Why Now Is the Time for Treasury to Embrace AI | Treasury Careers Podcast

Mike, CEO, The Treasury Recruitment Company: Welcome to this week’s Treasury Career Corner podcast, where I interview treasury professionals about their treasury careers. Each and every week, I talk to treasurers about how they built their careers, where they are now, where they see both themselves and the treasury profession going to next. Let’s get on with the show.

Mike: Today’s Treasury Career Corner podcast is with James Kelly. Now, what you’re gonna hear in a moment is my original introduction. So I introduce James back when I talked to him, he was the SVP of Treasury Risk Management and, and Insurance at Pearson. So you hear that. Then we actually do the catch-up at towards the end, so listen for that.

Mike: It’s where now he’s co-founded a company called Your Treasury, and they transform how multinational treasury teams operate through AI and cash flow mastery programs. So listen to today’s episode, catch up with James, what you hear at Pearson, then we go up to date. Some great value there. He talks about his passion for AI, and then his passion for AI keeps going, um, into his new role.

Mike: So enjoy today’s episode

Mike: In this week’s show, I’m joined by James Kelly, the SVP Treasury and Enterprise Risk Management at FTSE 100 learning company, Pearson. During his time at Pearson, James’ team have established a reputation as leading innovators, winning awards at the ACT’s Deal of the Year Awards for Pearson’s debut social bond issuance, as well as establishing comprehensive in-house bank structure, employing high levels of automation to drive efficiency.

Mike: Pearson also uses TIS CashForce, which is a leading AI-enabled cash flow p- lasting solution, which James will explain a little bit more about. Lots of words, I’m trying to say this all at once. But we’re gonna go back to the beginning of James’ career, how he first discovered finance, treasury, and then bring it up to date, and then we’re gonna talk about some of the passions that he shares as well for AI and technology a bit later in the show.

James Kelly, Co-Founder, Your Treasury: James, over to you. Enough of my voice. Over to you, sir. Sure. Thank you, Mike. So I started out as a bit of a, a finance geek. I did economics and French at, at uni. Went, did a couple of stints at investment banks in, in my summer holidays as internships, and then was looking for a job in September 2001, which wasn’t a great time to be looking for jobs.

James Kelly: I started thinking what else could I do, and applied for a whole range of things. And I then got a call, would I be interested in joining Kingfisher PLC as A group accountant. They were looking for someone who could speak French. And at that point I’d had four years where I hadn’t used my French. I’d left uni having thought, “Okay, French is gonna be a big part of my life.”

James Kelly: And, and so jumped at that opportunity. Thought, uh, I just got to the stage where I was like, I’m going into clients for three weeks at a time, getting to know them a little bit, and then you move on and you go to the next one. And I quite like to get my teeth into something and, and really be able to add a bit more long-term value.

James Kelly: And, and within about a year I was working on the cash flow, which was considered the, the most technically, uh, challenging of the tasks. And through that I was asked to help the treasury team with a few things, particularly transition to IFRS, and I was helping them with some work on derivatives and things, largely because I was seen as this sort of slightly geeky guy who could probably get my head around what was going on and teach people how to do sort of balance sheet reconciliations and that kind of stuff.

James Kelly: So I did that, and around that point, the then treasury accountant decided to move on and I was asked if I’d like to move across. And so having been thinking I’d quite like to specialize in something, I’d actually been thinking tax probably, I thought, “Why not? I don’t have any particular ties to anything.”

James Kelly: And I was incredibly lucky. I had two really top treasurers in the team in, in Linda Hayward and, and Nick Fevier. And actually that team’s amazing. You’ve got Grainne, who’s at Walgreens. Yep. Gary Burton, who’s gone on and done some really great things. So it was a really excellent team, really thoughtful, competent.

James Kelly: People really looked out for each other, but were also really clear about what they could and couldn’t do. It was a really great grounding and I think one of the things that I… In fact, two things I took from that. The first was just how incredibly generous Linda was with her time and, and helping upskill me.

James Kelly: And they joke that a six-year-old asks 75 questions a day, and I was very much in that camp. And it was like, Treasury is completely new to me. I’m trying to get through year-end I had two brand new green people when I started, both pretty much out of uni at that point. My first instinct was to try and go it alone and, and then get them to do little bits, and I tried that for about a half a week and realized that’s never gonna work.

James Kelly: I’ve got three people on my team. I’m gonna have to make sure that everyone’s carrying something, otherwise I’m gonna fall over. So that was a really good lesson, and it doesn’t matter, you know, who you think you are, the, the reality is you need to work with others to, to get things done. And I think Nick was brilliant at just never closing anything down.

James Kelly: Everything was… Everything’s always possible. It’s just you’ve gotta find a way of getting there. So I moved to, moved to Sky And I moved primarily to get more front office exposure. So the front office team at, at Kingfisher were pretty well set. It didn’t really feel like there was gonna be an opportunity to move across, and I wanted…

James Kelly: I knew that if I wanted to move up, then I needed both front and back office experience. Around that time, my, my first child was born, and it quickly became clear two things. One was, so Mike moved on, was replaced by Simon Morley, who’s amazing, and was incredibly thoughtful and generous with his advice and guidance.

