Live from the Adam Smith Awards, Treasury Today talks with Ravi Patel, Senior Director Financial Risk Management at Bristol Myers Squibb and Marianna Polykrati, Group Treasurer at AVRAMAR about treasury's top priorities: automation, cash flow forecasting and where AI is really being used. They also cover real-time payments, ISO 20022 progress, building the next generation of treasury talent, and hedging against geopolitical and supply chain risk.

Hello, everyone, and welcome to the Treasury Today podcast. I'm Liz Lumley of Treasury Today.

So we are coming to you live-ish on the hottest day of the year here in the UK, but it is the hottest ticket in corporate treasury, the Adam Smith Awards. And I am joined by Ravi Patel of Bristol Myers Squibb and Marianna Polykrati of AVRAMAR. Welcome, both of you.

Since this is the Adam Smith Awards, we're going to talk about top priorities for corporate treasurers right now. So that's what I'm going to start in with, that really broad question. So Ravi, I'll start with you. What are your top priorities right now?

Yeah, we're continuing on our journey of automation. So this year we're looking to implement CLS and exploring where we can find AI automation opportunities. And then lastly, we're continuing to be a partner for the enterprise through strategic partnerships.

What about you, Marianna?

Look what Ravi said, but I would say something else, something that I think it's something nobody have ever heard, cash flow forecasting. I think it's the hot topic of the last 20 years for treasurers. So we're trying with the cash flow forecasting after a period of data cleaning and trying to see what Ravi said with AI, how we can help this model better our cash flow and see how it goes. So basically, for us, cash flow forecasting.

I agree with you, but I have a feeling we're going to talk a bit more about AI during this podcast. But talking a little bit about real-term payments, and since you brought up cash flow forecasting, and removing friction from cross-border payments, how can banks kind of support the corporate partners? Marianna, do you want to start with that?

Look, whatever new innovation comes from the banks is more than welcome from the treasury community. We have, I mean, they always put us in a very nice structured way to think about things differently.

I always like what the banks are talking about technology and how they can do the cross-border and instant and everything, but usually most of the times it doesn't work exactly as I say.

So it's a bit, you try have to a bit not that big expectations, but I think we're going towards the right direction right now.

But what about what's going on with cross-border payments and ISO 2022? How's that working out?

Look, with the ISO 2022, what we're seeing is that after the VOP, the verification of payee, that they're trying to help you with the beneficiary, with the ISO 20022, you're trying to structure more your data that you have on your suppliers and all the vendors that you pay.

We have found as a company that we were missing a lot of data. So what we have done, we have focused a lot on the master data and fixing all these issues. So we will be able to implement the ISO 20022. Now, if you're asking me if it's correct, I believe it's towards the right direction.

Maybe not as it is, but we... We will have to see and discuss and negotiate with the banks. In Greece, the banks are still not ready, though it's November 2026. We're having the first pilots from the Greek banks in mid-July and session starting in September.

So as I say, maybe they're hoping that it's going to be a bit pushed back as they were hoping with the VOP, but I don't think it's not going to happen.

Interesting seeing that insight. I mean, what do you think about sort of bank supporting their corporate partners.

Just to add, the real-time payment innovation is really moving quickly. I've seen from our banking partners integration between blockchain and traditional finance in a way that we haven't seen before, enabling real-time payments, cross-border between entities, all these different types of technologies.

The challenge though, just to echo, it's not necessarily adding that much value to us above what we're already doing today.

The traditional lanes or rails are working pretty well for us. We don't have necessarily a pain point that gets solved by the innovation that's coming out. So I don't think we'll be a leader in this adoption, but I think if it becomes regulatory mandated or it moves to be more mainstream, I think we would move into it.

So I've been to a number of events throughout the year and of course everyone's talking about AI. And someone texted me about an hour ago and said, everyone's talking about agentic treasury. So we'll get into bigger with that.

I mean, in terms of like artificial intelligence adoption and the different tools and different strategies you can use it for, like where are you seeing the most use?

Yeah, BMS is at the forefront of AI adoption. We have been for years, just given the science aspect of it. We're using it to cure cancer and to find medications that'll be more likely to succeed.

And because of that, on the finance side, we benefit from some of the tools that they're able to adopt and we can leverage some of those. So we've been trying to experiment with as much as we can.

You've got the gold standard AI tools, the ones that are curing cancer. Can we use it for some cash forecasting?

We're in the exploratory phase. Individuals across treasury and finance are trying to find the best ways to leverage AI.

You know, we found some creative uses across different teams. So for example, one creative use we came up with was putting our work instructions on a SharePoint and then having an AI agent access those instructions. And then for new employees, they can just ask a question to the agent, how do I place a trade? And then it'll give the response in a way that can be easier to find, easier to access, and hopefully more helpful.

And then through that, what we're trying to explore is can we, if there's gaps in there, maybe we can use that chatbot to update the instructions. That would be the next level.

But we're just starting to toy around with these, see what works, and then implement those more broadly as we find more success cases.

Is it more of an augmented agent or are you moving to autonomous?

For now, it's augmented. I don't think we're at the point where we're going to have agents sending wires.

