#81 How can make businesses use of ChatGPT w/ plenty of use cases

Hello and welcome back to another episode of the Process and Automation Podcast. As always with the Automation Guys, with me here today is Arno. Hi Arno.

Hi Sascha, how are you doing?

Good, good. Very exciting stuff prepared. We have done some research and we we prepared top 10 use cases for ChatGPT in a very specific industry. I'm going into that in a bit. So yeah, it's, it's nearly impossible to, to have missed the huge hype around ChatGPT, I believe. I think, and generative AI, of course, the last couple of months, the progress and the updates and the, the rhythm of updates, it's incredible. So yeah, the first wave of excitement It's clearly come down, I believe. So obviously everyone was using it, incredible growth of usage for ChatGPT. But now the sort of the first wave is over. There are lots of people looking more and more into ways to really make it work for their business. And that's what we, what we want to dive in today. So the 10 use cases we have picked are related to banking, the banking industry, financial services, specifically banking. So, so that's, yeah, that's the, that's the reason why we have chosen like a real industry and to showcase that, yeah, ChatGPT can be used not just for experiments and generating here and there some emails and letters. So some real use cases. We can look at. Yeah, so banks are very often at the forefront for adopting cutting-edge technology and, and using that technology to provide better customer service. Very important for financial services and banking is the compliance. So they're using the technology to, to meet their compliance requirements. And now with the rise of AI, banks can leverage that generative language tool, tools like ChatGPT, to of course improve productivity and streamline their operations and de-risk their business and enhance their services. So yeah, let's dive in into this really 10 great use cases. Arno, do you want to get started with the first?

Yes, thank you, Sascha, and Like I said, not a day goes past where somebody doesn't mention generative AI and ChatGPT. So the first one on our list is obviously customer service. Banks can use ChatGPT to effectively transform their customer service by providing real-time support via ChatGPT-driven chatbots. Obviously ChatGPT has got a natural language processing capability or NLP capability. We've talked about that in the podcast over, over the sort of last year or so. And, you know, it allows it to, to respond very quickly to customer queries, customer complaints, and things like requests for information. And it does this very quickly and efficiently. And again, you know, we've kind of chosen the banking industry, but we feel that this is probably omni-industry, so this probably applies to other sort of industries as well, not specifically just in the financial services. But we've decided to sort of kind of focus a bit on banking today because, you know, that's sort of the place where you can see some of the real use cases come out. So again, if we mention some of the use cases, you know, like customer service, you know, that probably applies to kind of many industries out there. B2C sort of facing industries. The next one I want to touch on is fraud detection. Now, we spoke about fraud detection in sort of previous episodes as well, and ChatGPT specifically can assist the banking industry with detecting and preventing fraud by really analyzing the vast amounts of transactions and sort of underlying data and, and, and, and identify suspicious patterns within it. And of course, you know, this can help banks to protect its customers and their financial assets. You know, they can reduce losses from— that's a result from fraud. And, you know, you can even, with inside the bank, you know, personnel can, can set up alerts. So, so security professionals actually get notified of, of suspicious activity. Of course, ChatGPT has got a very, very good API, so it's not just limited to what we see online with the OpenAI interface. It's got some solid API integration, so some of these alerts that I'm talking about can actually be integrated with other systems as well. So it's not just sort of, how does this look? Well, you have to go to this web interface. No, it doesn't work like that. You know, you can actually embed it into your systems and you know, it's not just limited to your kind of internal systems. Of course, you can use these APIs as well to start off workflow processes. In this instance, that could potentially open up a fraud case and for somebody to actually look into it and, you know, proceed with dealing and actioning that case.