James Kelly: But that, but Simon’s experience was on the front office side. They decided they were gonna bring in a designated treasury manager. Mark and I would become back office. Yeah. And it was, it was a long commute and everything else. So it felt like it was the right time to, to try and find another role, and I had a couple of offers, but the most interesting was with Rentokil in, in Crawley.

James Kelly: And I remember, like sweating on that decision ’cause I had an offer come in, and then I had this Rentokil thing, which looked like it was probably going to happen, but I was a bit earlier stage in, in that process and trying to decide do I go for the, the known or do I gamble? And I’m incredibly glad that I gambled.

James Kelly: I joined as deputy treasurer to Nigel Roberts, who I think we had a really good working relationship. He’s incredibly structured, very good at prioritizing and has a really clear idea of strategy and communication. And can you explain for our international listeners, Rentokil, you and I know them, they’re household names in the UK, but if you’re in the US you might not know them.

James Kelly: So, so Rentokil… A- and actually, amusingly, Rentokil are quite large in, in the US, but they go under a variety of different brand names. Yeah. So Rentokil are effectively pest control and hygiene. So effectively you may see the Initial brand sometimes in, in washrooms. So if you go into a public restroom, often there are Rentokil hand dryers or soap dispensers and things.

James Kelly: Some of that’s been sold since I left, but a lot of that’s there. But they’re very large in, in pest control. And so in the US, they’re, they’re Terminix and various other brands. So Western Pest Control and- Yeah … various other things. Yeah, so I was there five years. It’s a pretty decentralized business. Very innovative in terms of the way that people work together, but there’s a real kind of culture of, okay, they’ve had this idea.

James Kelly: I’m gonna take that, and how do I make it a little bit better, which was really interesting. And I was lucky enough that my financial modeling skills are quite good- I was gonna say they, they w- they wanna talk to you. You’re, you’re a man in demand, which is great E- exactly. Exactly. So it’s one of these things, just trying to make sure that I really got involved in things.

James Kelly: And we did some really fun stuff, and I stayed there for about five years, and I reached the stage where I felt I, I wanted to work in a slightly different environment, somewhere that was a little bit more centralized, and also given the experience I’d had in DCM, I wanted to major on that a little bit.

James Kelly: Yeah. And so I moved to Associated British Ports, which at that point was the, the third-biggest issuer of US private placements in the UK. And then this role at Pearson came up, and it was a, “We’re pretty heavily c- decentralized. We wanna build a centralized model. We’re looking for someone with multinational experience,” and, you know.

James Kelly: Yeah. Cool. Yeah. S- and so we- And what was Pearson, what was Pearson like when you joined as a company? What was the situation there? That was seven and a half years ago. The first couple of years from my perspective was quite a kind of tidy-up exercise. I joined and we had over $2 billion of cash on balance sheet and over $3 billion of debt And we just needed to respect that our net debt was probably going to be about a billion or, or maybe a bit less for the foreseeable future while we went through transformation.

James Kelly: We had a couple of large, very ambitious efficiency programs that we run, we ran. I got pretty heavily involved in the first one, which was how I took on insurance, having looked after insurance elsewhere. Had a colleague who was looking after compliance who took this on and said, “I don’t really know where to start.”

James Kelly: So we agreed between the two of us that I’d take on that part of his cost saving task. And it was really interesting, quite challenging at times. Made you a better- You had… Oh, absolutely. Unbelievable transformation effort. Just really interesting because you’ve got this transformation going on in the business.

James Kelly: You’ve also got from a finance perspective, we were trying to transform all our processes, move to standard ERP processes, et cetera. And I was going into this thinking I need to make sure that I really think carefully about how best to design processes- Mm … and practices in a world where accountability, for example, for the balance sheet is going to be very diverse.

James Kelly: You can’t just go to somebody and say, “Tell me what’s gonna happen in the US,” because it’s actually, it’s five different teams in five different locations, and they’ll all know a little bit of the, the puzzle. And so trying to pull all that together and make sure that you’ve got a kind of coherent story and a manageable process, was really interesting, and that’s part of the reason why we put, put Cashforce in initially so that we could start to build more of the, the forecast ourselves, and then go out to the different teams and get them to interrogate it rather than have hundreds of stakeholders who we were trying to get feedback from and, and try and collate it.

James Kelly: And you’ve always… How come you’re so passionate about it? Where does that come from, first of all, if you like? So I, I think it’s one of these things. I think the first thing is I’m massively practical, so I, I don’t have a philosophical love of AI. If that makes sense. Yeah, it does. It’s a… I love it because from a selfish perspective, it’s a massive help.

James Kelly: It’s one of those things… A- and I I think what’s transformed is the ability to be able to do things yourself. So with things like RPA, you’ve had the opportunity to automate bits of processes. Again, if you automate bits in your TMS, you can make strides forward. But generally speaking, you’ll need some kind of consultant in.

