Name: AI Meets Finance: How to Drive Real Impact, Not Just Hype
Uploaded: 2025-06-04T15:32:46.292Z
Duration: 54 min 33 s
Description: AI Meets Finance: How to Drive Real Impact, Not Just Hype

Transcript for "AI Meets Finance: How to Drive Real Impact, Not Just Hype": Now I've never seen the chat fill up so fast. Welcome everybody to this webinar for AI meets finance, how to drive real impact, not just hype, where I'm joined by Christina to get into some core topics, including use cases. So quick wins for AI, where you can apply AI directly today and where you can't. A methodology. So kind of, Christina's got a great framework that you can use to assess the plethora of AI tools and vendors out there. I know it can be very much overwhelming when looking at these tools. Mistakes to avoid, not what to do is always useful, particularly when you have security concerns around AI as well. And then the bit I'm really excited to get into myself is the philosophy, why building that capability matters, more than just buying AI or just adding AI. And then I think also in a really important part of any AI discussion is the embedded AI. So when you look at tools, for example, treasury tools like Agicap, what level of AI functionalities do they have built into that tool? And then last but not least, there will be a q and a at the end. So please do add, your questions in the q and a section on the right hand side. Quite a few of you have already successfully posted in the chat. The q and a should be just along to the right of the messages column in the top right of your screens. If you have questions as we go, do chuck them in there. If Christina and I fail to respond to them, mainly Christina, then please, we will come back to them at the end and discuss them, individually. So I think that is all of our household tasks upfront. You have the chat. And you have the q and a. Very much looking forward to getting into this. I'll just quickly introduce myself. My name is Nat. I head up the partnerships team here at Agicap, and I'm very, very excited to be joined for this discussion by Christina Chen, CEO of First AI, one of The UK's leading, AI adoption agencies. There she is. Christina, do you wanna quickly introduce yourself and then maybe say a little bit about what your company does? Okay. Great. Thanks for having me here, Nat and Educap. Hello, everyone. So my name is Christina Chen, CEO of First AI Group. What we do is we help CFOs, we help finance teams to adopt AI and implement AI. We are specifically a Microsoft AI cloud solutions partner. So a lot of what we're doing in the Microsoft space. So whether you are buying off the shelf solutions from vendors like Microsoft, in that case, will help you adopt. So adoption programs, training programs, change management programs to make sure every employee of yours isn't getting the most out of your tools that you buy. Or if you decide to build something custom, configure something, implement something, again, typically around the Microsoft environment is where we operate, then we can, we have the experts to do to build AI, especially AI agents, which is all the talk right now. So that's what we do for the CFOs and the finance team, and I'm looking forward to getting into this, conversation here. We're excited to have you, Christina. And, Christina, you've been doing this now for seven years, eight years. Is that right? Yes. Yeah. So, a very early adopter of AI, so with a lot of wisdom from implementations, across the world, I'm gonna say, but mainly focused in The U in The UK, I know, and lots of lessons on what not to do and what best practice looks like. So, conscious we have a mix of Agicap customers and Agicap, new people on this call as well. Also conscious this is very much not the focus of, the webinar, so I'll just breeze through these quickly. But for those of you that don't know Agicap, we are a next generation treasury management solution focused on the mid market segment. We were born in France back in 2016. Since then, we have expanded to 12 different countries, got offices in six, and we have over 8,000 clients, in 2025. We are the market leader for what we do, both in mid market treasury and cash flow. And in terms of what our tool can do for finance teams, primary objectives of or or or the primary objective of the Agicap platform is is twofold. Really, it's about centralizing all of your cash data as a business in one place, and then being able to take action off that centralized data, to increase your cash efficiency. So for all of our customers, we typically connect to their bank accounts, pull in their bank feeds directly into our system, and then we'll also connect their ERP with their accounting systems, or, a spreadsheet if they don't have an accounting system even, and pull all of that data into Agicap as well to create a live cash flow forecast built based on some rules. For example, maybe customer payment patterns, could be one rule. Or if you always know, your your payroll is always gonna be paid on the same day of the month. These various different rules can be pulled together to automate the cash flow forecast, which we can then align with the budget by bringing that into. We give direct bank visibility in one place, a centralized overview of all bank accounts, and then you can also execute payments, and payment approval flows from within Agicap as well in this bottom left hand corner for our AP automation module. And then we also have our accounts receivable automation module in the bottom right, which is a workflow, or a module really designed to identify late paying customers and come up with automated workflows to encourage them to pay more promptly or, switch to direct debit so you have greater payment security. So that's a very quick overview of the Agicap platform. As I said, very much not what we're here for today, but if you do wanna learn more at the end, you will have the opportunity, to book time with me directly. Or at any point during this webinar, you can click book a demo, in the top, middle top of your screen. So just a quick recap in terms of what that means for the finance people on this call. If