What’s New at CFI | AI Prompting for Financial Analysis

Meeyeon (00:00)
Hi everyone and welcome back to another episode of What is New at CFI. Today, we are hitting on a topic that is so topical. Everyone wants to talk about it. We’re going to be talking about one of our latest AI courses, AI-related courses. And today I have here with me Ryan Spendelow, VP of Content and Curriculum here at CFI that you have seen and heard many times. Welcome Ryan.

Ryan (00:24)
Hey, Meeyeon. Lovely to be here, as always.

Meeyeon (00:29)
And today, we are talking about our latest course, AI prompting for financial analysis. So that right there is probably going to draw in a lot of you that are listening, because that is what everybody wants to do in finance today, is learn how to prompt AI to make us more productive and more efficient

at a 10,000-foot view, what is this new practice lab on AI prompting for financial analysis all about?

Ryan (00:56)
Yeah, good question to start. And just to clarify for our learners, so this is a practice lab. It’s a short course, for lack of a better word, that’s aimed to develop a specific skill in a certain area. So this should take learners no more than about an hour to start and complete. So it’s a very, very focused bit of content. And…

the reason why we developed it is that AI, as everybody knows, is starting to show up everywhere in finance workflows. So whether that’s in research, analysis, or even preparing PowerPoint presentations.

But the real issue isn’t the AI tool itself; it’s how analysts use it. So AI is very, very confident. It’ll give you an answer even if your prompts are vague or if the assumptions behind your analysis is weak. And the danger is that, in finance, weak assumptions can lead to flawed analysis, bad decisions, and, really…

core financial consequences. So, the purpose of the practice lab is to really help analysts avoid that threat. And instead of treating AI like a shortcut to answers, this lab is really focused on helping CFI learners to use…

AI as a thinking partner and we use that term a lot throughout the lab. The lab will learn us through a really structured process where they first have to identify some assumptions or the assumptions, but…

behind a set of financial numbers. And then they use AI to stress test those assumptions, but finally, they need to apply their own professional judgment. The goal isn’t really about just teaching prompting, it’s to help analysts become more disciplined and more thoughtful in how they use AI in their financial analysis.

Meeyeon (03:04)
And the lab introduces a key framework called CAP-AJ. Could you share with us what it is? First, I guess what it stands for and why it’s useful.

Ryan (03:16)
Yeah. So CAP-AJ is actually a framework that we developed here at CFI to help specifically financial analysts use AI in a really structured and responsible way. so CAP-AJ stands for context assumption prompt,

which is the CAP part, and then assess and judge. And those first three steps happen before you even interact with AI. So the first is context. You have to understand what you’re analyzing and what information is available or missing. The second is AN. AN stands for Assumption. Clearly stating what must be true for the numbers that you’re reviewing to be reliable. And the third is prompt.

This is when you actually design a focus question for the AI tool. Now in the lab, we use ChatGPT, but this is really applicable for whatever conversational AI tool that you’re using.

And so once AI generates a response, the financial analyst then moves to the final two steps, the AJ part. The A, or the second A, A is assess, which means assessing, reviewing the AI output, and extracting the key themes that really matter.

And this is the most critical step, the J part, and J is judge. And this is where the analyst applies their professional judgment. And this means deciding from that AI output what to accept, what to question, and what might require further investigation.

And I guess the reason why this framework is so useful is that it keeps the analyst in control of the analysis. AI becomes a tool that supports your thinking because the dangers that we rely so much on AI that the AI actually ends up replacing your analysis, which is a place that we don’t want to be in as a finance professional.

Meeyeon (05:20)
Yeah, it’s so important to make sure that AI, at least I think the purpose of it is to support you in your thinking, but it shouldn’t guide you in your thinking. The guiding and steering should be done by you, the human, and the AI kind of just supports you along the way. But it should never at one point be the person that is guiding the analysis, driving the conversation, and you’re just following along and trying to keep.

Ryan (05:44)
Yeah, absolutely. You, as the final analyst, are responsible for the analysis, and so you need to retain that professional judgment. AI is a productivity enhancer, but it doesn’t replace your financial analysis skills.

Meeyeon (06:00)
And maybe we can give our listeners an example of what kind of finance scenarios that learners are going to work through in their practice labs without giving away the whole thing, maybe just like a related type of example.