And I think that's where the power of ChatGPT then comes in to really link it up with other systems across the process and across the company. So just as you mentioned, just to chat there, it is already quite valuable. But to really get the huge benefits, makes sense to plug ChatGPT into your data, not using the data which is available out there, right? And then it really gets exciting. So yeah, and that brings me to the next topic. The next use case is around loan origination. So Yeah, when you open up a loan, you probably experienced it yourself, so it can be sometimes a very complex and tedious thing. So, so generally there are multiple steps involved, so it doesn't— sometimes, of course, nowadays with the apps you just get a loan decision straight away, but behind the scene there's plenty of stuff going on. So it's, it's basically collecting customer data, analyzing the credit store scores, assessing the risk, processing will then continue. Yeah, and then we will get to get a decision. So, and by using or leveraging ChatGPT and NLP, so yeah, as part of that process, banks can automate many of these tasks, making the loan origination workflow much quicker and ultimately easier for customers and for the employees. And of course, when it's easier for the customers, that, that will result into customer satisfaction. They will go for another loan and for obviously other services of the bank because the experience was so great. So, so yeah, when customers apply for loans, ChatGPT can then provide that real-time guidance throughout the process. And that could happen not just on the customer side, it can actually happen as well later on in the follow-on steps internally. So ChatGPT can then guide your employees was the same kind of guidance. So yeah, and really plugging that into your applications will make it really— in your application data will make it really, really valuable. Yeah, so that's loan origination.

So the next one I want to touch on was actually two. So the first one is wealth management, and all of us have been on YouTube to see how ChatGPT can make you money by investing wisely. Yeah, you know, this is not investment advice and I'm not a registered financial advisor, so I wouldn't recommend doing that. But there is a, there is a degree in here that ChatGPT can actually assist banks in providing personalized wealth management services to, to the customers by effectively analyzing the customer data and providing, you know, customer investment advice based on sort of the customer's profile and their individual financial goals and of course their risk tolerance. So again, it's within the sphere of knowing your customer, you know, knowing trustworthy sort of, sort of investment options that's out there, what's inside the bank's portfolio and really just leveraging that to create these sort of investment recommendations Then the next one, and this is one we've touched as well before in some of the discussions around RPA and sort of workflow automation in general, is compliance. And, you know, of course, compliance is something that's never going to go away. It's a very critical aspect of banking. And, you know, there are kind of significant financial penalties and, of course, reputational damage that can occur if there's breaches in compliance. And, you know, banks are looking at ChatGPT to help them with their compliance requirements against sort of the regulatory requirements out there. And again, this can be by monitoring bank transactions and identifying sort of potential compliance validation. So all of these things are describable in a model. ChatGPT can obviously use the outcome of these transaction monitoring and then flag those potential violations. And of course, you know, this, this can help banks avoid costly fines and penalties associated with compliance breaches and also more critically protect their reputation.

Very cool. Yeah. So the next one is financial planning. So many customers rely on banks to provide that. Financial planning services, and this is really another great opportunity for using ChatGPT. And yes, so that these banks can use the, the model to provide very personalized financial planning services, including budgeting, debt management, and for many people also the, the retirement planning. So because it has access to all the data, it can come up with that very personalised recommendations. So yeah, that brings me to the next one as well when it comes to data, KYC and anti-money laundering, and know your customers, that's what KYC stands for. Sorry for sometimes not going too much into the real world, so I'm dealing with those kind of terms every day. Yeah, so KYC, know your customer. And AML, anti-money laundering, is another very big important area for banks. And it's actually not— yeah, it's really critical to mitigate that financial crime risk and one of the core elements of compliance for banks. So yeah, ChatGPT can assist those banks automating these processes by again looking at a large amount of customer data, also looking at the personal information and the transactions history. And yeah, we'll come up based on that analysis, are there any risks and compliance issues with that? Yeah, and it verifies that customer identity. So you know, you have to scan your passport and provide your other information, address and all that kind of stuff. So yeah, so the system can check all of that if that is all available. If someone is on a sanctions list, a very, very hot topic at the moment, yeah, and flag that suspicious transactions or generally suspicious behavior. Yeah, so very important to reduce the risk of errors and yeah, and it will really improve the accuracy of your compliance in the bank.