James Kelly: Yeah. Whether it’s to build the automation in the TMS, whether it’s to build the RPA, and getting hold of capital to, to bring people in can be difficult. And one of the key challenges is that treasury is so varied that you’ve got, you know, 50 processes or something, not five. And so saying, “Okay, I’m gonna spend my budget this year in order to try and automate this process,” and then you go, “Brilliant.

James Kelly: I’ve automated one of my 50 processes. The other 49 are still crap. One is just absolutely brilliant.” It doesn’t really make sense. And so you need to have the tools to be able to go after a variety of different bits I think to date what’s happened is that the TMSs have been pretty good at automating processes where there’s a high degree of standardization.

James Kelly: Broadly speaking, we all execute FX deals in much the same way now. Interest rates swaps, there’s a kind- there’s a market convention. They’ve all got to be registered with a repository. There’s a very standard process. Once you get onto things like P&L reporting or to some extent hedge accounting, I know you can do it in the TMS, but again, it’s quite heavily customized often.

James Kelly: You know, you get into territory where it’s quite a high degree of customization for each individual firm. That’s not very easy to do in the current business model because, A, there probably aren’t enough consultants out there for everybody to constantly have someone building things. Mm. And, and B, you never have enough budget.

James Kelly: So where things are really starting to unlock is by combining different tools, by using a large language model with Python or a large language model with Power Automate or in a variety of different contexts, you can really unlock a lot of value, and it’s happening at, at, at pace, which means that you can, you can now start to go at things that mean that you’re able to get to a position of ease.

James Kelly: And I think the position of the treasurer hasn’t necessarily been one of ease over the last few years. Mm. Most of us still struggle to get full visibility of cash balances, let alone build a coherent forecast that covers everywhere. The reality is that there’s a lot of satisficing decisions that we have to make.

James Kelly: We’ll focus on the key markets because that’s what we’ve got time for. We can’t focus on everything all at once, and that means that inevitably you’ve got some blind spots that you’re not able to keep an eye on. Mm. And the nice thing about some of this tech is that you’re able to really make a difference.

James Kelly: We’ve now got all our, um- All our banks’ websites linked to a web scraping tool. So we now pick up whenever their credit ratings change. That automatically sends a message through to us to say, “By the way, this has happened.” So previously we had a model where we’d check once a month, we’d check what the movements were, and effectively we’d look at what was it before, what is it now.

James Kelly: Okay, it’s moved. If, do we do anything with it? That’s not necessarily very timely. You can set things up to monitor and those other things, and so I find that really interesting. So we’re gonna put your LinkedIn details in the show notes. I think there’s some great takeaways for people. What are the takeaways you’d like…

James Kelly: You’ve heard the podcast a few times, so it’s guys starting their careers or people a bit further through their careers. What are the takeaways you’d give them, James, from today’s show? I, I think three things probably. I think in terms of people starting out, try stuff, be adventurous. I think one of the great things about people coming into treasury is they’ve got the opportunity to bring different perspectives and different ideas.

James Kelly: Mm. And that’s really fresh. So ask questions, make life difficult. Ask, “Why are we doing this?” Because that’s how we get better. I think the second piece is if you’re trying to progress in your career and trying to take steps forward, don’t be afraid to try something different. If what you’re, if what you’re trying isn’t working, it’s almost certainly not you, but sometimes it’s the situation that you’re in.

James Kelly: So think about how do I take a side- sideways step, try something different, try a different tactic in order to get to where you want to get to. And then I think the third thing is in terms of makes somebody different and what makes you exceptional that gets you to the top, it’s an openness to working with others, trying new things, and really being thoughtful about…

James Kelly: It’s really important to make sure that you get the basics. You need to be tech- a technical expert. You need to be a really strong communicator. You need to be thoughtful. You need to be someone who people want to work with- Mm … and who, open to ideas, give people a, a chance and see where things get to. So as you say, be adventurous.

Mike: There you go. So I’m welcoming back James Kelly. When James and I… You’ll have just heard the podcast. It was amazing. We were talking about AI and treasury. James was at Pearson before. But things have moved on a little bit, so this is one of our update podcasts. You can go back through our archives. We’ve got other ones, which are great.

Mike: Have an original podcast, then you catch up with people. This has only been a year since I spoke to James. James, you’ve sat back in your chair, armchair. Not much has happened really since we last spoke, right? Yeah. Over to you, sir. Yeah. It’s been a completely quiet year. Completely normal. Yeah. I left Pearson, left the life of a corporate treasurer, and I seem to have become a, a teenage girl as I spend my time now waiting for the phone to call, the phone to ring, or checking in on proposals and things.

James Kelly: No. It’s, it’s been really interesting. I’ve set up my own business with, with a guy called Dominic Lynch, who’s- Uh … former treasurer of Bitpanda and GoStudent. We brought in a data scientist from Canada. The idea was we wanted to be a varied group to begin with- Yeah … and we wanted to be international. So, so the name of the company now is…

James Kelly: It’s called Your Treasury. Yeah. And the idea basically is we’re called Your Treasury because the idea is that we want to be a support network for treasurers. So if I think about my team at Pearson, I was really lucky. In the end, I was able to bring in a data scientist a- and I got carte blanche to, to do a lot of the work that I, I do on sort of automation and Python and things like that.