what I've done here is I've kinda split out the different functions of the finance team, and the bits in blue tell you kind of where Agicap platform can help. So cash management and cash flow forecasting, we got bread and butter. Treasury reconciliations, so really matching, your invoices against your bank payments, your supplier payments, your credit control, and then your budgeting and your control and your financial reporting. But let's get into the first part of the discussion with Christina. So, Christina, here you talk very, very clearly. You've got kind of four mindset shifts on the screen here. Which do you consider to be the most important, when getting started for finance teams? Yeah. Well, so just to set set the scene here, AI, everyone's talking about AI. Right? I think also for for CFOs and finance teams, there's a lot of pressure, coming down from, you know, sea level, from board level, from the business level where, you know, finance is often seen as a, a cost center, but at the same time, it's really critical to drive a lot of the intelligence, drive the strategy going forward from the financial numbers. So there's a a a we see a lot of tension wanting the finance team to do more at the same time as having less. Right? That resourcing balance, that prioritization balance really is a a a play from from that place, a lot of CFOs are are looking out for innovative solutions to solve that, and AI is perfectly kinda suited for that kind of tension and that kind of balance. One thing I wanted to say before any of the use case and sort of diving deeper is because AI is very different from a technology perspective, it does require a mindset shift. So if you compare AI with, you know, previous technology waves like clouds, like Internet, etcetera, A lot of those set behind IT desks. Right? You know, cloud is you know, have any background in the server rooms, etcetera. But AI is on the desktops of everyone. Right? And that really requires people think about how we work, how we impact search these days. And Google is very different. Google even comes up with AI answers first. Now you no longer think about search terms. You think in natural languages. And just seeing how fast everyone's kind of behavior change in, in a year or so, in several months. So just coming back to that then from a business sense, when we talk to CFOs, it's quite often to, we wanna talk through, actually, how do you wanna approach this in the first place. Now to answer your question, which is, you know, favorite, this they're all my favorites. That's why they're there, but I'll just pick one to talk about, which is the second one that's screen here. From moonshots to 1% everywhere. Quite often when we start conversations with c level, you know, CFO, t, finance teams, because there's so much noise about all these amazing AI technologies out there, and and they are truly amazing, People want to get there, when we can understand. But that really often grounds people to, you know, be be stuck, really, because it's not very easy to go straight from where you're standing all the way to, you know, all singing, all dancing, everything automated, everything autonomous, smart AI agents, etcetera. So we come back to this from moonshot to 1% everywhere concept. 1% everywhere, if people don't know, is, popularized by the British cycling team. They won many Olympics, and the coach that came in really thought about was really driving this. If we can just be 1% better everywhere, you know, the pedals, the the seat, the, you know, what the cyclists were wearing, etcetera, and eventually, these compounds. We have a, you know, finance audience here, so everyone would really get that point about compounding, the power of compounding. Right? So it's really thinking about how AI can even start small. Quick wins, small wins where it can save some time here. It can automate there. It can improve our efficient, accuracy. There that that's the 1% everywhere mindset. The firms that we've seen that started with the 1%, started with the smaller workflows, are the ones that, in fact, looking back, are further ahead than the ones that really had a one big AI initiative, going on. So that's the one I wanted to give a shout out to. Nice. Yeah. I completely agree with that compounding effect. And I I think that starts really with, basics like, you know, getting good at your prompt engineering through to, improving, your data. And can we just dig into that bit, a bit more as well? Practical data. I feel like data is such a blocker for so many, tool implementations Mhmm. And improvements in finance processes. What does practical data actually mean? Yeah. So I did let me just tell you a story first of, about data. So I was just speaking with an insurance company, CEO of an very large insurance company, one of top three worldwide, and we were talking about AI and data as a conversation topic always comes up. So you're not alone if you're thinking, oh my god. Our data is not there. We can't really, take advantage of AI. Everyone's in that boat. So don't feel like, okay. I'm behind because and we can't do it. Even in that insurance company, he he was saying actually, Christina, of of Fifth Floor, I have a whole floor, of people sourcing out data, and they've been there for years. And we don't see this ever ending, that data cleansing data, sourcing project. Right? Because there's just even more data being created at the same time. Don't let that perfect data, gonna, again, stop you, from implementing some of these solutions. So what does practical data mean? I'll give an example. We have, a firm that a finance team we're working with that was looking at something very specific around expenses. Right? So they say, okay. We have expenses coming through, need reviewing, compliance, checking with our policy, etcetera. And they they did have a system. They had the SAP system. But the way that the data was coming out of that system was not it has, you know, various legacy from many years ago, had different sort of entities, acquisitions, companies. So the labels weren't quite right. The categorization was a bit messy. People putting free