Yeah, and the next one I want to touch on, Sascha, is obviously acquiring new customers and the experience associated with customer onboarding, because of course it is a very critical process for banks to attract and retain customers. And these processes can be time-consuming and very complicated. If you work on a sort of customer onboarding, customer journey process for banking with all the products And, you know, you might have a multi-portfolio type customer. This is where sort of ChatGPT can assist you to improve that customer onboarding experience by automating, you know, much of those processes, which will obviously result in reducing waiting times for customers to be onboarded. And, you know, ultimately this streamlines the customer experience, which would lead to a better customer retention. And some of the things that it can assist customers with is sort of filling out the onboarding forms, getting answers to questions about kind of like the banking rules and the process. What's the next step? Addressing some of the concerns about switching banks. And this, all of this can be done in quite a natural language type speak, you know, rather than a kind of, you know, form submission. And it's very hard to, interpret what it's trying to say. And ultimately it can lead to sort of helping the customer to very quickly open new accounts. And, you know, using this ChatGPT model for onboarding, you know, it can actually help banks as well with the onboarding process. And this includes checking customer data for accuracy, you know, flagging any issues that need to be corrected, You talked a bit about identity verification earlier, Sascha, but yeah, this is something you can automate as well to make sure you get the customer identity, customer identity sort of figure out and all of that is done in a sort of compliant way. And then I think a key one here is sort of just offering personalized onboarding. For customers, you might have a segment of customers that you want to onboard differently and give them a different experience. So this would definitely help with that as well. And then also kind of from a customer concern perspective, it can actually help you to address those customer concerns very quickly, very effectively in the same way, predictable. So it's just kind of like a nice cohesive onboarding journey. Which results in obviously customer satisfaction going up. Customers will make recommendations on that. It's so easy to, you know, go out on social media and complain about a, you know, a bank, for example. And I think all of these things will just kind of mitigate that and just safeguard banks against that. And I think that's a very, very powerful feature of using ChatGPT as part of that sort of onboarding cycle or some of your onboarding processes.

Yeah, no one really wants to fill out these long forms anymore. So yeah, so I think we all are chatting away, isn't it? So if that can be done easily with ChatGPT or with the support of ChatGPT, it's really good. But I think it feels sometimes it's a long way to go, isn't it? So with all that excitement, We're really at the start of all that. So some obviously making starts into playing around with it. That's probably the right term. Some maybe do something real with it already, but the majority is just having a quick look at it and then saying, yeah, that's nice. But it would be really nice to see more of that really happening. And I think this is what this episode is all about, isn't it? To give everyone ideas and it can actually be done. It is, it is real.

Well, I think, I think also that again, the API integration and with ChatGPT version 4 coming out with sort of the ability to start processing voice commands and things, it's not that far away where you will be able to speak to an Alexa-type device and be able to open a bank account, for example, or launch a complaint because, you know, all of that natural language processing that happens behind the scenes with the large language models, you know, it will be able to detect your voice. If your voice is linked to an identifier, then obviously that takes care of, you know, your identity element. I know there's subsequent things that probably needs to be dealt with as well, you know, sending in a selfie and your passport and blah, blah, blah. But I think it just opens up so much more than just being online as well, you know what I mean? So I think once we get to the kind of voice activation and voice processing of this, you know, there's gonna be several kind of new ways to actually, you know, do onboarding, for example, or getting help, resolving queries, you know, just these sort of things. So I think that we are in the sort of early stages of this, but all of that is possible.

