Parth Desai - Open Banking Challenges & Opportunities: Why AI is the Essential Business Enabler

Good afternoon, everybody. My name is I've been working the field of payments and compliance in the field of banking for last 25 years. My journey started about 35, 30 years ago. Infin AI were in shorts and trying to do things, but I think that did not work at that time. And I started a company and few years after that, and one of the first customers we had was a bank in Germany.

And I had, I, I kind of, my second home was down start for a few years. So trying to make things work with using AI. And so I'm good. It's nice to be back here now. Why am I here?

Basically, I would like to talk to you about the opportunities that we all have in the field of open banking, 19, especially, which is because of the PSD two, right? PSD two has forced upon banks, especially banks to open up their APIs, and that has created issues for them. Some challenges for them at the same time, there are some, there are some which banks also have.

So I would like to address the challenges and talk about how we can exploit the opportunity at the same time with AR as one of the, so with D two, all of you know, regulation allows now third party providers PPPs as they're called to access the customer's data, which are held at the bank. So the banks will have to allow these other parties to access the customer data, which was originally kept by themselves and was kind of proprietary kind of proprietary to them now because of this data, which should be accessible, the data is no longer an issue.

I mean, data, everybody will access to data. And the, it basically a level playing field between the banks themselves, the new banks, new banks coming up, especially in Germany, all over Europe and the third party providers. So all of them have access to all the data which is available. So that is because, because of that, what happens is data is no longer an issue. What is important is so what is important is going to be the customer, right?

And then now the, the bank has to decide, is it going, going to treat PhD two as a challenge, as a compliance issue and just do minimally what is required to be done or take it as a strategic opportunity. If you only do what is supposed to be done or what you are, you have to do, then you will remain a banks will remain just a holder of the client data, but being accessed by all of everybody else.

And the, the direct customer relationship, which is banks used to own so far will be at risk. So all the customers, all the PPPs and other banks will also have the data that you hold and off your customers. And they can start providing services perhaps better, perhaps worse, but at least there's a big risk that you may lose direct contact with the customer. And because of that, you may lose the opportunity for further revenue generation. Now.

So, as I said earlier, the main issue is customer centric, who is going to focus on the customer. Are you going to be able to capture the customer, keep your customer, and also get more customers from other buyers that is the opportunity you have. So if you are able to capture and retain customer, it becomes a strategic opportunity and you need to do whatever necessary to go in that direction.

Now, if you take strategic approach, then the question of open banking comes up. I mean, many people use PSD two and open banking as of each other. But I believe that open banking is one level higher than easily to be easy to forces banks, to do minimum things. Whereas open banking is a strategic approach with some of the banks and some banks are taking, which I will go one step further and make open banking as a strategic way forward. But what does that mean? Basically it means three things.

You open up your bank, not only for customer data, for account information and payment initiation only, but also for other services, for example, direct updates, mandate management, loans, effects, standing orders, stuff like that, which is of value to your customers. You want to allow all the banking services available on these other services also, but on these, the main differences you can charge for these services, as opposed to the PSD two PSD two services, which you cannot charge. So that is where you can change the whole thing and convert into an opportunity for doing that.

So what I would do if I was a bank open up services at value, all value added services charge for them also connect to other banks bank themselves can be a TPP. It can act as a FinTech and then they can also get the access to all other customers of other banks in the country or actually Europe wide. Yeah. So they can also offer products and services everywhere right now. Actually there's an example of your, there is a, I G Netherlands has come up with a application called your in UK, right? UK doesn't have any base in UK, but it can offer base or this PP provider PP regulation.

It can provide service in UK and also provide services to not only your customer, but also other banks customer. So this is a strategic opportunity. You can open up your services, start making money or other services, become a TPP and connect to other banks, get that data and start providing services to not only your customer, but other customers. Also. Now this is where the problem then come, turns out. The main problem will become a customer engagement. That should be your focus.

The only focus is you can then access all the data through Europe-wide and then focus primarily on customer engagement. You not only have to interest to customer, but also engage them and delight them. What you need to do is the person who has access to the customer, who is able to go to the customer, offer interesting solutions for the benefit of the customer will be the party that will bring. And that should be the focus for the open banking from the time now.

So there are significant challenges, what you need to handle this, not only huge amounts of data, but also be able to serve and like millions of customers. So that is the challenge. So how do you, how will you be able to access the data, manage the data and at the same time manage multitude of customers. I think one of, one of the sessions yesterday, there was some point about some data being in different systems was not able to the customer data was not in sync. One of the problems one of the customers had, but that is just a point in case that there are issues that we need to address.