James Kelly: But that was quite a hard slog. I had a lot of fights with IT to get access to the tools that we use. It took a lot of engineering in order to be able to bring in the data scientist, et cetera. And really I felt that’s not accessible to most people. Well, they do A- a- absolutely. We all completely unfund- funded business cases always land well.

James Kelly: So it’s one of these things that this kind of offers the opportunity for people to learn about things, co-create, use our data scientists, things like that, so that they’ve got access to tools that maybe they wouldn’t have otherwise. So the way that we’ve set the business up, we’ve got… In Microsoft Azure, we’ve got all the key tools that we think people need in terms of large language model, Python environment, we’ve got automation tools, and effectively the way that we work is that we help co-create ideas with people, and then we deploy them on their network so that they’re then available for them to use.

James Kelly: Everything is designed to be very editable and something that they can maintain themselves, so that effectively you’re not having to wait for the market to bring all, all of these ideas to you. You’re not having to wait for a TMS vendor to bring something in. A- and I think that’s important because actually a lot of the things that drag treasury teams down are not the things that TMSs currently do.

Mike: Yeah. You talked about it. I’d re-listened to it. I said I’d re-listen to our episode, and to anyone that’s just heard it, they’ve just heard firsthand you talked about in TMSs, there’s standardization of data, it’s great, but actually in treasury there’s not a lot of standard data sometimes. And it sounded like you said, and again, you said there was a specific gap w- with these little challenges, and you and I- Yeah

Mike: have talked about this. Find a little problem in treasury, get buy-in, and then you prove its worth and then, oh, hang on, we’ll invest in that because you’re saving money. Oh, wow, hang on, iterate and iterate. Can you go- Mm … talk to, speak to that for a moment? Yeah, so one of the, one of the things that large language models like GPT and Gemini and Copilot, et cetera, are really good at summarizing information and translating between different methods.

James Kelly: Yeah. So, you know, things like account opening- Python and a large language model are quite capable of reading a PDF form. Yeah. They’re quite capable of sourcing information from databases and from their training and that kind of thing to suggest here’s how the form should be completed. Using Python, you can then get the form completed in the PDF.

James Kelly: Some of the models are allowing you to do that at the moment. We’re not quite there yet generally- Yeah … certainly for, with the enterprise ones. So you’re in a situation where effectively if someone sends you a, an account opening form and most of the information’s generic- Yeah … then you can just self-serve on that- Yeah

James Kelly: rather than have to go through, fill it in, and that moves you from the position of being doer to being reviewer. Yeah. And I think one of the big challenges for small treasury teams, and most are small, is that it’s really difficult to be both the in-the-detail doer and the strategic thinking- Big ideas

James Kelly: reviewer. Yeah, yeah. Exactly. And so the more that you can take some of these basic tools which, or basic tasks which really are only being done because there isn’t a kind of an obvious automation tool- Yeah … that allows you to translate… KYC is another example. The request comes in through email, through forms, all sorts of things- Like, it’s the same questions.

James Kelly: Yeah. They just need converting into a standard format, and then you can use automation. So there’s, there’s lots of opportunities to help people with what feel like little wins, but actually when you add them up- Yeah … add up to a, a big chunk of time. Um- And you talked… Sorry, not jumping. You, you also talked- Yeah

Mike: you and I have talked about this repetition, if you like. Yeah. A lot of repetitive tasks in treasury, and you’ve even described it yourselves that manual workload for treasury can be, say, 20%. Yeah. And that’s one day a week. And if you could then free that up, you’d have all that extra time. But what does that look like in practice?

James Kelly: Can you give us any examples? Because I know you have got some examples you worked on. Yeah. So the one that springs to mind, actually, is some work that we did on cash forecasting with a company. Now, this isn’t the sort of classic we put in the machine learning tool and nobody had to do anything anymore thing.

James Kelly: This was just we took the existing process, and we redesigned everything so that effectively we replicated what were the business teams doing- Yeah … right? What emails were they sending out, all those other things, to populate the… They were using spreadsheets. Populate the spreadsheets that were coming in.

James Kelly: And then for the central team, like, how could they make sure that the spreadsheets were received on time, automated chasers. If something looked odd, then we had automated validation, so it would send something through immediately to say, “Mike, I’m not sure that your submission looks quite right. Would you mind answering this?”

James Kelly: Flag it. Yeah. Flag it. And doing that immediately rather than doing it later makes a huge difference because you’re much more likely to get an answer there and then because people are in that mode. Whereas if you wait three hours, it sets a precedent, and you probably find it’s another three or four hours before someone comes back.

