Pular para o conteúdo principal

We'd prefer it if you saw us at our best.

Pega.com is not optimized for Internet Explorer. For the optimal experience, please use:

Close Deprecation Notice

PegaWorld | 40:05

PegaWorld 2026: The Next Era of Banking: How Mizuho Bank is Pioneering AI-driven Automation

Mizuho Bank is proving tradition and innovation can coexist. By digitizing paper-heavy processes and deploying predictive AI agents, they are redefining customer engagement and operational efficiency. Now the bank is taking a bold leap, deploying predictive, enterprise-grade AI agents to revolutionize engagement and efficiency, proving that even regulated industries can lead the future of intelligent automation.

Hello. Good afternoon, and thank you for joining us today. In this session, we will share a Mizuho Bank initiative on Pega and their journey toward AI agent and will have our time Q&A at the end. So let me introduce Mr. Arai and Mr. Maeda from Mizuho Bank. Over to you. Good afternoon everyone. Thank you for coming to our session. My name is Toshiaki Arai at Mizuho Bank in Japan. First I would like to thank Pega. Thank you very much for inviting us to PegaWorld and giving us this opportunity to speak regarding regarding the relationship between Pega, Pega and Mizo Group. We have been partnering across various regions, entities and business areas for several years. In particular, our business area retail Banking area. Pega has provided us support over the past about five years. As a result, we have achieved a major transformation in both customer experience and business process, including back end operations.

We need to build up furthermore. So we think that AI becomes essential. But miso is a bank. We are required required a high level of accuracy. We think that AI should not be implemented without proper controls. In this point, Pega is very reliable. We should put AI in the right place inside the clear Pega workflow. In this way, we can control AI and improve customer experience, productivity, and work quality. At the same time, we believe this approach will enable us to achieve further evolution. We would like to introduce this point today. For more detailed explanation. Is my colleague, Mr. Maeda. He explains this session. He's a key man in this business area. In my company. Mr. Maantic, please. Thank you. My name is Kousuke Maeda. I will talk about today's agenda. First, I will give a short overview of Mizuho Bank and the role of our retail business.

And next I will talk about our work with Pega. I will explain the issues we had before using Pega for account opening and change requests that we introduced and the fact that we gained and. Here change request means such as. Such as changing a customer's name or address or handling Lost and Found or reissue Cashcard. In the second half I will talk about the future we want to create with AI. I will also explain the risk of AI and Heather Pega Platform can help us control those risks. Let me start with Mizuho Bank. And Mizuho Bank is a global company of Mizuho Financial Group. We provide many kind of financial services to customers in Japan and outside of Japan. Mizuho Financial Group, Inc. is the parent company of Mizuho Bank and the group. And there are also companies such as Mizuho Trust and Banking and Mutual Security as a group.

And we serve both individual customers and corporate customers. Our service includes deposits, runs and payments, asset management and business succession, corporate finance, M&A and capital market services. Miso has a strong base in Japan and also has a global network to give more value from the customer and from the customer's point of view. Miso has a company based structure. In this structure, each company has clear responsibility and authority for each customer, group or business area such as retail, corporate, global, business, and asset management. This helps each company respond quickly to market change and customer needs, and it also helps us combine the power of the bank, the trust bank and the security company so that we can provide better solutions to customer issues. Next, I will explain our retail business. And the retail business is not only one business area, it is an important base that supports the whole group.

The base of customers is very important. The deposits from these customers are also very important. They support other areas such as asset management for individual customers and solution for corporate customers. In this way, retail business supports stable income and future growth for the group in this way represents a stable income and future growth for the group. A Stuart Daly business. Our goal is to become the financial partner most chosen by individual customers. We want to be easy to use and safe to trust. To reach this goal, our most important theme is customer convenience. We combine face to face and service branches with digital and remote channels. We also use group company service when needed. By doing this, we can meet many customer needs in one place. Our team focused especially on improving the customer experience by seeking better customer convenience. Customer contact points have become wider than before.

In the past, the main contact points were the branch. Now we also have remote channels such as contact centers and digital channels such as the web and mobile apps for digital channels. We improve the user interface and user experience for branches. We use the strengths of face to face service to and give careful service to each customer. At the same time, we want to connect data across channels. For example, we want to share application information and inquiry information across channels. We are stressed out customer two step process in one channel and continue it in another channel. Our aim is to give a seamless customer experience and build long term relationships with customers. Now I will introduce our work with Pega Group and Pega have been no sorry and have been partners for more than ten years. And this history is not only about introducing Pega services.