But you have to be very, very careful about this, right?

I think what may be one case that I heard, which could come in a few years, I don't think we're there yet, but maybe you could have a KYC bot at BMS that talks to a KYC bot at a bank. And maybe they have a predefined set of documents that they can share with that bank and they can talk and get 90% of the way there.

I think maybe we're a few years away from that, but that is a more, I think, treasury friendly autonomous use case.

What are you hearing, Marianna?

Look, treasurers, I think we're quite risk averse. So I think while we are the ones that like to innovate and do a lot of things first, I think we're quite sensitive in the risk and compliance that AI could bring into our companies.

For us as a company, we started, not playing, I won't say the word playing, but I would say testing a lot with AI over the last one year and training the team, simple uses, co-pilot, how we can improve things, starting from mails.

I have seen from other corporates, not my corporate, that they have very good chat boxes and they have the policies and procedures, which is very interesting because you can get rid of a lot of the noise and the phone calls that happen in the organisation, so it helps you more structure. So you say, what is the proper way to do an expense report or an expense? How do I do? It shows you the whole policy. So this is one of the things that we have seen.

What we are currently looking at, we are trying to build cash flow forecasting with two very nice advisors and do our own cash forecasting using an AI model, using our historical data.

So we just kicked it off a week ago and hopefully by the end of October, maybe we'll have something much more interesting to discuss and see how it works and how it reads the data. The problem is that usually you don't have a very straight cash flow or normal cash flow.

They have a lot of things that are happening, a lot of irregularities. So maybe the historical trends that they are leading are not the correct ones, right? So you have to be in and start fixing a lot the model and playing around with it. So we will be testing it. But we're not negative. We're quite positive on doing it.

And by the way, since you said making payments, I was at an IMD session last year with a great professor. And he said he was playing with agents like one, one and a half years ago.

And he had the agent looking for books. And what happened at the end is don't ever have your credit card inside. It ordered the book. The agent ordered, he was trying to find the best book and the agent ordered the book. So the book came to his door two days later.

And he said, so be very careful with the agent AI as he was playing one and a half years ago. So yeah, we have to be quite careful.

Yeah, I was speaking to a corporate treasurer in Germany last year, and he said he built his own agent, but he just doesn't he doesn't trust it yet. It's all Frankenstein creation where we're getting to.

I have people that don't trust the TMS, and they go and they do it on Excel files as well. So it's not something, I mean, they don't trust the results from the TMS, and they go and they test it in Excel files. So this is a matter, I think, of culture and what you think about it, right?

It's what people feel confident with and safe with. Yeah, it's interesting. I'm going to save the long-term forecast for the end, but I want to talk a little bit about talent. I mean, since we just talked about AI, sometimes talent comes up again.

I mean, who are the next generation of corporate treasurers, Ravi? What do you see?

Yeah, and this is very timely. We're interviewing right now for my team.

I think given AI has changed a bit of the requirements that we would look for in the ideal candidate. Obviously, you want the finance background and skill set, but having a little bit of exposure to AI, but also to Python or some kind of computer script is very helpful.

I'll give an example. So if you ingest all the data into AI, you can create a dashboard. And that's great. And you can ask AI to update that dashboard. But AI is a probabilistic model. It's giving you the highest likelihood of what the right answer could be. And it could change that answer in the future. You want X plus Y. You don't want it stretching X plus Y plus Z.

So what you want to do is to take that and to put it in, have AI create the Python script that can then run that and update that and refresh that more frequently. But to get the right people who can take you from point A to point B is going to be the challenge I think we face because that's not intuitive, based on the way we've been doing things, it's going to be an adaption we have to make, something that we could have only done with technology resources before.

Now I think the hurdle to implement some of those things is a lot lower. And so we can do potentially much greater things if we have the right people.

I see. I mean, what are you doing for the next generation? I think corporate treasurers have lost the Halloween costume war, you know, the firemen and cowboys, like little children dreaming of what they're going to be.

It's very interesting because we have seen in Greece that there are a lot of the people and the students are in the university, they don't know what treasury is. So personally, myself and a lot of other good corporates, we're going to the universities and we're showing simple things, what treasury is and what it's not.

We have the technological part that it's quite, as I say, it's a bait, because the new generation they like more tech-shavvy, they don't like doing manual things, they don't like doing repetitive things, they do not understand why they should do it. They are more tech-shavvy, but they're also very dependent upon the technology, so sometimes when you take it, they look at you like they're lost, right?

I have a great team that is a combination of young and old. So I have all generations. I have 50, 40, 30, 20. So what I try to do a lot is we try to mix these people and bring experience along with the new technology so they share.

So you have to build a culture that also engages the people that are of an older age, right, in order to see the new technologies. And they feel very great with the new technologies.

Why? Because they make their lives easier. And we're trying to train the young people coming in. To have more creative thinking, because this is what you can have the model, like Ravi said, it can bring you different results, right?

But at the end of the day, you have to have critical thinking, creative thinking, in order to understand what is right and what's wrong. So this is what we're trying to train to the new generation.

