Tuesday, December 19, 2023
HomeBankPodcast: Lazard saves 100K hours annually with UiPath

Podcast: Lazard saves 100K hours annually with UiPath


Financial institutions including asset management firm Lazard and technology provider Fiserv are looking to business automation platform UiPath for automation, AI and added.  

Through UiPath, Lazard, which has a market capitalization of $3 billion, saves 100,000 hours annually, UiPath Chief Product Officer Graham Sheldon tells Bank Automation News on this episode of “The Buzz” podcast.  

Lazard tapped UiPath to automate pitch decks — which traditionally took its teams two to three hours to create — using generative AI for analyses and formatting of pitch decks, Sheldon said. 

It took six months to implement the UiPath technology into the fintech and has reduced errors and improved Lazard’s overall consumer experience, he said. 

Listen as UiPath’s Sheldon discusses how financial institutions can improve operations by automating mundane tasks with AI and generative AI.  

Get ready for the Bank Automation Summit U.S. 2024 in Nashville on March 18-19! Discover the latest advancements in AI and automation in banking. Register now. 

The following is a transcript generated by AI technology that has been lightly edited but still contains errors.

Whitney McDonald 0:02
Hello, and welcome to the buzz of bank automation news podcast. My name is Whitney McDonald and I’m the editor of bank automation News. Today is December 18 2023. Joining me is Chief Product Officer of business automation platform UiPath Graham Sheldon, he is here to discuss how financial institutions can implement AI and automation to improve employee and client experience. Absolutely.

Graham Sheldon 0:24
Well, thanks for having me, Whitney. My name is Graham Sheldon. I’m the Chief Product Officer for UiPath. And what that means is, I sort of look after the vision, the strategy and the roadmap for what we build at UiPath. And I spend a lot of time with customers trying to create what’s coming next for us in terms of AI and automation UiPath, we build a business automation platform. So we build these awesome software robots that help people either work with them, or work on their behalf, to make them more productive. And, you know, what we’re really trying to do our mission is to sort of help people achieve, you know, achieve more, both in individuals and together with the rest of their company. And, you know, that really helps them save money, with cost savings, it also helps them operate better. But it really also just drives these sort of end to end experiences for customers, for employees, for investors. I know you worry a lot about, you know, the the financial world, and we try to help them out as well, by putting the best of Ai plus automation together in the solutions for those customers. Well,

Whitney McDonald 1:45
in today’s environment, we definitely can’t ignore AI. So we’re excited to have you. Thanks, again for joining us on The Buzz. Let’s say to kick things off here with a little bit bigger picture what really is AI bringing to finance professionals today?

Graham Sheldon 2:02
Yeah, so with AI today, people are asking these sort of fundamental questions, just like you said, like they they kind of understand because they’ve made of us chat TPT, that AI can do some amazing things. But they’re really starting to ask us now like, well, what can I really do about it. And so, you know, finance customers of ours have started to really seamlessly integrate intelligence into everything that they do. And it’s sort of fundamentally changing the way people work. So if you think about it, all knowledge work involves, you know, trying to understand what’s in text or in images, and then trying to help people understand what could be done better, automating pieces of it, that could maybe go faster, or be done more efficiently. And then really trying to, you know, do that at scale and help best practices so that every person can be better at their job. And that you can make maybe the every one is good as the best in their particular field. So, you know, it helps bring in like new ideas and helps you create things, right. It also helps you sort of transform and understand what’s happening, like really getting to the bottom of things, summarizing information, or being able to help you understand better what’s going on in a process or in an in an email. And then you know, sort of unleashing productivity, you know, the things that used to be routine, the things that are kind of annoying or hard to do get way easier. When you’re starting to use AI to help you understand help you get things done.

Whitney McDonald 3:43
Maybe we can break into some of those annoying or mundane tasks, those manual processes that AI can can help to add efficiencies to and facilitate those activities that you don’t necessarily want to spend hours doing are breaking down data, what are some of those manual processes that you are seeing AI replace? Even if it’s UI path? That’s that’s doing it?

Graham Sheldon 4:05
Yeah, absolutely. So you know, all financial professionals out there, they went to school to learn how to, you know, analyze data, or, you know, come up with, you know, projections or, you know, look at investments and things, think about, you know, how to underwrite it a great deal, right? They didn’t go to like, enter data. And a lot of them today, you know, spend a ton of time just getting like a document or a PDF, or just sifting through contracts and long documents trying to figure out, you know, how to apply these things and they’re, you know, copying it from one place and entering it into another, taking it from Excel, putting it into an email putting into a prospectus. There’s a lot of like, very laborious work, and it’s prone to errors. Right. So just give you a concrete example. One of the most frequent use cases for UiPath is for helping to process invoices. So you know, when you’re doing this order to cash kind of process, every customer is getting invoices from, from folks, they’re taking, making sure that all the right line items are there, that it shows up and doing either two way or three way matching, which is when you try to make sure that like the goods that you got match what you’re being you’re paying for, right? And you copy and paste from one document to another or into Excel and then running a tabulation to make sure that the totals are right, or the taxes are right, or the number of goods is right. And that process. Now with AI allows you to take a lot of that out of the hands of that person, so they can focus just on the stuff that’s really hard and interesting. You know, why is there a discrepancy between these two items? Where is it that we should be getting these goods from? Where might the next you know, cost savings come from, so that you take that kind of out of the realm of someone having to do it every single day and take hours to do it, and have ai do it for you? And