It is also a history of using them, improving them, and expanding them. The teams and areas that use Pega worldwide, they include Japan and other countries. Pega has become one of the important basis that support higher quality and more standard work across the group. Example includes trade, finance and internal electronic approval and routing system. It through this project, and this group has built strong and strong know how on Pega services. This knowhow covers internal work such as applications, review, receipt of requests and progress control in the work we'll talk about today. We have taken this one step further. We use Pega not only for internal work, but also for customer facing work in remote channels, especially account opening and change requests. Let me explain. We have been before at that time, opening and change requests in remote channels still needed many manual tasks, and this creates a burden for both customers and employees.

There are three main issues. The first issue was that the customer system and the internal system were separate. Because of this, accepted cases were printed on paper once a day and the next work was done from the paper since it was once a day. It took at least two business days from application to completion. Second issue was paper based manual work. It was hard to count cases and control progress. It was also hard to see which case should come first and which case had stopped. Because of this and the work burden and the management burden were large. We also need many outside workers. The third issue was paper mill and customers could apply by mail. We saw we had to receive the mail, check it and handle missing items and input data by hand. Long processing time caused more customer stress, more inquiries, and more work.

This made the worksite even busier. Then Covid 19 came and we accept more face to face applications. We became concerned that the old way of working could not handle the future value. To solve these issues, we need to review the whole work flow and build a new system. We had two important points. First, we wanted to do manage the whole process in one system. From the time customers start an application on a smartphone or web screen to the time our internal work is complete. Second, we wanted to make manual work automatic as much as possible, such as registration work and document creation. Three. So these are also important because of Covid 19 and the state of our system. At that time, we needed a fast release. So we did not want to build our own workflow system from zero. We looked for an existing platform with strong results and high trust level.

We learned that another department was already checking the Pega Platform. This gave us a clear image of what Pega could do. After that, we carried out a proof of concept with Pega. we can have that. We could recreate the model. We wanted to inverse account opening and change requests. One reason we valued Pega was it Center-out business architecture. We can connect in a flexible way with many. System by API. This can improve not only work speed but also long term service level. Of course, not all internal systems can connect by API, but even then we can combine other methods such as LPA by using API, LPA, and other methods together. We can manage all this work in one place on Pega. We also care about maintenance after release. We wanted to use standard functions as much as possible. Pega gives us practical advice on this point.

Also with the group already had a Pega expert team. We could get help from that team when needed. This gave us confidence. We introduced the system step by step. First, we manage the process from application to completion in one workflow. Then we implement the workflow and added automation with API and RPA. This helped us reduce rework and and and reduce the system to real work. All three issues were improved by Pega. For the first issue, the customer system and the internal system were separate and the process took more than two business days. After moving to one process on Pega and same day, completion became possible. At the first scan issue, we had paper based manual work and many outside workers with Pega. We moved to a paperless process and managed all cards in one place. We also use API and API to automate work. As a result, productivity improved and we reduced outside workers by two thirds for that issue.

We had a work button from applications. With Pega. The application screen became easier to understand and our process capacity improved as a result. We are able to stop receiving application by mail and remote mail costs, but the real value was not only lower cost and lower work hours. With a workflow, where you can see where cars are stopped and where the bottleneck is. This helps us find the next point to improve. Also, because the process is standard, the way we check cases has less variation. This helps reduce mistakes and rework, whereas the workflow has become the best for improving both productivity and quality. Next, I will talk about UX and AI. At Mizuho Bank we use AI in many kinds of works. Our basic idea is this work and work that AI can do should be done by AI, and people should focus on creating value that only people can create.

I will introduce, I will introduce three examples. The first example is generative AI for internet search and document creation. We use it to search internet rules, procedures and manuals. We also use it to help create customer proposal materials and internal approval documents. The second example is generative AI in the contact center. In a contact center, we need stable service quality and lower burden on operators. After customer identity check is finished and customer information and past history are shown at once on the operator's screen. The composition is changed into text in real time. Related. Manual and Q&A are also shown automatically based on the composition. This reduces waiting time during the call. The composition is also summarized automatically, so after the code, we can show that in this way an operator can focus more on the customer. The third example is visual LLM for financial work.

