outbound to the client.

So we manage that to operate within the same Platform journey. And we do have the features of the chatbot. So at any point of time, you might have, you might have witnessed in the demo that when the client service officer talking to the client or servicing the client, they want to look up any knowledge base article or the procedure or policy information. Hey, how do I do this? USD domestic wire or how do I do the foreign wires? So they should be able to seamlessly look up the chatbot, which is integrated through Agent Key with rag model, with our foundation model, which will be talking about in the in the next with further slides from tarmac. So with that, the back end is our downstream systems, where we have a data lake layer, where we have the reporting and analytics for our operations and client service business.

So we also enrich the experience with the entitlement based, role based access security, where any person or any user logs into the system they drive with their business function, what primary group they are part of, what secondary group they are part of. We merge the primary group secondary groups and then paint it in on the single screen. That's where the dashboard you might have witnessed. That shows the queue, page and list of all transactions, like work list work, cuz they would be able to seamlessly do a bulk transfer bulk merge bulk, the escalation bulk update all the bulk case wide actions are actually automated in this same platform. This taken from the customer service framework. Inspiration, all the inspiration. Actually, we got it from the customer service framework, but we did on our Citi ecosystem with this next gen Platform. So with that, I will hand over to my colleague Tomek, who will talk about the metrics of this platform.

Thank you. Thanks for explaining all the technical details and architecture behind our platform. I will talk about the platform a bit from the perspective of product Owner. These numbers are more than just the data. Data. History of successful transformation of our client service organization from a very manual process that was disconnected, as you saw in the video, into one centralized platform and one unified client servicing model. The first metric I want to talk about is number of users that we have. So currently we have over 200 client service officers onboarded, which means we have all North America and Latin America client servicing teams on the platform. We started onboarding back in December with a pilot group of about 30 client service officers, and then we scaled. Till May to onboard all of our 16 client servicing teams. And that process wasn't completely smooth, to be honest.

We had some problems along the way. We had to stop at one point because of performance issues that we encountered some slowness Platform instability. I will come back to that a bit later in the presentation. But as of today, feedback from our client service officers are over are mostly positive. They are happy, especially at this pace of improvements that we are making to the platform every month. Just to give you some idea, we are implementing between 20 to 30 new features every month, new features or improvements to existing features, which is a really good place. And client service officers can see that we are listening to them. Their feedback is getting implemented and they are more and more often surprised how much new functionality there is on the platform. Oftentimes when they come back, they come to me asking for new features. I have to explain to them that this feature actually already exists on the platform.

So from product team perspective, it becomes more and more challenging how to make sure that all of our 200 client service officers keep up to date with what we are adding to the platform. But overall, it's a good problem to have. Next metric. This is my favorite metric because I think it shows best the progress of the we made with our platform. Number of unique client interactions going through the platform. Our current annual run rate stands at about 190,000 of unique client interactions. And by that I mean just the parent level case, if you would like to count. Typical interactions with clients requires a few child cases to be created. So if you would count all the cases that are going through the platform. Currently we are at half a million cases on an annual run rate basis, and we are going to probably double that number by end of the year as we onboard more users and workflows on the platform.

And last metric I want to talk about is our AI assist rate, which currently stands at 55%. And this metric basically means that over half of all of our cases are created automatically by our in-house developed AI solution with client service officer in the loop. For now, we hope that by the end of the year, we will take out the Client Service Officer from the process and fully automate the routing of the of client emails to our operation teams. Um, and together we need to improve our AI to closer to 80% and gain confidence in its accuracy. But we are getting there. We clearly see improvement every month. So we are on our way to achieve that. And really, if you combine the last two metrics, number of cases that will go through the platform at 1 million plus 80% automation through AI, you can get a good idea of how much potential this platform brings to our client service organizations and operations.

Um, we'll be able to deploy additional capacity to grow our business and manage our cost base. And as of now, based on my end client service team's estimates, we are providing about 30% productivity benefit to their organization, which is a very meaningful improvement. Um, however, implementation of this size is never easy. It never comes without its problems. Just to highlight a few challenges that we had. One is definitely the number of teams involved in this project. My product team is dealing with multiple client servicing teams, servicing multiple markets in North America and Latin America, and working with multiple tech teams. All of them are spread across different time zones West Coast, East coast, Latin America, India. So I think it's really good to have product team based somewhere in the middle like Poland. It's really helpful to bridge all the teams together. And then next challenge was to create one unified client servicing model.

Previously, when client service providers were operating in outlook, everyone had their own process how to handle things. It was possible because in outlook there are no limitations. You can do whatever you want. And with Platform like welfare station, everyone needs to follow the same process. So it was very challenging to agree with our all of our client service officers on a single unified model, because every user thinks that their process is unique, their market is unique. But we managed to pull it off at the end of the day. And last problem, I want to talk coming back to the our problems with the rollout of this platform. Um, when we onboarded around 100 users on the platform, we experienced some platform instability and slowness, increased number of production incidents. So we had to actually stop the rollout for a full month and fully dedicate one Sprint one month to stabilizing the platform, making sure we can scale to around 500 users.