text, descriptions, etcetera. So even though you had a system, they had a system in there. It was quite hard to actually work with the data. But what they did was they didn't let that stop them. They did use AI on top of that data. They exported and sort of used API to extract that. And with AI, you can't get more than 50%, sixty %, seventy % there with, you know, the the remainder 30% still needing human to go through, source it out, you know, check, etcetera. But that's significantly better than a %. So instead of waiting for the clean data, the clean categorization, etcetera, you can start with the data you already have. And then what they did in parallel is, yes, they source out their categorization, the, you know, the new policies, how you submit your expenses, etcetera. And so going forward, that messiness would reduce over time, but then you're not held up on that. Super, super clear, and great example, Christina. Thanks very much. And what about, that fourth one there on the right hand side kind of from buying tools to driving adoption? What does it mean to sort of actively drive adoption? And do you have any kind of examples of teams that have done that really well? Mhmm. So adoption, let's just talk about what that means. Adoption in fact, adoption management, change management is a whole new discipline. And it's not that new, but it's kinda emerging discipline that people are starting to take more and more seriously. This is all about change management in this sort of day and age. There's new technologies coming out, new market dynamics, and how does the c level how does the leader in the businesses really drive change so that every employee is taking, you you know, is coming along with the change. So in this AI case, as an example, is instead of just training or in addition to training, are they coming back to their seats and really leveraging AI in on a daily basis? So we talk about adoption is different from training. Training is sort of passive. Adoption is having a program where there's AI champions, in the organization, in the peer group that really advocate for that. It's about, you know, very personalized, approach to to AI, for example, your personalized prompt. Yes. You've learned how to prompt, but do you have personalized prompt to tailor to your workflows? And that's what adoption programs do. Mapping out all your workflows. Right? Okay. You can actually write a smarter prompt that's probably 200 word long, but once you have that, it will take away that workflow instead of just, you know, you've been up to a a training and then, you know, we go back to Tech Tipiti and ask the ask the question, can you give me this? So that's what adoption is. You ask Nat you ask about what a good example looks like. We just finished off a project. So because we one side of one half of our business is running adoption. We just finished a project with an investment firm. So their finance team is five people in the finance team, specifically. We run a ten week, adoption program for them. What does that mean? They had Microsoft. So for them, they bought Microsoft Copilot just like ChattyPC in Microsoft. It's in their Outlook, in their Word, in their Teams, meetings, etcetera. And then mapping out again, sitting next to the five individuals. These are your personal prompts. These are your use cases. These are your workflows. At the end of that, usage was was a %. So that's a big, sort of very strong foundation. Let's put it this way. They now have a personalized library of more than 80, prompts. So they can really use for the specific tasks they know they do on a very regular basis. They now have AI champion within the team because it's not just top down because the CEO, CFO said so. It's gonna come from bottom up as well. So they have someone on the team who's just genuinely really fired up about AI. Now she's the one that carries that going forward of the ten week program that when, you know, new features come out, hey. Did you know we can do this? Oh, I saw, you know, Nat was using that, prompt of clicking that button or using AI this way. Hey, Tom. You can look through it that way as well. That's what adoption a good adoption looks like so that that that starts to embed into everyday workflow for, for a long time rather than just, you know, a training period. Okay. Yeah. Great example, Christina. And, I like that you've you've pended on me to be to be the leader in our adoption. I'm not sure that's the case at that chat, but I do try. Cool. Let's get into this framework, kind, that you that you've got for, our attendees because, obviously, I think we hear about new AI tools for individual things every single day. How do finance teams go about selecting tools? And Mhmm. Can you talk us through this framework, what it what it means, first of all, and then, what the individual factors, what those definitions are as well? Right. Okay. So, there were last count 11,000 AI tools out there. I'm sure there's more coming out every every week. And we have a a team that's, you know, tracking that. So it is overwhelming. This framework very similar is a similar framework that many CTOs would be familiar with. So I know from a CFO perspective, buying tools is not in your, core responsibilities, but really CTOs, have been doing that for a long time, but we're looking at how that applies in this environment here. So, typically, for as to mid market, this audience group here, mid market finance team, you'll be looking at buying. So what's build versus buy? You buy something off the shelf from a software provider like Agicap. They have the best of breed solution for a specific pain point you're trying to solve. That's buying. Build, typically, typically only happens at very large enterprises. They want those something very bespoke. It is costly to do. That scene so that's why I bill and buy on on this slide here. And for mid market, typically, 99% of the time, you're under buy. Now AI is changing that slightly. Now if you could go to the next slide, AI's market has got a third column, which is configure. Configure something that's prebuilt. What do I mean by that? When you're buying, a