Financial work has many special terms and complex rules and rules in the rules and long standing work practices. A general AI model is not always enough in finance. Wrong understanding and unclear answers are not allowed. So we are wrong. We are working on visual LLM. The aim is to understand financial knowledge. Special terms, road rules and internal data such as procedures and manuals and to support high level judgment in real work. Mizuho Bank. LLM is Blend with three layers. The first is a financial specialized LLM for work such as inquiries and document creation. The second is a task specific LLM for Azure such as procedure advice, credit decision support and approval document drafts. The third is a collaborative expert LLM, where several tasks and specific models work together to support complex and complex engagement in the finance specialty. LLM practice tests have already reached the internal level in areas such as deposits and loans, foreign exchange and financial analysis.

We are now studying how to make it more advanced and use it in real work. Based on this AI example, but we want to create with AI in account opening and change requests. Answer. Answer. Answer is a customer centered application experience. Today, many application methods use fixed form. These forms often ask the same question to every customer. They do not always consider the customer's situation or transaction details. As a result, customer, say may have to enter more information than needed. When input is bad and wrong, input increases. Some customers prefer the end. Inquiries also increase in the end. Internal office work increases. We want to change this by using AI. We want to understand the customers purpose and ask only for the information that is really needed. For example, if a customer says, I moved, I can get the customer to the address change process.

If I can, if a customer says my phone number changed, I can get the customer to the phone number, change process and related services when needed. This creates a more personal and more natural experience. It also helps the back office if we reduce wrong input and we can relate. Sorry, we can reduce related checks and returns and rework. So I can improve both front side convenience and back office quality. Sorry business handled very important things and customer assets, customer trust and friends that support customer needs and company activities. And so finance needs a very high level of accuracy and trust. One error, one mistake or one right decision can cause loss. It can also have a large effect on a customer's life or a company's business. And financial institutions must follow laws, regulations and internal rules in a strict way. Mistakes can reduce customer trust and diminish social trust.

For this reason, finance needs not only speed and efficiency, but also accuracy, transparency, repeatability, and clear responsibility. So should we put AI into every part of the application process? Our answer is no. It is growing very fast, but the risks of a wrong answer such as acceleration is not there. The frameworks that help us use AI safely is the Pega workflow. Pega Workflow Automation has set in detail where AI is used and where AI is not used for account opening and change requests. Our step from application to completion within the workflow. So for each task, we can decide whether to use AI or not. This means we can control the safe. Sorry. And this means we can control the scope of AI in finance, where it is practical to keep the change area small, test it step by step and then expand it. Today part of checking and visualization work is already automatic.

In the future, AI can support a wider area and further human judgment remains in the place where it is needed. Also, when we can limit the scope of AI, it is easier to measure the effect. We can compare input error rate, return rate, processing time and inquiry rate, and before and after introduction. This helps us find the next improvement point. Next, let me share an example workflow and AI use. The key point is while sharing and it handles customer response and understanding of customer needs, the workflow handles and safe and correct processing. For example, a customer says, my commute time became shorter and from this context, I may think that the customer has moved and may need an address change. And I then ask, do you need to change your address? If the customer says yes, I want to change my address, I create an address change case and issue case ID.

From this point, the workflow decides the next step. For example, the next step may be address input. The workflow already defines that postal code and already and oh sorry. And address needed based on. And based on this AI set. Please enter your postal code and address and that's. The workflow shows the next item check, such as whether the phone number also needs to change. Asks do I also want to change your phone number? If the customer says yes, the workflow moves to the phone number input step. If the customer says no, the workflow moves to final confirmation and then complete the receipt. In this way, AI gives natural composition. The workflow gives the right process. The right order required checks and expression handling. AI does not move forward by making free decision. The workflow defines the next question and the choice that arose. This helps reduce the risks of wrong guidance.

Sorry, let me organize Jeff Frantz. It has large benefits and it can improve customer convenience and work productivity. At the same time, it has risks when risk is high definition. At the same time, sorry. And AI may create answer that seems right but is not correct. In finance, we cannot ignore the risk. That is why we believe AI must be controlled by the workflow. The workflow can define the order of tasks, the decision rules, and execution process. It can also place the AI use points clearly inside the workflow. For example, it can be used for input, help, summary and natural wealth for the next question. At the same time, final confirmation and registration can be controlled by the workflow and when needed by human approval. Rabobank and customer trust is the best of the business. We can't allow serious mistakes. So the workflow is a guide for sale and AI use.

It is also an engine for continuous improvement. AI is not something we introduce once and then finish. We need to keep adjusting it as well. Rules and customer behavior changes with Pega. We can run a cycle introduction, monitoring, and improvement in a safe and steady way. Let me close by summarizing the main points. In the past, our work was mainly paper based and manual. Because of this, the work volume was heavy and the customer experience was not convenient enough. While we were looking for a solution, we met Pega together with Pega. We introduced the Pega Platform. It enabled us to manage the whole process from application to completion in one connected flow. Through this, we improve the speed, quality and productivity in important known face to face customer work such as account opening and change requests. Now we are moving to the next stage. We want to use AI to create a more customer centered response.

