In this episode of What’s New at CFI on FinPod, we discuss the power of AI for finance, focusing on AI-powered scenario analysis and its impact on financial modeling.

We explain how AI is not just enhancing traditional scenario analysis but also making it more efficient, detailed, and accessible for finance professionals at all levels. The conversation breaks down key components of scenario analysis, the differences between traditional and generative AI, and why these advancements are crucial for those in FP&A and corporate finance roles.

This episode offers a dive into the evolution of finance, providing practical knowledge that can help you leverage AI to improve your financial models and decision-making processes. If you’re looking to stay ahead, you won’t want to miss this episode.

Transcript

Ryan (00:13)
Hello there, and welcome to this latest episode of What’s New at CFI. My name is Ryan Spendelow Senior Vice President of Training and Curriculum here at CFI. I’m joined by my fellow subject matter expert, Glenn Hopper, he’s also a fellow cycling enthusiast. So, good morning, Glenn. How are you this morning?

Glenn Hopper (00:34)
Ryan, I’m doing great. Just wrapped up the Tour de France. Waiting to see, I’m waiting to see if Tade is going to ride the Vuelta next. We’re getting way too deep for our typical fans now, aren’t we?

Ryan (00:41)
Yeah!

I know. I hear he’s not riding the Olympics, which as we record, this has just started today. So it’s going to be interesting to see what he does. But as much as I’d love to spend the next 10 minutes talking to you about professional cycling, we’re actually here to talk about the latest course that you’ve created for CFI, and it’s called AI-powered scenario analysis. Do you want to just give us a little bit of a background about the course?

Glenn Hopper (01:08)
Sure. So first off, think scenario analysis.

Real finance geeks like me and many of our listeners, we love building models. So any excuse to build multiple models is that much more exciting. in the course of scenario analysis, I talk about things that we’ve all used throughout our FP&A careers, but with new technology right now, I hope to show through this course how much quicker and how much more efficient and how much more detailed you can get using generative

AI to do the same stuff we’ve been doing and to be to add new levels of depth and to be able to not just create a scenario analysis with multiple scenarios, but to be able to interact with each scenario. So it’s really, it’s a really exciting course for me because it’s really, it shows hands on how something that we’ve done for years is really being transformed by the power of AI.

Ryan (02:06)
Brilliant. Now, I’ll share a little secret with you, Glenn. I actually took the course this morning and completed it, and you’ll be glad to know that I got 100% in the assessment. In doing it, I see that it’s probably suitable for not just FP&A analysts, but also probably financial analysts in general. But can you just tell us exactly what scenario analysis is and which key components of scenario analysis do you consider to be the most important?

Glenn Hopper (02:33)
Yeah, so as far as

what scenario analysis is. I think it’s summed up perfectly by a quote by a famous statistician named George Box. And I guess I’m showing my level of geekiness here that in my world, there are famous statisticians. But George Box said, all models are wrong, but some are useful. So to me, what scenario analysis is, we don’t have a crystal ball. We don’t know what the future holds. So we can take our historical data

Ryan (02:47)
Hh-mm.

Glenn Hopper (03:04)
and model that out and we can take what we expect to happen and model that out, but we don’t know what the future holds. So with scenario analysis, it’s really asking what if, and it’s, you can think of things like what if our main raw goods supplier goes out of business? What if there’s an economic crisis or some other global event? What if our sales drop 30% ? Well, if you’ve just modeled out, you know, status quo, nothing.

Nothing changes, well then you’re not prepared at the full level of what you do if something happens. And if you’re going to use data and if you’re going to use your models to make decisions, you can’t

think you have the perfect roadmap for where it’s gonna go. So scenario analysis allows you to evaluate the potential impact of different future events for, know, it could be a business, it could be an investment, or it could be a whole financial portfolio. You’re looking at, you’re kind of stress testing. Okay, if everything goes normal, it’s great, I get my 12 % return, whatever, but if something happens, you know, how much of an impact is it gonna have on what I’m planning? So I think what people take out of scenario analysis in general, and one thing going back to the course that I think

we show is how much easier this is. what people need to think when you think of a scenario analysis, the first thing you do is build out your base model. Then, you look at all the potential scenarios that could happen. And you’re not going to think of these black, all the black swan events, but what are the most highly probable events that are going to happen or at least somewhat likely? And then you model out each one of those. And then you kind of stress test the original model based on these different scenarios,

complete a scenario analysis, gives you a more rounded picture of whatever the various potential futures are. And then, instead of just going forward on hope and faith, you’ve got this basis to inform your decisions. How risk-tolerant are you? Do we need to hold back a cash reserve because we think it’s likely that maybe our raw goods supplier will go out of business or there will be some kind of economic crisis. know, so it’s, it’s really just painting that full landscape of potential futures through scenario analysis and,

and what I love is we’re getting asked as finance professionals to do more and more and to tweak these things through the year. And with AI, we’re seeing how much quicker we can do it and how much more informative it can be.

Ryan (05:28)
That is a fantastic explanation of it. I guess my next question is, how has AI actually revolutionized the field of scenario analysis? Because I guess it’s had a really big impact.

Glenn Hopper (05:44)
Yeah, and you, you know.