solution, quite often, because AI so AI's intelligence comes from the context. Right? It comes from your data. It comes from, or even better, you're not just your numbers putting out a finance system, bigger contextual, maybe your emails, maybe your files elsewhere, maybe your core, core discussions. Right? These are, the context that AI can really thrive on. That is quite hard for, just buying off the shelf solution to achieve most of the time. So that's number one. So a lot of teams now looking at, okay. How can I buy, but then configure or tweak it, tailor it somehow? The second reason why there's a new category that I wanna talk about, this configure here, is the cost of build has just dramatically reduced. Right? I think before, if you was to build anything AI or, you know, software solutions even, it you know, you've got whole development team going, you know, machine learning engineers, data scientists, and most businesses, if you if that's not your core capability, you should not be investing in that. But because of the AI the way the AI is set up, there's a lot of agents and agents being prepackaged, prebuilt, right, into for example, in Microsoft environment, there's libraries of now prebuilt AI agents. You can really and you've seen chat g p t. Your own GPTs, you can almost like app stores can get them. So there are now teams that really use leverage that piece and take the prebuilt to start configuring because they know by just adding doing the last 10% of the work to configure it actually drives more than 50% of value because that's where the contacts and intelligence come from on their own systems. Okay. So, basically, what we're saying is that the old way of, buying software has gone out the window. It's not just build versus buy anymore and build only the domain of large enterprises. Actually, with AI, it's buy and then configure, a little bit yourself. So it's used the two in tandem. It's, well, let's say it's a mix. Right? There's no it's it's not that one solution is right or or the other is wrong. It's, you know, everyone should consider all three. I think in this mid market, it's either buy or cons consider. Okay. Very clear. Thank you, Christina. And then let's get into the kind of the practical things. So so I think everyone wants results. You know, finance is very results driven, part of, of a business. Where can, finance teams get quick wins today? Where do you see AI as having the biggest impact for, finance teams, today? We get asked this quite often. You know, where can we get the best stand out? But, typical, it this you know, people are familiar in the AI space would know high impact use cases. I think it's the other way around the title, actually. So on the left hand side is high impact use cases. These are repetitive tasks, quite structured data, time consuming, and you'll be doing that over and over again, especially monitoring a lot of work. So generative AI is anything to do with document, reports, research type, and sort of topics are really good. The more traditional AI machine learning is really good with data, obviously. So that's, still very much where you can best leverage AI, where the low impacts or lower priority, let's call it use cases because AI is really developing every day. It's on the right hand side. They're one off tasks, or unit once you do every course every year, you know, may not be, kind of high priority. There's a lot of judgment work of reasoning behind. So, again, AI's reasoning powered to GTE and clawed, etcetera, is, you know, growing exponentially. But there's been data still suggesting that, you know, for strategic judgment and decisions, it's better to use. Humans are are still better in that case. Good to know we're still better at something. So, yeah, just to reiterate that, the low impact, cases are on the right guys and the high impact ones are on the left. So still that kind of repetitive, tasks, you know, just a bit of, scenario planning decisions for anything that's, got lots of data, I guess, and lots of standardized data is very high impact, whereas one of tasks, or strategic judgment calls, you should probably, involve yourself as much as possible. And then do you wanna just quickly talk us through this, AI agents for late, AI agents library that you put together, Christina, as well? Yeah. So this is a a a very popular, kind of request. When we work with CFOs, they often ask, you know, because you work with those all these finance teams, what are the top use cases? Where where are they? So this is the top library the library of the top use cases where AI and AI agents, prebuilt agents have been used. This is, you know, our, our proprietary data. I'll just take some of, this to walk through. There is, invoice review assistant at the very bottom right, the last one. That's what I was talking about, when expenses were going through. There's very similar use case in invoice review. So we actually have one company that is a supply chain submit, sorry, logistics provider, fast growing. They had a few thousand invoices going through every month. And using AI, they could review a lot of duplicates, again, check compliance, etcetera, and they flagged once the system up or the agent was up, they flagged 21, you know, duplicates, within the system in the first batch and, you know, started really dropping down their AP cycle as well. So these are the, you know, we can take a screenshot of this. These are the top use cases we've seen with finance team. Okay. And if we just kind of dig into that a little bit more, are there any kind of here, we've kind of got it feels like we've got all the use cases that the finance team might go through. Mhmm. Are there any quick wins you see finance teams as kind of crazy not to be taking advantage of? I know quick wins is always what what we we all we all here for quick wins. It a quick win on a slide can be very different for for every company depending on, like, situation, data, etcetera, etcetera. But I just call one out. There was, the, the budget variance explainer. So that's budget or board report. If you are a business, and we just came off a call with one, who