AI has strong benefits, but it also has risks. In finance, we need frameworks that can control AI. That framework is Pega Platform. By using the workflow, we can place AI in the right work and in the right way. We believe this will help us achieve both customer centered experience and self correct work in finance. Going forward, we want to grow the trust based relationship that we want build with Pega. We want to contribute to group customers and to society. This concludes my presentation. Thank you very much for listening. Thank you very much for coming to this session. I'm Ayano from Pega, Japan. I will I will do my best to support this Q&A. So if there's anything that you can ask for the the Mizuho Bank, the presentation. My name is. Paula. I'm from Pega. First of all, Chicanos IT PegaWorld ni go Sanka. Itadaki.

Hontoni. Arigato. Gozaimasu. I hope this was understandable. Thank you very much. My question would be. It looks like your AI is not the Pega kind of the eye that we deliver with all our tools in Pega GenAI. What is your reason for kind of using a different. Are you using actually Pega GenAI features here? Just want to make sure that your question is like the AI is not the Pega. It looks like, at least from the slides, that the Pega AI GenAI features are not used. It is kind of different product, a different set of technology. Okay. So why. Pega. Open API a. AI agent. Okay. So now. We are creating the AI using the OpenAI. And then, um, the, the. Also they are. Using AI, but actually we had a kind of POC with, uh, this is my comment. Uh, we also had a POC with Mizuho Bank.

And then at that time we used the GenAI, I mean the AI in the Pega. So we work on that together. But currently they are the Mizuho Bank is creating the AI external AI. From Infosys, I have a question. As we were seeing what you were showing in the slide, it looks like you have used a lot of different LMS. So which LLM models that you have explored in order to integrate that with the Pega workflow? I'm so sorry I couldn't hear. I'll repeat which different LM models you have integrated within the Pega workflow. As of now we have Gemini. We have open. AI. Cloud. Open. AI. OK. Okay, so what was the efficiency that you have received with open AI in the workflow? Sorry. I want to make sure that your question again, your question is like how workflow is working for that with AI or the different question.

If I understand you have used open AI with the workflow. So if I understand we have used you have used open AI as part of the workflow. So what type of business efficiencies that you see as part of the workflow after using open AI. Can you No. Candy or no risk. Credit portfolio. No. Unexpected loss. No. No. No. Sorry. In short, maybe I don't. I'm not translate perfectly, but it's not released yet. And then they are planning to use that for the financial, uh, area kind of wide widely. And then um. And also they want to manage the risks. So we cannot provide any outcomes for now, but we are trying to make it efficient and then working on that very hard. Okay. Thanks. Thank you. Thank you. I'll go. I'll try to use Google Translate. Might be easier. So the Japanese banks are still quite paper heavy.

Um Mizuho tried to go out of paper heavy business process and try to do more automation, more digitization. Um, how far is Mizuho ahead in terms of other banks in Japan in your understanding? Of. I know. Your. Mother. Sent you. To Dinesh Karanam. So. You took. This. Unfortunately, it's only 20% or 30%, and then we. We. We will get to that goal as as soon as possible and then won't. And won't make it Possible as much as possible. Yeah. Yeah. Oh. That's good. Thank you. It's actually very good compared to where we are as well in the digitization. So I think if you guys are taking so much initiative, even 1,030% is really fantastic. So keep up the good work. Thank you so much. Thank you. Thank you. So again, thank you so much for coming for this session. Thank you very much. Thank you very much.

Recurso relacionado

Produto

Revolucionando o design de aplicativos

Otimize seus fluxos de trabalho com rapidez e com todo o poder do Pega Blueprint®. Defina sua visão e veja seu fluxo de trabalho ser criado instantaneamente.

Pronto para criar um Blueprint?
Escolha o mecanismo de reinvenção para suas necessidades.
Para fluxos de trabalho e design de aplicativos

Reimagine seus processos e transforme qualquer fluxo de trabalho em um aplicativo pronto para desenvolvimento, com confiança.

Pega Blueprint™
Para concepção de estratégias de marketing e CX

Visualize jornadas do cliente e estratégias de engajamento em todos os pontos de contato e coloque-as em prática.

Pega Customer Engagement Blueprint™
Compartilhar esta página Compartilhar no X Compartilhar no LinkedIn Copying...