Now, there are, I will, the data problem can be converted into two. We divide into two seconds. One is about data quality. When you have lots of data, lots of disparate data in multiple languages, multiple formats, structured unstructured, which all need to be understood. And it has be put together to make sense out of it. Yeah. And humans cannot address this data and not only have multiple types of data and problems with data quality, but you have lots of data.

You will have lots of data, all over different systems, across different bank and different in-house systems, across different banks, different channels, different interfaces. You will have to manage this income. So what you need to do, because now, sorry, this data has lots of data with back quality. You have to use it to understand yeah. You to understand the data and understand the customer. If you can understand what I would call customer Coon, you are able to understand the customer and understand the needs of your customer. You will then be able to provide the service home.

That is the first important thing. So you are able to engage the customer. If you understand the data, understand the, what is containing the data, understand the behavior of the customer based on the data that is coming up through open banking, integrated to you understand what a customer is doing and would want to do. And that is where the customer cognition is. One of the example is think the best example of what customer organisms is, for example, Uber, right? So Uber understands. It's a very narrow scope compared to the banking, but it has mastered the customer CS.

It understands where you are, where you want to go. And just from a click of button or press of button selling where you want to go, everything is automated. Just say where you want to go.

The, the taxi will come to you. If you get in, you go where you are going, you can track it. You know where it's going to arrive there. You can do whatever you want during that period. When you reach there, you just walk out. There is all, not even a hassle of making a payment. So this is a kind of experience that customers would want. And I getting used to more and more. So you need to apply that in the field of banking. So this is where the AI disciplines come into training. There are two disciplines, which are interesting.

Now AI, as some of you may know, covers many different aspects, right? There are some disciplines of like sensory AI basically tries to use simulate human behavior. Yeah. It tries to replicate human behavior and the human behavior can be characterized characterized into sensory behavior behavior regarding your mind and brain and motion. Right? So voice vision is all sensory behavior movement. Robotics is about movement. And the natural language processing is about the human mind. You need to be able to understand the reason about that. I will come to that.

So if you really want to act like a human being, you need to understand, read the information, understand the data, understand the meaning behind it and reason with that meaning. And also remember, and learn from that. That is the key function that machine learning and natural language combined together allows you to do. Right? So we talk about AI with the machine. Learning with natural language is something that is added on top of it, which can really get value out of data and get the meaning out of data and make them make the software reason like the humans do.

So how do you like and understand the customers? Right? So let's take examples of what do let's. Let me explain what I mean by customer engagement. You need to be able to provide proper services to the customer at the right time. For example, you have, you can take consumer or a business as an example, right? Whenever there is a consumer working in the field, for example, a student, his behavior is very different.

He has, he has lot less money. Doesn't have disposable income. He's trying to spend every money he has. Then as he becomes employ, he starts earning more and more. He has more and more disposable income and start using that for restaurants, entertainment and things like that. As he becomes, as he becomes, he gets married, has children by house, again, his behavior and things change. So you need to be able to understand that drag the journey of the customer and understand what he doing, how we, what his requirements are.

And you can understand that, learn that automatically through machine learning, looking at the peer analysis, what other people in that group are doing. So by merging understanding a particular behavior or vertical customer understanding in which peer group, which fits in, you can then predict his behavior, recommend something to him. So if you can provide the meaningful services at the right time, there's no point pushing for mortgage and to a student too, we can't afford it.

But the question is at the later stage, if you can understand from his data that there is sufficient money available, or for example, he has, he's buying something in mother care for children that you know, that perhaps he needs perhaps a mortgage, he needs a car, he needs an insurance and things like that. These are the kinds of things that you can understand automatically using machine learning and reasoning to actually understand the customer and provide the right service at the right time. I mentioned the highly personalized service, for example, the case of Uber, right?

It knows exactly where you are, what stage you are, what you want to do. And it kind of makes that easier.

Also, you need to be able to connect the customer through multiple channels. It should not matter if you want to talk to you, you should be available as in when required. I'll come to that a bit later. Again. Now in this current world, everybody is used to instant gratification. They want to know answer immediately. You click up a button, they want immediate information. So this is where the voice recognition, the voice trade solutions can really help to actually make things easier.