has a private equity sitting on their board, and that's up to budget review, board pack, review, type exercise every quarterly, and that's where they can use AI to this AI agent to first pull out their own numbers from systems, run the first pass on, you know, what the variances are against budget, etcetera, and put it into a format, template so then the team can go and really comment on that. But what's more interesting is they also so that was step one, getting the data. So prepopulated four packs already, or budget sort of, review already a big step, step up. Step two was they then started recording meetings in a discussion. So as the CFO was sitting down with the different functions or they were reporting or they were having group meetings, leadership meetings, those meetings were recorded. And then when they were discussing, you know, q two budget versus actual, for example, a lot of the insights from meeting transcripts can start populating into their, again, predefined, budget, in this case, also, ballpark reviews. And that's where, again, it's not the final. I was just saying, AI does make mistakes. It still does. It has to be always be checked. But getting the team there 50% of the way, 60% of the way, and then they can then really comment on that. That was a huge win for them. That's just quite a simple plug in some meeting and some number, data sources for them. Yeah. It makes a lot of sense. I think also one of the quick wins I've seen as well is definitely on some of those. If we think back to that previous slide where you're talking about this highly repetitive task, stuff like reconciliations, which is incredibly painful to do in most Yeah. ERP systems, can be really sped up by AI at a sort of phenomenal rate. Mhmm. Because the data is so repetitive and the matching is so obvious, for an AI, but quite time consuming, for a person to do an ERP system manually. I'm just gonna pause here to say people, if you do have, questions, stick in the q and a, particularly on this kind of range of AI agents. It's quite a lot to digest on one slide. Please do stick them in the q and a, and Christina and I will come back to them at the end. And, likewise, I see some questions coming in the chat as well. Try and stick them in the q and a, and then we can deal with them, all at the end. So common mistakes, what not to do. Christina, let's get into, some common mistakes, maybe with some examples of, things, going horribly wrong. Don't call out any names, but we love to hear that. What are some of the most common mistakes you see CFOs particularly making when implementing AI, and how can the people on this call, avoid them going forward? I would just call out number one. That is the biggest, and it's not horribly wrong. I think there were some data either Gartner or one of the big consulting firms have quoted 20% of the AI pilots are wasted right now. As of today, so far, you know, not going ahead, you know, not going anywhere. I think that's a highly underestimated, percentage. That 20% is gonna go up to even more. People were in experimentation phase, which is great. I think that pilot purgatory, which is you're forever stuck in that pilot mode and is now moving really into production, moving into scaling, moving to actually helping the teams. That's quite common, that we see across the board, across different industries in the finance function. This is a combination of, you know, just the market moving very quickly, but also uncertainty. A lot of CFOs are saying, well, I don't know, you know, if this was the right pilot, right tool to pilot for us. Maybe there's a better one coming out next week. Maybe there's a cheaper one. Maybe, you know, how does that link into maybe our existing system? We're gonna add AI later. You know, all of these questions is leading to pilot purgatory right now. And we again, coming back to the 1% everywhere concept, the mindset is, again, rather than looking for the perfect solution, having that mindset that in AI, things will just move. You you're not buying a solution for the next five years. It's very unlikely. So, you know, look at the vendor lock in situation. You know, make sure you're not locked into a vendor, but don't let that, again, hold you back. Okay. Super, super clear, I think, on the pilot perpetuity and, the vendor lock in. If people want to, avoid pilot purgatory, do you think that's really linked to making sure they have an AI champion is what what are the key ways to avoid, the pilot purgatory? There are many ways AI champion for sure, AI strategy to start with, you know, from top why why the why. Why are you doing this? Typically comes down to the why and the what next. So when you think about pilot going through is why are you starting the pilot in the first place? Why are you looking for a tool or a solution or building something in the first place? Why are you trying to solve a business problem? Another one that people have forgotten is the what next. They hadn't thought before they started the pilot. They hadn't really thought about, okay. If the pilot was to test a, b, c, d or, you know, experiment on, you know, a, b, c, d, what that would look what a, b, c, d are in the first place, and then what what does that mean for us going forward? Does that mean we buy? Does that mean we you know, not if they don't meet? So having thoughts about this and plan it out and agreed upfront, scoped out, let's say, at least, scoped out upfront really reduces your chance of wasting a lot of energy on a pilot and then because I hadn't thought about, okay, we're gonna do it after. It kinda just was a nice little experiment. Okay. Nice. Thank you for that. And just on that number two, waiting for perfect data, we obviously talked, quite a lot about that earlier, in this webinar, and how to avoid that. Do you know just, any LLMs that are kind of really specialized in in cleaning data? I feel this is one of the the biggest thing that holds not just finance teams, but also sales teams and marketing teams back, when it comes to getting more from AI. Oh, that's such a good question, Ned. I wish I had an answer. I'm sure there are there's million