James Kelly: That’s almost the next business day, and suddenly you find that your whole process has been held up by 24 hours because of the fact that there was- One glitch … one glitch. So we took a process that was taking 40 hours a week across six finance staff, and we took it down to three hours, and the quality of the output was improved The speed of everything obviously significantly improved.

Mike: The exciting thing about where we are is that so many of the key questions that existed 12 months ago have been answered. Yeah. And we’ve already put your LinkedIn details, but we’re approaching the end of the show again, as we did the second time round. And I said to you earlier that I’d re-listened to our show, and I said we were really good.

James Kelly: We like… We were eloquent. It was about you were eloquent. Your reflections now and maybe from an AI tech sp- to wrap up today’s show. Yeah. So I g- I am gonna use an AI angle, because- That’s great … it’s one of those things that the higher up you go in, in an organization within treasury, the more important it is that you’re really good at getting the most out of your team.

James Kelly: It’s less about individual performance, it’s about the collective that you’re able to facilitate. And so you have to be good at delegating, you have to be good at working with others. And if you’ve got people that you’re not getting the most out of, it’s going to be obvious and people will start asking questions.

James Kelly: Now, we’re now in a situation where most of the AI tools are good enough that you should be using them. If you’re turning around and saying, “I’m not- I don’t have time to find out how to use this,” or, “This isn’t important to me,” you’ve got what’s effectively a free tool sitting there- Yeah, that can do it all for you

James Kelly: that can drive significantly improved performance. It is one of those things, it does take a little bit of effort to learn to get from I’ve experimented and I can do meeting notes and that kind of thing, to a level where it’s able to write reports for you in your style of a really high quality, being able to integrate data and being able to combine data sets and that kind of thing But if you’ve got somebody in your team who’s got amazing data skills, and who can speak multiple languages, and who can take notes for you, and who can follow up action points, and trigger actions, and all sorts of other things, would you really leave them sitting in a corner and ignore them?

James Kelly: Gathering dust, yeah. Yeah. You should use them. So- James… Yeah, go on. So I think it’s one of those things, there is a little bit of a call to action there, because I think we are now at a point where there’s a level of maturity- Yeah … that means that now’s the time to get involved. And if people still need a decent translator and someone that can be their guide, they should reach out to you, and you’re there to help them.

Mike: Absolutely. Thank you, sir. Thanks very much. Thanks, Mike. I’m welcoming back James for his third conversation with me. James, we’ll talk through briefly his career. I spoke to James 15 months ago, and what I wanted you guys to do was hear about, and I reached out to him recently and said, “15 months in the world of AI and tech, it’s a bit like dog years, isn’t it?”

James Kelly: It’s like so much changes, like, in just such a short period of time. Maybe if you would, James, we’ll get people maybe to listen to the first show, but just for simplicity’s sake, how did you first start? Bring us up to date, and then let’s talk from there. Over to you, sir. So when I was at Pearson, I did a course in data science, and used- found that incredibly useful, because a lot of the challenges that we had effectively were grounded in data challenges.

James Kelly: Either data that didn’t talk to each other, data in different pockets, et cetera. And I learned i- in the old world, how to do everything by hand. So how to, how to pick data up, join it, clean it, all the techniques. And then when ChatGPT came along, that gave me an opportunity to bring more of the team up to speed, because it’s a, it’s a data science tool that’s built by data scientists.

James Kelly: Yeah. A- and so effectively, all of the concepts that I’d learned, I could train the team in without having to actually teach them oodles and oodles of Python for them to be able to do that. A- and effectively, we had a lot of success and, and, you know, won various awards and things. And when I came to a position where I felt that, you know, I’d done as much as I could at, at Pearson, and it was time to hand over to the, the next generation to take it forward, that actually I thought, “Well, it would be fun to do this, almost make it my role.”

James Kelly: Yeah. You know, come in and, and help other treasury teams to do a similar thing, and use my experience as a treasurer to plug those gaps. Particularly because I know from experience that there are two key problems for, for treasury teams. One is access to IT resources is always difficult. And two, actually treasury teams are normally pretty busy.

Mike: Yeah. And so, uh, and not specced to have extra senior resource on tap. And so you felt there was an opportunity where people were, were taking on transformations or had major changes due to M&A or that kind of thing, to be a resource a- available to, to help with that kind of surge demand. That was coming on.

James Kelly: Uh, yeah. And so we’ve been able to help some really big companies, a lot of kind of multinationals, and they’re generally NDA’d, so I don’t tend to talk too much about names, but you’ll have heard of a, a, a number of- Awesome. You and I last spoke about this on, on tape 15 months ago. Do you think the, the doors to AI and tech are, are sort of slowly coming open a little bit more that, you know, the treasuries- Mm

James Kelly: and finance teams are, are… Oh, hang on. Yeah, they’ll talk to you. Yeah. There’s a massive difference. Right. So if I go back 15 months ago, effectively you had the early adopters who were either looking at trying to work out how could they do things. Yeah. And so a lot of what we did at that stage was coaching and kind of ideation and, you know, use cases and all of that kind of thing.