You can talk about AI in two ways. So what everybody’s talking about for the past, what, year and a half, two years now is generative AI. All the ChatGPT and Claude and Llama and all these large language models that are out there. But really the impact of what I guess I’ll now call traditional AI, which is the machine learning algorithms that we’ve been using for over a decade now, probably closer to 15 years. But with machine learning, we’ve been able to, you know, if you have these tools at your disposal,

you’re able to do a faster, more powerful data processing. You’re able to, you know, in the past, even if you could have gotten access to Federal Reserve data and macroeconomic items and weather, whatever, you know, whatever these huge data sets are and all the different variables in them, you know, if you’re trying to jam them into Excel, it’s really hard to do, but with AI and with, with data science, it’s easier to build these models. So we’ve been doing that for years. You know, another thing that machine learning gives you is…

computers are very good at pattern recognition. So it may find some correlation, some nuanced relationship between variables that helps inform your models, you know, more real-time analysis. If you had to update all this data manually before being in an AI world, it would take forever and it just wouldn’t be practical. And then also because we’re able to put more data, more features, and parameters into our models, we’re able to get improved prediction accuracy. There’s automation in what

do natural language processing. Now you can with, with AI, you can bring in unstructured data. could be earnings calls. could be, you know, social media sentiment. There’s all this unstructured data. could actually incorporate into your models, which imagine, you know, again, trying to do that in Excel without machine learning, you can’t do this kind of stuff. And you can, so just a super complex models. but the problem was if you’re not a data scientist, if you don’t have access to a team of developers who are building these

models for you. You know, if you’re if you don’t have that, you’re kind of sitting on the sidelines and thinking, wow, that sure would be cool to have all that. I don’t have access. Well, that’s what’s exciting about generative AI. And we’re still in the early stages. And, we’ll see even in the courses, I show you some of the limitations of what you can do with generative AI today. But what the promise of generative AI is and what we’re getting to more quickly than even I might have guessed. And I’m a techno-optimist, you know, but is you’re you’re going to knock down those walls and

is going to have the ability to run these complex models that we didn’t before. So all the stuff we’ve been doing for years has been great, but now we’re just democratizing that and reaching out to a much broader group of people who will be able to do it.

Ryan (08:31)
It’s just incredible, isn’t it? So, Glenn, can you explain why this course is particularly beneficial for financial analysts and corporate finance professionals?

Glenn Hopper (08:42)
Yeah, you know, it kind of goes back to what I was just saying. 

If you didn’t have access to write code, didn’t, you weren’t, you didn’t used to be able to do this. And now, you know, there are tools out there and not to show for any particular software or anything, but there’s tools out there off-the-shelf software tools that have built-in scenario analysis features where you can pull little levers and tweak things. But the truth is, and that’s great. It’s more than what we used to have, but every business, every industry, every geography, you know, every there, there’s just too many unique situations.

So, an off-the-shelf, out-of-the-box tool isn’t going be able to handle for all these. But if you can, on your own, interact with generative AI that contextually understands what you’re doing, then in that case you can.

Kind of you’re building custom models, can interact with the data. So what I, what I hope to show in this course is, you know, you’re not, this is very different than traditional software. You are working with the AI as if it is a coworker, someone that’s helping you work and you working with you on this. And what I think people will get from the course is I don’t have to rely on some third party software. have this, you know, robot teammate that I’m working with that’s helping me do this. And I think through the course, you see, you don’t treat this like

software, you’re not just clicking buttons and pulling levers, you get a real teammate. And I think that by the end of the course, you see how successful it can be. even doing things like running Monte Carlo simulations on our data that are running 10,000 cycles of different scenarios and helping you build models. It’s just super cool technology that’s out there. And hopefully, this course helps make it accessible to anyone, even if they haven’t used generative AI before.

Ryan (10:26)
Glenn, in the course you’ve done a fantastic job at showing just how accessible this is to people. And one of the things, one of the many things that I found really interesting and really useful is the way that you interact with the generative AI and the prompting that you use. And so what really resonates with me just there is that how you explain it as your partner, as your robo partner. And that comes through in the way that you interact with it, the way that you prompt it, the way that you communicate with it.

And it’s a real eye-opener to see someone like yourself leveraging the best practice almost approach to making sure that you’re putting in the right prompts to get the right information, the right results.

Glenn Hopper (11:11)
Yeah, and I guess what I would say on that is I’m on record as I hate the the term prompt engineering, because I think in a couple of years, it’s going to be saying I’m good at prompt engineering. It’s going to be like saying I’m good at Googling. It’s not going to matter what and what I hope comes through. Yeah.

Ryan (11:27)
Yeah. I still say that. I’m out of the Googling, so I’ve got a long way to go.

Glenn Hopper (11:36)
Yeah. So what I think that really I hope comes through in the course and that people need to think about it’s I’m not going to hand you a sheet that is here’s your prompt library. Just copy and paste this into the software because it’s not like traditional software. If that were the case, they would just have buttons you push. So what is really cool with interacting with this software is it’s not a specific…

this is how you, this is the prompt you use. It’s more of understanding what the system does and what its strengths and limitations are and knowing how to ask it questions. So, you know, every question could be different and every response could be slightly different. The numeric responses shouldn’t be different, but the way it handles them will be.

Ryan (12:19)
Hey Glenn, that is fantastic. I’ve really enjoyed taking this course. As I said, I took it this morning. I think our CFI learners are really going to find it beneficial. And I really appreciate you taking the time out and doing such a great job at explaining what the course is all about and why it’s so beneficial for our learners. So I know that you’ve got a couple of other ideas and a couple of other AI courses in flight. So hopefully…

we can have you back on another What’s New at CFI podcast sometime in the near future to talk about our upcoming AI courses. Would that be okay?

Glenn Hopper (12:51)
Anytime, Ryan, always a pleasure.

Ryan (12:54)
Brilliant, all right then. All right, thanks very much for joining us on this latest episode of What’s New at CFI. My name’s Ryan Spendelow and I hope to see you on a future What’s New sometime in the very near future. Take care, bye!