no. Millions would be over at it. There are lots. I know it's frustrating. We're gonna go 11,000 tools, Christina. Lots of tools out there, but I think it's more, a lot of service per providers as well, you know, data cleansing, consulting, service professional services. I don't have an answer for the perfect l and m, I have to say. It, you know, it, again, depends on your data type and where that lives. If people in the audience have heard about it, please put in the chat because then we can all share. And then I'm back. Share, your your favorite AI tools, not just for data, but if anyone's seen a good data cleaning AI tool, then, we'd love to hear about that one particularly as it can often be the key first step. But, yeah, keen to hear both of us what everyone's preferred AI tool is, whether that be Claude versus Chat GPT, or something more exciting. Please do stick it in the chat. And then let's get into the philosophy part, Christina, why building capability matters more than just buying AI. The competitive edge isn't having AI. It's having a team that knows how to use it. I guess if I were being really, cynical here, I would say, you know, this is a nice theory. But what's actually the starting point? How do you apply that in in a team context? Mhmm. First of all, I need to just set the scene on what we mean by building team capability. I don't know if if you guys have come across. There's a study that was done by BCG and, I believe, with MIT and a few of the sort of US, universities. So they put through, over 700 BCG consultants into an experiment of, you know, a group of them were performing tasks with AI and a group without. It was quite interesting data coming out and as expected, you know, people with AI, tools were performing better, on some of the tasks. But it was really interesting. There were certain tasks that people with AI, BCG consultants with AI, were on average 16 worse than the ones without. And that just goes to show, it's not the AI itself, is if people know, first of all, how to use it and when to use it when coming back to limitations. Right? You are, at the very beginning, asking about what use cases was, you know, viable versus not. In this particular study, the ones the questions or the task they asked the BCG consultants that were really under underperforming were questions like they had a report of, kind of figurative company, to analyze. And then, you know, write a, again, a figurative CEO memo to recommend to the CEO what to do with the strategy, and you're looking at the past data, etcetera. And because, ironically, because the consultants had access to AI tools, they switch off their brain, really. That's what happens in that kind of the post interview. We're talking about this is and the people were less, rigorous in terms of analysis, critical thinking, etcetera. And that, again, just shows it's not just everyone could use the tool, and you maybe had the tool. It was value destroying in that case. So I just wanna set that scene to know what capability means, or team. And coming back to that's almost the box number one here on this slide. Right? Understanding AI trends and limits. Because it's still for a lot of people that's a bit of a black box, how it works, etcetera, especially not in the technical field. So how do we do in practice is there's there's training and there's adoption. Really, there's nothing compares with, people just use it every day. If people can be using it, they will encounter it. Oh, right. That wasn't you know, I I asked this question. It was giving gibberish out, you know, to me. They've tweaked it this way. Oh, I didn't know I can click this button now on my Outlook. It will start scheduling, meetings for me, for example, or getting my agenda meeting agenda out. Things that people start discovering on their own if you sort of set them off on the adoption path. So that's number one. Number two is just very practical, prompts, with the interaction, really workflow design here. Again, because people are not just, you know training is an an adoption are very different. If you give, your users, employees more personalized prompts and workflows, AI powered, it really helps them to use it. And then process integration, use case identification. The the box number three and four is where perform where we start to guess performance. Box number one and two is where we see a lot of companies start from productivity. That's the shift from productivity to performance. We you know, when we use ChatTBT and we use Copilot, you know, etcetera, etcetera, it quite often is a productivity tool. It's a personal system for you and me. Great. But that's individual level. At company level, you know, c level, they are looking at how do we make the business transform. Right? And how do we move the needle in addition to saving some time on employee, work. Right? And that's when process integration, use case identification, so going deep by function, by use case, that becomes interesting. And then number five is that loop. Right? It's continuous. You go continue to identify next process, next use case to use AI, and that's the 1% again. The compounding starts to happen. So if you have that loop, your workforce is naturally being kind of immersed into this. You know, they're doing number one, number two every day on personal productivity, and then they see their function being transformed, ideas coming through. And for some of the best company, you'll still see their the the employees themselves come to the executive level about, hey. Have you thought about this? The, you know, ideas that the probably the top level hadn't thought about. Super, super clear. Great story about the BCG consultants. I definitely see that as one of the main threats of AI is that we will just switch our brains off. And I guess, it kinda goes without saying in this slide as well. For CFOs, they really need to be involved in this process throughout. If you wanna achieve that one percent compounding in the continuous experimentation stage, you've gotta know where to keep experimenting, where you're getting more value. So it kind of starts with, deciding which