Whitney McDonald 6:16
that’s exactly the next question I was gonna say is maybe we could talk about some financial institutions that are doing this. Well, where is that automation really being seen? Where’s the AI being leveraged? We could get into some examples there. Yeah,

Graham Sheldon 6:29
you bet. One of the customers that I got to talk to really at length is Lazard. So you may know them, they’re global financial advisory and asset management firm. And what they have done is they’ve automated, an extremely time intensive process, you know, transforming information from multiple different sources to create these pitch decks, right. And these pitch decks, you know, it takes two to three hours, typically for 1000s of their employers, their bankers to create. And they they’ve automated this sort of coupling both the automation to get the information from the different sources, from their internal tools from some external sources of data from Excel from SAP from these other line of business systems. And then using generative AI to sort of tweak the those numbers to analyze and to create suggestions for the banker to sort of select the right data and then to format it in a way that you can sort of answer complex queries about maybe a deal that you want to where you can hand it off. And so in just six months, Lazard implemented that automation. And now they are able to answer those questions more accurately, they’ve reduced the errors on that enormously. And then they’ve really enriched the client experience because they can answer much more in depth questions and spend the time really delighting those employees. And so they’re estimating they can save upwards of 100,000 hours every year, and expect to be able to, you know, grow their operations as a result of that savings.

Whitney McDonald 8:15
That’s a lot of hours. And is a lot. Um, no, I know that you’ve mentioned client satisfaction, or that you just mentioned and we’re able to quantify what those savings looks like, what it looked like, but when when it comes to leaning into AI and productivity is one thing that that you’ve been mentioning, how does this really improve overall employee satisfaction as well that you’re able to, like we’ve talked through at the beginning, get rid of those annoying, mundane tasks, and really focus on maybe more interesting or intricate pieces of the data that AI helps to pull out. But how is this in improving employee satisfaction?

Graham Sheldon 8:55
Yeah, you bet. So for I’ll give you another example, actually, the customer example, Fiserv. They’ve automated a whole bunch of processes, especially in their customer service. But they’ve also expanded this now into HR and legal and it and nearly half of them are focused really on the end customer experience. And if you think about in the customer, like in the call center, or in customer service, there’s a lot of turnover. And that’s because directly these employees don’t get to work on the things that they love to do, which is really making people happy and and satisfying their needs. There’s nothing more that these folks would love to do more than to actually just solve a customer’s problem or help them get something done. And so what Fiserv did was they created this center of excellence. Coe, I’ll maybe refer to it as that funded by the IT group, but they basically enrolled the whole population of employees to go develop these automations for themselves. You And by sort of bringing them along and giving them the tools to fundamentally change their own experience, they were able to obviously drive, you know, things like efficiency. So they average handle time went down by 50%, after they did some of this, this work and processing, you know, 50 cases daily for every agent and saving, you know, 10s of 1000s of hours every year. But they also moved employee satisfaction, because you’re not focusing on trying to, you know, you’re on the phone with the customer or trying to get the answer to them. And you’re looking it up from different places. Now, the automation is bringing that to them, so they don’t have to hunt around for it. And when it sort of here’s what’s going on, you can get the right information from a knowledge base, right. And building that into the desktop console with robots, you know, lets them quickly gather that information in ways they couldn’t have done it before.

Whitney McDonald 10:58
Now, when it comes to the technology itself, I know that UiPath has been working on a few things to allow for this type of automation to meet these market needs to improve efficiencies throughout institutions. Maybe you could talk through a few what those products are that UiPath has been working on. Yeah,

Graham Sheldon 11:19
absolutely. So AI powered automation is what we call sort of this. Well, how we bring together ai plus automation in our business automation platform. And we keep we’re sort of continuously working on the entire platform, bringing AI to help make that product better. But there’s a couple things I wanted to maybe highlight that people may not have known about. The first of which is autopilot. So at our at forward, in October, we announced that generative AI powered assistant will basically allow folks using generative AI, and something we called specialized AI, which I’m sure we’ll get back to in a little bit, are actually helping people automate and bring this more naturally into their work experience by just letting them use natural language to do some key things. So for developers, creating automations is kind of a laborious process, like you have to, you know, put together all these different steps to gather all that information. Now you can create workflows with natural language, you can also create if you’re a developer create expressions and applications just from a form. So think about taking a paper form, showing a picture of that to the AI and having it come out with a digital form equivalent of that with all the automation behind it just takes a couple of clicks, then, for everyday users, you can actually use autopilot to get stuff done, to actually perform tasks on your behalf that make use of those automations. Things like being able to look at a document and extract the key information, and then enter that into another place. All that takes is you telling autopilot, Hey, move this data from this form to this other place. And those are the kinds of things that you can now do with autopilot that you could never do before. There’s a lot of other things too, that we’re working on so that people can create their own automations that make use of generative AI, so they don’t have to use our autopilot, oftentimes, they want to create their own generative AI experiences. And so making use of our generative AI in their own automations is as easy as sort of dragging and dropping in components that will use models like GPT, four, or from vertex from Google, or from bedrock from their models in Amazon. And putting that together. In a platform like ours requires that you have a lot of trust. So we have an AI trust layer, that helps make sure that only the right information goes to those, those AI models, and that we’re able to marry that with the automation capabilities. So that humans are always in the loop for the important things.