James Kelly: You don’t see so much about use cases today. I mean, there are still people who ask what are the use cases in treasury, but I think as people have got more comfortable and the technology’s matured, it’s become much easier for people to understand and think about how they can use it The other piece that’s important is the concept of the agent.

James Kelly: Now, agents themselves, I think, are not particularly interesting concept in and of themselves. But the, the key piece here is that effectively it, it puts the technology at the fingertips of the treasurer. Right. So even 15 months ago, if you wanted to plug together a, a, a complete workflow, end-to-end FX management, including recommendations and execution, for example, you’re probably still bringing in tools like Power Automate.

James Kelly: Now you can kind of do end to end within an agent wrapper without needing to go to IT and say, “Can you give me an extra license for this?” Give me a line. And, you know, a- and therefore, that, that puts a lot more autonomy in, in the hands of the treasurer. And when, when you say that, James, when a treasurer is coming to you, or treasury teams and things- Yeah

Mike: what is the problem that you, that they’re most likely to bring? What are the, what are they giving you in a, in, in a, in a plate and you go, “Oh God, yuck”? I think it’s almost always a data problem of some description. It’s either we’ve got data in our TMS, but it’s not connected to our reporting. We’ve got data in our TMS and our ERP, and they’re not talking to each other.

James Kelly: We’ve got rubbish data coming from the operating companies. We want to improve our cash forecasting. It’s almost always related to how can we get better insights about the world in which our treasury operates- Yeah … in order to drive better outcomes. And last time we spoke and we, it was the headline of the episode, you know, 37 hours down to three- Yeah

James Kelly: about a cash forecasting process, again, without NDAs and things like that, but can you talk through how, just summary level- What actually happened in that process and how you changed it? So on- Yeah … cash forecasting, effectively we move from a decentralized model where the business partnering teams prepare the forecasts and they send them in.

James Kelly: Yeah. And effectively they pull data from the ERPs, they talk to their, their commercial teams, and then they come up with a view, submit an answer, and off we go. A- and that can be very labor-intensive. If you’ve got 20 people all spending two hours, you’re at 40 hours a week. We’ve got plenty of examples of that.

James Kelly: Yeah. And then you’ve got a central team who probably have to spend about 10 hours chasing, reviewing, amending, updating, consolidating, reporting. So it’s really easy to, you know, that example, you’re at 50 hours a week. You know, we’ve got examples of people who are spending up to 100 hours a week. And if you reverse engineer that and say, “Well, actually, what we’re now going to do is we’re gonna pull all the data from the ERPs ourselves.

James Kelly: We’re also going to use the historic information to project forward, and we use different approaches for different types of cash flow because they all behave slightly differently.” Yeah. Yeah. But effectively, once that’s configured once, you’ve then got a model that works forward. Yeah. And so effectively, the, the forecast production exercise is effectively click a button, and that gives you the group.

James Kelly: You’re then in a situation where the key thing is that you need to understand what the assumptions that build that up are And then that gives an opportunity to go out to the operating companies and say, “Right, review this. These are the assumptions we’ve made. What have we missed?” Yeah. And typically that’s, you know, 15 minutes a, a, a business, something like that.

James Kelly: And so, and then there are, you know, if you’ve got 20 entities, often there are 12 that don’t really matter. And so it’s quite quick to get an answer back, make any changes, a- and then there you are. So you can quite quickly get to kind of three hours a week across the, the whole finance stack rather than 30, 40, 50 hours.

James Kelly: And- And the outcomes as well tend to be better as well. Sure. Because you’re able to get more deta- you can get as much detail on the least important as the most important. Yeah. Whereas typically what happens is the treasury teams will slightly discount the less important ones because there isn’t enough time and resource, and that’s part of how you end up with, like, cash buffers building up and that kind of thing.

Mike: Yeah. Because there’s just, you know, there’s a few hundred thousand here and a couple of million there, and before you know it, there’s a, there’s 100 million around. Spirals. Yeah. And what happened to the work, if you like, the, the… You said you’ve reduced the level of work. You know, some people were worried if you like, that that’s immediately gonna be job losses and things.

James Kelly: But as I’ve, you know, been talking to people, they’ve, we’ve said it’s freed people up to do more interesting and more value add. Yeah. Has that been what you’ve seen and things? Yeah. There are very few, uh, particularly where you’ve got that decentralized model on cash forecasting. Yeah. Generally speaking, the, the people who work in the businesses don’t treat it as a core part of their role.

James Kelly: It’s something that they have to do Yeah But it’s n- it’s not really something that they particularly want to be doing. And then if you look at the central team effectively having to coordinate that chasing exercise, that’s just… It’s not value added. It’s not something that you need, you know. Yeah. I mean, I’ve got 10 years of, of training between accountancy qualifications, my degree, and, and my treasury qualification.

Mike: The idea that you need that level of, of training in order to chase people down for a submission, you know- It’s not valuable … it, it’s not a good use of people’s time. Someone listening today- Mm-hmm … and they’re thinking, “Well, we’ve got a process.” Yeah. How does someone identify whether a process needs improvement and investment to get that better, or where does it need a reset, rethinking right from the beginning?