processes, need the most focus and, in steps three and four, and then also staying on top of exactly what's going on, who's having success where, and where you should get the team and the champion to continue to focus their energy. Cool. And now we're gonna switch to embedded AI. So how to unlock quick wins in forecasting financial planning using AI and Agicap. I'm gonna show you guys a few examples of the AI use cases within Agicap. So Christina's talked a lot about, LLMs, how you could begin to drive AI adoption, identify use cases, which tools to go for. You've obviously got AI first tools, and then you've got your existing tech stack. I think, Rosa, you've asked a question there about, balance sheet reqs into an ERP like NetSuite. This is layering in AI or embedding AI into existing products. And Agicap is doing this every single day as we continue to develop our products. So I'm just gonna run through a few examples with some quick videos of where and how AI is working within the AGICAP tool. Within AGICAP, we have this AI assistant, which will, or does enable our customers to do direct data analysis. So this is helping you ask questions, helping you build graphics, and explaining how to do so. And here, you can see a quick question's been posed. If I just pause the video, will I have enough will I have enough short term cash if I make a payment of a hundred and £50,000 to explain how this is working? This is taking all of the client's data in a secure platform, so there's no risk that you have in the same way that if you just pose this question onto your own, OpenAI account. We'll have enough short term cash to make a payment of a hundred and 50 k in three days using your connected, systems and your connected bank accounts to carry out this calculation. And then here's it. It's explaining how it does that, and it's giving you advice based on that. You can use this, same AI assistant to then decide what action you could take, also build graphics off it, to show the consequences of what you did if you made this payment, at a later date versus now. So there you see just one example of how we have an inbuilt LLM focused on our own customers' financial data. I think if I go to the next slide, we also have, one of the great AI use cases, I think, that we'll talk about. You could also use the AI assistant for product tutorials. So, historically, this is obviously the domain of the chat function where people invest in new tools. They ask in the chat, how do I do this? Or they ask their account manager. Actually, with AI, which has an overview of all the different use cases of, all the 8,000 Agicap customers, We're now feeding that information into our AI system so that our customers can ask these questions directly. How do I make a payment on Agicap? Basic functional question that will be repeated a lot, and the AI assistant will explain very quickly what you need to do and how you can do that, through this, quick, set of instructions. So really speeding up, the, adoption of our own tool and how to do, how to execute certain actions within the AGICAP platform. We also use, AI human collaboration in the AGICAP chat function. So that same rule applies where not only do we filter certain answers, to, give AI answers or where we need human support, but we also have, suggested answers being designed by the LLM for our, people on the chat function provide to our, audience, our customers. Then if we get into the more exciting bits, around categorization, for example, within Agicap, and data consumption, for our very big customers, we're beginning to talk about very, very large datasets, and also international teams. And here we see, for example, one process being used in one geography and, another process being used in another geography. And then they want to apply the same processes across new markets. Maybe they've done an acquisition. Maybe they've onboarded the French or the German part to, their new team. And here we can just directly use an LLM to translate this template that they have for their UK Team into, something that the French team can directly use I guess. And here you can see the template created in French and in Italian, and I think there's also a German one that we see. So just three templates immediately created in three different languages, just at a click of a button. Then I think my favorite example, personally, is kind of the use of, machine learning to categorize data. I think, one of the stories I heard from one of our CS guys was we took three years' worth of data for one customer and transformed this, into automatically categorized data. So with new categories using Agicap, machine learning based categorization module, which were then checked with the customer. And for now, every new transaction that goes into AGICAP, and this is thousands per month, about 95% are automatically categorized in a satisfactory way, based on the categorization, that this, controller built with the machine learning, thing. And I think the key difference that really kind of shows the power of AI, they had one FTE working on this over that three year period, and it was their sole role. They had so much data that was to categorize all the information. Now they work with the AI. The AI does all the categorization, and they do the analysis. So it's kind of and the AI did the analysis in hours, you know, not years. You're talking about three years' worth of data that can be 90% categorized in hours. And you talk about, Christina has mentioned, you know, getting 60%, seventy % of the way there. I think categorization is one example of a really data heavy task that AI is unbelievably powerful for and can enable all the finance people on this call to spend much more time on actively, doing analysis rather than filling in, data points on an Excel spreadsheet. So that brings us kind of to the end of the webinar on the the discussion points. Let's get into the q and a. Have we had some questions? I did see, there was a long one from Christina Kelly, which maybe we can kick off with. From an audit perspective, is there not a risk that overuse of AI agents may bring into question the CFO finance