Nowadays, I, I would be. Now, if you can look at many students, they just, instead of using voice and WhatsApp, instead of typing data, send voice messages, it would be great to find that asked by app straight away saying, what is my balance? And the balance is show on my, on my app, right? So things like that. So that is where voice is coming in. And that is the natural way of communicating with any anybody else. So if the computer or a app or a phone mobile is simulating a human behavior, you need to make sure that it is interacting like a person used to do.

For example, in the 30 years ago, when the bank started, used to go and visit the branches and talk to a better machine, to understand everything that you wanted to do, you were talking to a person then came up internet banking, or before that telephone banking, you had to punch the punch, the numbers to get that job done. It was actually the, as the time progressed, the, the banks have been pushing away the customer more and more away. So was face to face. Then there was telephone banking, internet banking, and how web all these were on key.

And they were pushing away the customer more and more with the use of AI and natural language processing and voice recognition. You are again using the technology to give our proper feedback to the customer at the right time. And we always have to remember that you always have keep in mind the context in which you are operating and the context, right? So it is important. That context is kept in mind. Who are you talking to? Why are you talking to them? What is the context in which you are talking? And a lot of times people forget the common sense aspect of it.

Sometimes computers or see, for example, act very dumb cause they don't remember the common sense part of it. And sometimes they just ignore that. So now if you have lots of data and lots of customer, we believe the, the technologies of machine learning and language become critical. Humans just will not be able to address these kind of things in a real time, in a more digital world, millions of terabytes of data is generated every second and this needs to be analyzed and understood.

And no matter how many people you throw at it, you will not be able to put people want 24 by seven instant response. And that is not possible without technology. So let's look at what machine learning can do Now using machine learning, you can start analyzing the data and trying to understand what's the content. You understand the meaning of the data. You understand the context, you understand the behavior of the person or a particular company and see what's going on. And then based on the information you can first just provide the statistics and so information about how is he doing.

If I'm a person I would like to know immediately what I can, how am I doing with respect to other people? I would like to know, how am I doing with respect to the general expected guidelines or my peer network. So if I'm a student or if I'm a fresh graduate in 2025 earning hundred K what is my, what is my behavior? What is my performance? Am I spending too much on entertainment? And am I saving enough, not saving enough. All that kind of general guideline guidelines would be interesting from the prescribed guidelines point of view, for example, financial advisor, point of view.

I also from the peer group one, how other people doing in my London, for example, or in Frank work, how the, that was something we very interesting. And then based on the lifestyles, then the customers can be provided some recommendation on their own, through these systems. And this is where the machine learning can come in. When you are in metrics, it immediately tells you what kind of movies that you can recommend. That is based on the behavior of the other customers that have used similar movies.

They recommend you similarly, based on the peer network, sorry, the peer in your group through customer segmentation, it'll tell you what you can or you should do. And that becomes very powerful in customer, gaining customer trust and interest. Now other customer insights can also be used for, to offer products and services, understand the behavioral changes. For example, somebody has a BB, you understand the life studies you have to change and you can recommend and provide interesting products and services, which will help him and benefit in future.

Also for small businesses is an area where I think machine learning and recommendation systems would be quite useful. For example, in SMEs, the SME market, what we have is lot of entrepreneurs who are not financially savvy, they, they understand their business. They understand a particular model of the business sales manufacturing and all that.

However, the financial side, they're not at savvy. So if solutions are provided to make their financial ecosystem or financial life easier, that is much more validated. So now I mentioned voice, voice, voice based system. Now What has happening is that everybody wants to understand that the everybody would like ideally to work, talk to a human that is practically not possible. Now that can be using AI and natural language and voice recognition. You can substitute that by an Omni channel, mobile based or a web based solution, which can interact very easily like a human really.

And that is where the Omni channel, natural language cross chart board, or a voice with a system can help. Again, if I'm going on our website, I would like to immediately type in a question and I would like an answer. I don't want to navigate through a website to understand where this information is. Yeah. So you should be able to, I think something and state away the frequently asked questions or the right page, the right section, the right form state away can come in and answer my question.

So these are something that can be very easily provided to users to make things easier and engage it again for customer service. People should just state ask a question, go on a website type in something or voice through voice, ask the question and the answer would pop up on the screen state again for navigation, for example, making a payment, finding out a balance. They should just ask state and should get a number, not navigating the system type in the query from when all that stuff. It's the waste of time and energy.

And these are the places where we believe by adding natural language and machine learning. You can make the whole system simple, easy and interesting for the end user.

So you, as, as I said, easy to can your strategic opportunity. And by using AI as an enabler, you can seize the opportunity and be a winner. Thank.