James Kelly: How do you characterize those, if you like? So I, I think it’s one of those things. It’s thinking about what the problem is- Yeah … and then thinking about the available tools and how close can you get with your current process. And- Of course … and cash forecasting is, is an area where it’s probably the most stark, which I think is part of the reason why people talk about it so much.

James Kelly: Because if you, if you’re asking people to fill in a 13-week forecast, so effectively you’re looking at, you know, 65 data points, and then you’re looking at multiply by, let’s say, 10 categories. You’re then asking people to fill in 650 data points a week But is 10 data points actually enough, or 10 categories actually enough to get the granular information that you want?

James Kelly: Maybe you want information on the top 10 customers. But if you, if you ask that, you’re now asking them to fill out 1,300 data points a, a week, and the chances of the, the level of quality remaining the same is kind of relatively low. Yeah. If you then want, you know, information on major supplier payments as well, well, that’s another, let’s say another three or four categories.

James Kelly: You know, quite quickly you get to a point where you’re kind of going, “Well, actually, we could comfortably be asking people for over 2,000 data points A- and there’s just no way- That’s not happening … and so it’s one of those things that realistically you have to go back to basics and go, “Well, is there- how much of this information can we get from other sources so that we minimize the amount of information that we’re asking from the, the teams themselves- Yeah

James Kelly: and we get the, the really valuable stuff?” And it’s the same when it comes to FX, for example. People are reasonably good at picking out balance sheet exposures, but then there are all sorts of hidden exposures. Often when you talk to teams, there are things that they’re not quite comfortable about with intercompany, for example, or there are issues relating to procurement contracts that maybe have a fixed exchange rate but that can break if the exchange rate moves too much and- Yeah

James Kelly: and, and actually they’re not on top of like what all of those contracts are and that kind of thing. So there’s kind of, there’s lots of little bits where actually you can use technology to kind of go, “Okay, well, let’s review all these contracts. Let’s pass them through AI, have a database,” and then a series of kind of little warning signs that say, “Oh, by the way, you know, the exchange rate is now within 2% of the trigger threshold on this particular contract.

Mike: You might wanna start having conversations now before it actually triggers and, uh- Yeah … and you start having to pay money.” Yeah. And you and I have been, we were just talking before at various conferences, and I know that you’re on stage and giving examples where you can about something where a treasury team versus embraced AI.

James Kelly: Yeah. You know, are there anywhere you’ve seen where, where there was an existing problem, they tried something, and what was the result sort of thing? Are there any, any good ones you can give to the audience? So one of my favorites is actually one of our Irish clients does pensions reporting as part- Right

James Kelly: of their kind of year-end process. It doesn’t really sit anywhere, if that makes sense. Yeah. So it’s something that the treasury team have to do. They have to disclose the, all the assumptions and that kind of thing, and there’s a big report comes through from Wi- Willis Towers Watson, and that process was typically taking a couple of days, but at a year-end when timelines- It gets super tight

James Kelly: are super tight. A- and the accounting team wanted to be able to post everything and all of that kind of stuff. And so we worked with them, and we created a co-pilot agent process that effectively picked up all of that data from Towers Watson- As well as the existing template from the previous year. So it would roll the, roll the previous year forward and then populate based on all the information.

James Kelly: And we created what’s called semantic mapping. So effectively we told the tool where to find all of the different- Yeah … data points, and that meant that the actual time to, to prepare the note condensed to about half an hour. Wow. So- From what? What was it before? It, it was over two days. So it’s one of those things that it meant that year-end process effectively became much more pain-free.

Mike: Yeah. And I really like th- that kind of year-end piece in particular because I know from experience that, you know, whenever a number changes, it’s a real pain because it flows all the way through the balance sheet. Yeah, ’cause there, it starts there, so yeah. Yeah. All, all the notes and- Yeah … disclosures all over the place.

James Kelly: And so having a process that, like, picks up all this stuff, again, it- it’s, that’s what I call, it’s like surge work. It’s not steady. It’s extra work that comes at times where you’re busy. Yeah. And if nothing else, reducing the amount of work at times where you’re busy significantly reduces risk. Yeah. I know that I’ve always felt that there’ve been times where everyone’s flat out on things,

James Kelly: and you kind of feel I’m really hoping that nothing major happens in the world Yeah, ’cause otherwise Because, because we’re not really able to monitor to the level that we normally would do. Yeah. I think it allows you to kind of cope with that kind of situation better. And, you know, the world is littered with plenty of examples of, you know, half-finished M&A integrations and things, n- not because of any major problem other than the amount of work required in order to, to get over the line, and the amount of coordination is just too tricky.

Mike: And so these tools become really helpful. KYC and account opening, for example, become much easier using AI to fill out forms, et cetera. And again, I know, know we’re pushed for time, but wanna make sure. So you told me that, and we just discussed it, that your training, if you like, starts by letting people play with the tools you give them- Yeah

James Kelly: and then look around treasury tasks. Why in particular do you start there? Or also for anyone, again, listening, if they’re wanting to encourage their teams, they can see the benefit, how would you encourage that? See, we typically start outside treasury. Right. We typically just get them to, to play and explore and we’ll give them a, a, a case study to work on, but that isn’t a kind of treasury process.