Gopal, maybe you want to talk a bit more about. Um, yeah, just to give an overview of how we evolved the performance issues. Right. Um, so this is our baby. The baby was sleeping, crawling, walking and running. Now it's running right? When it is crawling, it all occurs, right? Um, starting from the, any, any kind of enterprise platform at this scale is never be straightforward. It's never been from day one, you get the transaction processed within the SLAs. So we had numerous challenges and it was a great experience, great collaborations with multiple teams. Great learning to. One of the biggest challenges we faced was, um, going, moving to the angular responsive UI with the experience layer had a cried for stability in terms of integration of our components, right? We have our domain APIs, we have APIs. So how do we tune that integrations and the API to respond within the SLA of 3 to 4 seconds?

That's a game changer. We kind of tuned several areas. We looked at comprehensive view of from the client side. Moving to the middleware, the downstream systems. What are the areas that had a key pain points? We noticed, especially the connection pool tuning, increasing the performance in terms of the vertical and horizontal scaling on the infra Jdbc tuning for the DB queries. And the API needs to be more asynchronous and it is already stateless. We had a challenge in tracing the request from the session for ETag where when when the client logs in. When the service officer logs in, initiate the transaction with the session. We want to make sure that subsequent tasks and sessions going in into that system seamlessly going through the same ETag. Otherwise it loses the connection and it will be a disruptive experience. So we had a challenge to maintain that heat tag through our upstream layer of the enterprise gateway of APG, API integration, nginx and reverse proxy.

You have to onboard. We have to onboard all the APIs. Also, we have to support the processing for high payload for emails and attachment size for uploading, listing and viewing the documents. We also have the communication email. Sometimes the client communicate with multiple email chain with going through the same thing. So we need a increased performance tuning on those areas with high availability scalability, resiliency. So how did we solve the resiliency issues when when the transaction is flowing through. How do we process the domestic rules versus non deterministic rules where the agent tick AI. And how do we scale that. So we did we did with our harness the engineering, as I said earlier, repeatable, sustainable, predictable outcome with our AI agent aka automation for NLP. So we don't want to wait for the external system wherever, wherever there are places where you want to autonomously operate and then going this asynchronous, event driven messaging.

We do that integrations, we streamline that. We also refactored our code with engineering, harnessing that particular, uh, sequence to, uh, have a parallel task in the sequence, not in one block of code. So there are a lot of, and also, more importantly, the DB queries we. The platform tried for stability in terms of the sequel's SQL queries because we created our DB views, right. If the team, if the client service officer is part of 15 teams and I want to pull the transactions from. 15 work views from the backend and then render it on the angular on the front end that seamlessly, um, need to retrieve it. The progressive loading with pagination on the front end with angular. So those are the great experiences that this product evolved, um, going through all these layers, comprehensive performance tuning recommendations. Uh, I think that kind of Um, made this journey great success.

Uh, tarmac, over to you. Thank you. Uh, the last slide I want to talk about is, uh, let's look at our roadmap. Okay. It's not working. We move on. Okay. Thanks. Uh, so last year, we, uh, the focus was on building foundation foundations of the platform. Bare minimum needed to allow the platform to the users. Uh, but we also managed to build in some AI features, such as automated email triage with sentiment analysis, understanding what kind of transaction client wants to implement, and also a chatbot that is connected to our knowledge platform so that client service officers can ask any questions related to how to execute their processes. This year we are mostly focusing on the automation. And there are two main themes that we are. Two main workstreams we are doing. One is adding more integrations to the platform. Just to give you one idea, we are integrating, for example, DocuSign.

Today, our client service officers need to prepare a letter of authorization manually in order to save it in PDF, log in to DocuSign application. Upload the document, search for the client. Send it to the client. Wait for notification that client. Send the document back. Download it. Upload our workstation. Next month. All of this is going to be integrated as part of a workflow on our workstation. So we have multiple integrations like that planned for this year. The second workstream is AI related. So we are improving our AI solution to not only be able to recognize what kind of transaction client wants to initiate, that we want to go a step further to automate the process, to be able to extract all data elements needed to initiate each transaction type. And that is going to enable a lot of productivity, productivity benefit not only on the client service side, but also on our for our operation teams.

And obviously this year, we are also heavily focusing on implementing our feedback that we are getting from client service officers. And next year we are mostly going to focus on rolling out this platform globally to our client service teams in EMEA and APAC. We are actually starting with EMEA right now, and I hope that we will launch the pilot in November of this year. And obviously, we'll keep improving our AI solution. I don't think this works will ever end. We have a long way to go to fully automate our assisted channel. And that's all. I will open it up for Q&A unless you have some. I just want to end it with. I don't know how many people in the room saw the Google partnership we did with Citi Sky. Um, it's basically like an, you know, your advisor in your pocket and that entire platform will connect with what we do here, right?

It's through the advisor channel that's going to come through, and it'll still hit the workflow that we need to hit through this platform. Alan left. I was going to say a lot of our platform problems were Pega, but he's not here anymore. So I can't say that. But, you know, integrating into large enterprise is really hard. Um, any of this stuff is really difficult to do. So I'd like to thank Tomak and Gopal, especially for actually really building something that we really haven't had. And I really wanted to thank you guys for that. Thank you.