Whitney McDonald 14:19
With all of this innovation with all that you’re doing in UiPath, the the discussions that you have with your clients the feedback that you get the technology that they’re asking for, of course mix with the innovation that you guys have going on, what are you really watching for when it comes to AI and what the possibilities are looking ahead kind of to 2024 What are you excited for? What are you working on? Just kind of a future look here?

Graham Sheldon 14:44
Yeah, there’s a few things Whitney that really get us pretty excited about what’s coming in 2024 and beyond. I’d say the first one is that, you know, we really believe in this notion of specialized AI and some of the smaller More task or domain focused AI models. So, you know, with models like GPT, four, and some of the larger foundational models, they they’re built on the web data, very publicly large public sources of data. They don’t necessarily know your data, they haven’t seen your invoices, they haven’t seen the way that you work. And in order for you to get better, higher accurate models, faster models, ones that will help you explain how it got the answers that it needs to, you need to have a specialized AI model that you’ve trained with your own data on your own tasks. And so what UiPath does, in addition to making the use of the best sort of generative AI models, which continue to get better, is you need to think about your strategy for these specialized models that really know the way that your business works. And so the UiPath platform helps you build your own models, both with out of the box ones that we have for things like invoices and purchase orders and, and expense reports and those kinds of things. But also, for special use cases that you might have. So I know that we’ve got customers of ours who are building in the insurance market, they’re they’re doing this for claims processing, they’re building models to understand those kinds of documents. I’m also excited about sort of the multimodal so that’s that’s one thing is specialized AI is here to stay and you know, the the that’s that’s a really important part of everyone’s AI strategy. Now, the second is on multimodal models. So GPT, four v. And some of the models that Google just released Gemini. And some of these models that were actually building ourselves do an interesting thing, they they actually come back, they’re not just about text anymore, or they’re not just specifically focused on images anymore. It’s the combination of all of those things. And the really cool part about that is that what that does, is it can understand like what I’m looking at, like what’s on my screen, in addition to what I’ve typed, in addition to all the tasks that I’ve done before. And so I mentioned right at the top, how we’re helping people become the best versions of themselves. If you know that the way to delight a customer, is to send them an email after your order and say thank you for your order, I see you did these of these things, if we can build models that help people remember or can suggest the next thing that they ought to be doing, or that build in some of the knowledge from how their co workers and their industry do things, best practices, you can do some immensely cool things, to delight customers and to make employees more productive. So those are a couple of the things that I think, excite us quite a bit. I think the last thing I’ll mention is in the realm of automation, one of the things we get asked all the time is like how do I make sure that my robots aren’t going to break? How do I make sure that it’s going to be reliable, and that I can trust that it’ll. And in the AI world, one of the things that’s going to is coming is the sort of notion of self healing robots. So robots that have the AI built in, so that they can be more resilient to changes. As you probably know, applications get updated all the time. And when they do, you know, buttons move around, or things change fundamentally, that might break your automation? Well, if the robot knows what it’s trying to do the intent, it can actually go through and try to go around and complete the tasks without being told exactly the way that it was done before. What that means is it drives Total Cost of Ownership down makes those things more resilient to change your robot succeed more of the time. Those are some of the things. There’s a lot of hype around AI still. And there’s still a lot of distrust, and a lot of customers that they can use it responsibly. It’s critically important that when you use AI that you do so on a platform that’s well governed, that’s manageable, that helps you understand things, you know, who’s using which models, what data is going outside, to do those those things, what is my what are my models trained on? And then building humans in the loop. We didn’t talk as much about humans in the loop, but there’s some there’s some decisions, frankly, probably a lot of decisions that really require people to make them and it’s everything from when you’re about to write a check for $10,000 to make sure you didn’t include an extra zero When you’re making a critical hiring decision, you want to make sure that a human is in the loop when you are trying to make sure that, you know you have routed an email to the right place where we are, or a simple one where you’re about to just, you know, send an email back to a customer. You want a human there to be able to make sure that you’re saying the right things. And so having ai plus an automation platform like UI paths, to be able to make sure that your governance rules, your compliance policy, your privacy, posture and security roles are being well taken care of is critically important, so that you can trust that what’s going on.

Whitney McDonald 20:43
You’ve been listening to the buzz of bank automation news podcast, please follow us on LinkedIn. And as a reminder, you can rate this podcast on your platform of choice. Thank you for your time and be sure to visit us at Bank automation news.com For more automation news,

Transcribed by https://otter.ai



RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

- Advertisment -
Google search engine

Most Popular

Recent Comments