Ryan (06:15)
Yeah, sure. So, in the practice lab, we use accounting assumptions as the context to teach the CAP-AJ Framework. So we have some numbers from an income statement.

And we start by thinking what assumptions must be true for those numbers to be reliable in our analysis. But we really just use accounting as the context becauseit’s an easy thing to grasp this idea of what accounting assumptions must be true behind numbers on an income statement. But as we allude to in the… Did you hear that computer?

So, I’ll just keep going.

But as we allude to in the practice lab, the CAP-AJ framework is actually really quite applicable in many areas of finance. And so we give examples of how it can be used in FP&A, how it can be used in investment banking and capital markets, and so on and so forth. And we also support the learners by providing a couple of resources that they can download and take away. And one is a CAP-AJ cheat sheet really. So it gives examples of a poor prompt

from a financial analysis perspective. Then, we show an example of an effective prompt using the CAP-AJ framework for financial analysis. And then also, we demonstrate how the framework can be used in other scenarios as well across different types of financial analysis. So, it’s not just accounting that this works for; it works in many, many financial scenarios.

Meeyeon (07:59)
And as learners progress through this course, and I think in their day-to-day life, so as long as you are not about to retire tomorrow, you whether you started your analyst role yesterday or you’re midway throughout your career, I think we’ve all been exposed to AI for a fairly limited time. And we are still understanding how we need to shift our mindset when using it. What is the biggest mindset shift that you think analyst? And when I say analyst, I just mean,

everyone that’s working in an analytical role needs when using AI.

Ryan (08:34)
Yeah, and I think this is a really, really important question, Meeyeon. I think the biggest shift is realizing that AI doesn’t replace analytical thinking. It actually requires, in some ways, stronger analytical thinking. The best financial analysts aren’t the ones who just rush to answers. They’re the ones that ask the right questions and understand the assumptions behind the numbers.

When people first start using AI, I think there’s a temptation to jump straight to the prompt and ask the tool to analyze something. We’ve all done it before. I know that I have. But if you haven’t clarified the assumptions yourself first, you’re essentially letting AI define the problem for you. And that’s really risky in finance, because AI…

doesn’t really know the specific context of the company, the accounting judgments, all the business environment that you’re analyzing.

So, the mindset shift we emphasize in this lab is that analysts should do the thinking first and use AI to challenge or strengthen that thinking. I think when used properly, AI can surface the risks and highlight areas of uncertainty that help analysts…

explore different angles, but the final judgment still belongs to the analyst. And that’s really the core message throughout the lab. AI can amplify financial analysis, but it cannot replace financial expertise.

Meeyeon (10:01)
And last but not least, as someone that has experimented with AI in the past, in the recent couple of years, what surprised you most when you started to experiment with AI tools specifically for finance workflows?

Ryan (10:17)
I think the thing that surprised me the most is how compelling the responses from AI can appear at first glance. But when you dig into it a little bit deeper, how easy it is not to spot the mistakes that it’s making. And that really reminds me that

for the foreseeable future, AI is changing so rapidly, you’d be foolish to make any kind of predictions. But I think this is why core fundamental finance skills are still gonna be critical for any financial analyst.

If you’re moving into an investment banking role, for example, having financial modeling skills, valuation skills, accounting knowledge, Excel skills, they’re still going to be table stakes. You’re still going to need those because you need those to be able to audit the output that any large language model generates. Or whether you’re moving into capital markets, still understanding your asset classes and derivatives and risk.

If you’re moving into FP&A, you’re still understanding budgeting and variance analysis and scenario analysis and what have you. So those skills I think are still critical for any financial analyst. But yeah, what surprised me most was how compelling AI can sound and appear, but still be wrong.

Meeyeon (11:52)
So there you have it, folks. I think that this is going to be an incredibly important practice lab for anyone that wants to elevate their productivity. That might be you who’s listening right now. But remember, this is a concentrated course. So if you have an hour of time where you want to experiment with some new AI tools that can help your workflows and finance, this is it for you. It’s a condensed.

It’s a condensed course, so to speak, but it’s very practical, very hands-on. And we think that it’ll be a great benefit to your workflow. So we hope to see you in that course, or at least you’ll see Ryan in that course. So until next time, everyone, thank you for listening, and we’ll see you next time.

Ryan (12:37)
Thanks, Meeyeon.