I'm wondering what will be— nowadays we have these deep fakes, isn't it? So even with voice, you can't even distinguish anymore. Is Morgan Freeman now talking, or is he really talking, or is it a deep fake? So that will be interesting to see how that works with voice identity. So yeah, but that brings me to the next topic here, risk management. I think definitely with all these new technologies, banks definitely have to catch up with, with the advances of those technologies as well, not just for their own benefit of maybe processing things quicker or better customer service, definitely also from, from, from that risk side. How can someone actually abuse voice identity and that kind of stuff? So yeah, so yeah, with ChatGPT, banks can identify those and manage those potential risks by, by again looking at this vast amount of data. Everything comes to that, good, the data and structured data and sometimes unstructured data, but definitely clear, clear sets of data need to be available. And then yeah, we can identify potential risk factors. Also monitoring the transaction activities, we had that before when it came to compliance, yeah, really flag those suspicious transactions and identify potential fraud. So if you— so we know there's sort of this transaction limit of $10,000 or pounds or euros, whatever, whatever that might be in every country. But if you go in, for example, like, okay, I'm paying in $9,000 every week rather than doing this, I think systems like that and analyzing the data will come up with that very suspicious behavior. So, yeah, so the model can analyze as well market data and news. To assess potential economic risks related to politics, and specifically those risks which might have an impact on the bank itself. So if there's a sort of a huge strike coming up or whatever it might be, or a big catastrophe, a war or so, that obviously has all impacts on the bank potentially. So yeah, with the machine learning capabilities within ChatGPT, banks can continually learn from the past risks, the past classification of risks, and improve that management, risk management processes more and more. So, yeah, and the last one I like to talk about is the virtual assistants for banking. So, of course, we know, we sometimes see this chat thing on websites, and we We see that, of course, with all our banks these days, within the app or just by the normal website. So yeah, now with, with those technologies getting more and more available and getting more powerful, banks can provide the customers with, with such a virtual assistant 24/7, and to really help because it really understands, it can really guide someone with, with managing the accounts, paying bills, and, and making transactions. So a very, very powerful stuff is available to us very, very soon. I'm sure everyone will plug that superpower of ChatGPT, or of course there are also others. So we have seen Google Bard, maybe you have experimented with that, Arno, I'm not sure, but I looked it into recently. It actually caught up. It was very late to the party, but it has also some really good, has made really good progress there. So yeah, whatever technology you might use, ChatGPT is now, because it's sort thing of the time. But whatever you use, AI, generative AI, conversational models can really bring banking forward or help banks to, to, to go to the next level. So yeah, with, with that OpenAI ChatGPT handling customer services, service by automating responses. Yeah, the role of ChatGPT will be around providing that personalized experience, and I think this is where it's really, really strong, based on that user data which is already available. So yeah, plenty of really good things possible. Some of those elements are also, like you mentioned earlier, Arno, possible within other industries. Customer onboarding can happen in lots of different companies, not just banking, insurance, for example. So yeah, really exciting stuff and really important to look into it. So because companies who are not really taking this serious, I'm sure big banks, they will definitely take it serious. But everyone who's just looking at it and think, ah, that doesn't apply to me, I think it will be a problem going forward. So companies have to adopt it, otherwise they eventually be overtaken by the competition, right? So this is just a list of 10 use cases which showcase the power of what's going on at the moment. Anything that you'd like to add?

No, I think we've had some very good responses from our listeners asking a lot of questions about ChatGPT and generative AI in general. So I think we'll, we'll start to focus more on exploring the kind of application of, of these sort of technologies in future episodes and really sort of just try and track its progress and, you know, where we see the opportunities exist for this tech. And again, you know, if you've— for our listeners out there, if you have any questions or any comments you want to sort of add, just feel free to reach out to us. We are at theautomationguys.net or just feel free to link up with either myself or Sascha on LinkedIn. I believe our profiles are in the show notes or if not, you can just find it on the website. More than happy to sort of engage in conversation and, you know, explore this a bit further. I think it's quite an exciting sort of time to bring AI to people, and I think in a way that they understand it. So hopefully in this episode, provided a bit of a glimpse of the sort of possibilities that, are out there, current use cases that can be sort of applied using ChatGPT or any sort of generative AI technology. And yeah, hopefully this was very informative to everybody, and, you know, we look forward to sort of talk more about this subject in the future.

Yeah, and if you'd like to enhance on that and get really hands-on with automation technologies, combining that with generative AI, specifically ChatGPT, yeah, check out our website theautomationguys.net, join our community, and yeah, get your hands on a sandbox where you can experiment. And we will be back with another episode very soon. And until then, let's automate it. Unfortunately, that's it again with this episode of the Process in Automation Automation Podcast. If you like this episode, please give us a 5-star rating and don't forget to subscribe to this podcast so you don't miss any upcoming episode. We hope you will tune in next time, and until then, let's automate it!