team's accountability, liability, and governance? What would you suggest, be considered to manage this risk? Christina, this sounds like one for you. It sounds, like the kind of question, you must deal with, much more than me. So, when it comes to using AI agents, how can CFOs and finance teams feel confident that they're meeting accountability and live liability and governance risks? Oh, such a good question, Christina. There's a lot of vendors and, tools out there who don't have an answer for that, unfortunately. So, all from an audit perspective, the AI agents that are used need to be explainable AI and, you know, trackable, and and there's a whole, you know, separate responsible AI piece, to do. So in terms of what vendor you use, it's really quite important for you to be able to explain what had gone through the AI decision making, to get to the answers. That's I guess, Chris, if you have a specific example you had in mind that would be challenged, then we can answer more specifically. But, typically, that's why, you know, in big vendors like Microsoft, they they have built in, audit explainability, AI explainability in their products. Super clear. Thanks, Christina. And, yes, that is something you see in a lot of AI tools. Always check the terms. I think the same applies for security as well. You know, if you're getting, a different, OpenAI license, I think the more you pay, the better for security. So you need to check the terms and conditions, on the standard LLMs as well. Are there any other questions? I see in the chat we've been having a good discussion about AI tools for balance sheet recs within, NetSuite. And BlackLine BlackLine is a good tool. I would agree with that. Are there any more questions? P please do put them in the q and a. We've got a little bit of time. But if not, I'll just run through, or maybe, Christina, why don't you run through this, this offer for all the attendees on on this webinar? Although I know there is a limitation to the offer. So why don't you quickly explain that? Right. So we wanted to make this practical, this, webinar session for every practicing for practical for everyone. I know there's a lot of hype around AI, different tools, different vendors, etcetera. And I put out earlier, a library of proven AI agents that been built for finance teams, etcetera. So we're offering here really online call to do AI agent matching. What does that mean? It's basically matching your, situation to, already use cases that your peers are using and tailored to your specific setup. So once we've had a chat, then we know exactly what where you know, what system, what tools, what workflows, what, you know, size of team, etcetera, then we can match you to an AI agent if that's some that if that's a route that you're going down to. Normally, that's a a paid engagement, but we're making this free for the first twenty sign ups through our partnership here with Agicap. So 20 sign ups. Grab your phones, guys. Normally, a 1,200 pound engagement. Grab your phones. Get that QR code. Book yourselves in with Christina. It wasn't easy to get her on this webinar, so, I highly recommend you move as fast as possible. And if you wanna learn more about, you can book your own call or demo with me, or just add me on on LinkedIn. I'm very active in that. And if you have any questions about what our product can do, I'm very happy to respond on chat there. Or, if you wanna ask them directly, you can use the QR code here. And then last but not least, we have a couple more events coming up. The one next Tuesday is focused on AI times ERP. ERP is obviously, I think it's the biggest software market in the world, but it's probably the most prominent system used by finance people. We've had this discussion about NetSuite on this webinar. On Tuesday, I'll be running another webinar with two ERP experts of fifty years experience combined, to get into what AI is gonna do for, the ERP market. There's quite a few new exciting ERP tools that are AI first, which we'll be discussing in that webinar as well. And then Christina will be back with Agicap on Tuesday, July. If this webinar is not enough and you wanna do something in person and have discussions about how finance teams are using, AI, with people in the same condition as you, position as you, over some drinks and some food, we will be, continuing, this discussion with Christina on the July 1. You can also sign up for that. These slides will also be shared. I appreciate this is a lot of QR codes at this point. But so we will share the slides, and you can book in, whichever event you'd like to come to, yourself there. So book yourselves in with Christina. Are there any more questions? I have put Christina's QR code up again. Malcolm, I'm glad you enjoyed it. Rosa, glad you enjoyed it too. Thank you for everyone for your time for tuning in. Christina, we've got one more question that's just come in. Alright. Okay. Chris, have any major regulators published certain guardrails or guidelines when selecting the right AI vendor for our for organizations? Oh, there's the EU AI Act. It's not really guardrail for vendors per se. And JPMorgan had a recent, open letter. I don't know if everyone had seen, about calling for more regulations and more not regulations, but, for calling for vendors to trade to prioritize security over speed. I know that, you know, software vendors, we do, you know, wanna be competitive, you know, in the market with, you know, product launches, features, etcetera. It's very competitive market out there. So there's that. Christina, to answer your question, I think AI may have, you know, faster references than me in this case, but I know there's the EU, AIX and JPMorgan's open lesson may point to a few things. Super clear, Christina. Thanks. So I think that is it, with yes. No more questions. Just lots of thanks. Well, thanks guys for attending. Great to have you all here. You know where to come if you want more information on either, Agicap or Christina. Hopefully, there's no spots left for Christina's, and you will be adopting AI at rapid rates across all of your companies. Thanks, guys, for attending. Thank