James Kelly: A- and the reason to do that is to make sure that people have a common understanding of the art of the possible, and they’re not too much in their heads around solutioning how things would work in their particular environment, if that makes sense. By taking them into a different project- They’re a little bit freer to actually experience the tool and what it can do.

James Kelly: And then we start to bring that into a treasury context and say, “Well, we’ll get people to explore how far they can push the tool, how much it knows,” so that they can think about, like, how much could they share with the AI tool and get it to do for them. We then work through, like, the detail of prompting, and we have a, a framework called the, the CASH framework.

James Kelly: It’s context, anchor, specifics, and how. This effectively runs through, like, what are the key pieces of information that you need to be able to share with a, an AI tool in order for it to really know what it needs to do? Yeah. It doesn’t know where in its training data to look and where to pitch things. So the more background you can give around what is it you’re trying to achieve, how do you need to achieve it, what specific items do you need to include in a response, who’s it for, all of those kind of things.

James Kelly: And give examples is a great way of getting better out- outputs. That then means you’re in a situation where you’re able to move forward. And often we find that we work with some companies direct on transformation projects without doing any initial training, and sometimes we do initial training and then that leads on to other projects And if you were hiring a treasury analyst or manager per se today- Yeah

Mike: what would you be wanting to hear from them again about technology and how they’ve learned and things like that? What are you gonna be looking out for, or what, what should our listeners look out for? Enthusiasm and carefulness in equal measure. The reality is that we’re still in a stage where mistakes are made by AI.

James Kelly: Again, it depends how you set things up. Yeah. We’re quite careful about what tools we use for what, and I think that’s one of the things that people have to be quite careful of in the market because

James Kelly: if you use the wrong tool for the, for a task, you’re end- you’re likely to end up with much more variable and uncertain answers than if you, if you pick the right tool. So we would tend to say that for calculation, it’s better to use a calculation tool- Yeah … what’s called a deterministic tool. That’s either a, a- an agent with creativity turned down to zero, or it’s, it’s a Python script or something like that.

James Kelly: Whereas i- if you need something that’s gonna help, like re-engineer things, then actually you want that creativity. So it’s picking out the right tool for the right job. Yeah. But good communication, good written skills, enthusiasm for trying new things, but also carefulness and, and- Caution. Yeah … and caution because I think, you know, I had a, a quote from a workshop that we did a couple of weeks ago.

Mike: Someone said, “Well, this looks great. I mean, it’s done loads of work. I don’t know if it’s right or not, I haven’t had a chance to check it yet.” That is the danger because everything looks very convincing and it’s highly polished, but actually when you, when you scratch the surface it’s making sure that the, the actual underlying answers are right.

James Kelly: And you made this move, you know, at the time it was a brave move and step into the unknown and things like that, and you’ve encouraged people to try things and, you know, when they’ve stopped developing you’ve gotta make that move. Do you still give that advice in the same way now you’ve made that big move yourself if you like?

James Kelly: Yeah. Yeah, absolutely. And I think it’s one of these things that we see that with treasurers who are moving to new, new businesses, and often that’s a… When, when people are coming into new businesses, that’s often when they need a, need a bit of extra help. Yeah. So we work with a lot of people who’ve just made the move I’m yet to find anyone who’s disappointed by that Right.

Mike: That makes sense. You know, it’s one of those things that I think generally people have got quite a good handle on when they need a new stretch Shift. Yeah. Okay. And as we wrap up today’s show, you’ve given, a couple times we usually do the LinkedIn close. I’m not gonna do the LinkedIn close today ’cause I think it’s different.

James Kelly: We’ve already asked you two or three times, and you’ve given us some great advice, and people can see that, they can see it in the notes. But I was gonna ask you simply, what are you hoping to build with your treasury over, in your company over the next year or so? Where, where do you see it going next?

James Kelly: What’s your crystal ball for the, you, yourself and the company? I think we’re getting some really good opportunities, and I think there’s kind of, there’s two pieces that we’re doing. We’re, we’re working a lot with corporates, and we’re also working a lot with TMS vendors. Yeah. And so I think a continuation of that piece I think is really interesting, because the work with individual corporates allows us to get close to teams to, to really help at a personal level.

James Kelly: But by actually helping the TMS vendors build their products as well, we’re able to have a, a broader reach. And I think that that mix, you know, again, the work that we’ve done with the AFP allows us to be a voice in the industry, and I think that’s sort of where I would want to continue to go. Yeah. Uh, able to, to help at a, a detailed level, but also having a, a broader influence on the industry.

Mike: Amazing. James, thank you very much for catching up once again. It’s been an amazing 15 months, and, well, we’ll do it again in another year and a half. Thank you, sir. Thank you very much. Thank you.