I hear a lot about people talking about data analysts, I don't think that we need data analysts. Okay, the data analyst will be the model at the end of the day, right? The AI.

You need people that will have critical thinking in order to interpret these results and see the trends and understand much more the business.

So I'm interested in that kind of multi-generational team that you said. I mean, does that tend to break down different biases people have? Do some of the young people see that people in their 50s maybe are tech savvy and what happens with that?

Look, I had once the... If he listens to the podcast, he's going to be like, one of the interesting things, it's not my job to do filing. I'm going to do much more interesting things. So these are for you.

You have to start somewhere.

Yeah, but you have to help somewhere in the data filing. Still we are. We're switching to the digital era and everything. But you have to, it needs a lot of things. Yeah, they think that they are different. They don't understand any things. But if you put them together, it's very interesting.

The only problem is that they don't think like their mother and kid, and sometimes we pamper the young generation because we think like he's the age of your kid, right? So you pamper them, and that I think takes them away of their target, right?

But... I did, I did, I discriminated against my own team because I had a workshop like one and a half years ago for AI, starting to talking about AI with two very good instructors. And I had myself and I had one of my young treasurers as well.

And I left the older people of my team, the 40s and the 50s outside of this group. I'm 50, right? So it's a, and I left them out. And I said, what am I doing? I'm doing something very wrong. And I brought them on. And they enjoyed this session. They felt included. So it was a win with everyone.

Interesting. So we're moving on to everyone's favourite topic, which is geopolitical events going on right at the moment. Maybe we can bring in climate change as well and have a party.

You know, how is treasury with supply chain and shadow tankers roaming around the globe. How is treasury dealing in time?

Yeah, so treasury, we can't predict nor should we expect to predict the next year of local crisis. But what we can do is to improve the forecast, mitigate some of those risks using derivative, interest rate derivatives, FX, hedging, mitigate as much as we can through that period of time.

And we're not going to be able to protect against it indefinitely, but we can at least buy some time. So to the extent if tariffs were a short-term phenomenon, we could potentially protect against some of the market impacts from those tariffs, the dollar moves for a period of time. And eventually it catches up, but we can buy the business time to adapt to any kind of new environment.

I think there needs to be some sort of heat map of all the canals around the world. You know, which one?

It's going to be only red.

I don't think you will have anywhere green in this map. That's AI will bring you a good solution. So you will be able to be modelling and put in the model all these changes. What happens if we have another blockage and oil prices go up? Let's do a scenario plus. Let's do a scenario minus.

So start making all the sensitivity analysis. What will happen if everything goes bad?

So in Greece, we had the prolonged term of a liquidity crisis. So it started in 2008. For us, it ended in 2019.

Prior to the COVID, it started ending. We had the COVID then, but we recovered after that. We have to be very creative. Supply chain. We had capital controls. We could not send money outside of Greece. And we had to buy and source all our materials.

I had a friend, a Greek friend here who visited his family with money stuffed.

Money stuffed. We had, yes, we had money. We had money everywhere. It was everywhere really, except all the banks.

So you have to start, I think, the probabilities and thinking always have in a bucket. And what is this, what is that? When I do my cash flow, I have like a small pocket, like one amount saying other.

This is the other fund. You have the other for all the extraordinary, right? You cannot have a cash, you can have a cash, you can have an overdraft. You have to have it secured in some way because things, I had a great CEO at once that he said that the only number that we're not going to bring ever is the one that's the budgeted.

So it can go either way. It can go up, it can go down, but it's not going to be the budget to figure out what you have in your business bank. I mean, it's going to be moving either up or down. So you have to be prepared for these things.

Yeah, just to echo, liquidity premium is much higher now. You need to be able to get your cash when you need it and ideally across the borders if needed necessary at a dime's notice because of the change in geopolitical factors.

Real-time payments for everyone.

So we're going to go for the long term here. We've talked about a lot of issues today in this episode. What's going to be around five, ten, 20 years? I know in AI, five years is like a century from now. But where are we going?

It's hard to imagine ten years from now. I think the next five years, the AI is enabling us to build much greater things than we would have thought possible with the resources that we had.

We always have aspirational goals and what we want to build out, cash flow forecasting. And now we can, maybe we can't quite get there, but maybe we can get much closer than we could have before.

And that'll require the right people, the right talent and time But that's where I think we're heading in the next five years.

And then after that, I think it's to be determined how that shapes and how treasury evolves.

What do you think?

For me, remember I told you in the beginning, cash flow forecasting is like the top of the last 20 years. I think it's going to be for the next 20 years because it's the operational model. The operational model, you have people, you have departments, you have certain needs, communication. Communication is one of the things at treasury, I think we master a lot. If you're great in communication, you have open communication with all the departments, then I believe that your cash forecast is usually better, right? Because you have all the unexpected things that are happening.

So I still believe even with AI, even with the best systems that you have, at the end of the day, we're still going to be trying to forecast the model, try to predict things, and still working on cash flow forecasting improved, but still working on it.

Marianna and Ravi, thank you so much. And I hope you have a very enjoyable Adam Smith Dinner Awards tonight. And you both look gorgeous. And thank you very much.

Thank you.