What kind of skills does today’s CFO need to have? 

Chen: Number one, technical skills. I’m not talking about engineering skills, I’m talking about data. Over the past 30 to 40 years, you’re seeing the slow evolution of data being a more and more important component of the finance department. Finance started as mostly accounting, and spreadsheets are what empowered the finance function to be more forward looking. With the introduction of data warehouses and ERPs around the ‘90s, that became an even more important source of data. That trend has only accelerated, particularly over the last 10-plus years or so, where companies are just becoming more data driven in general.

Increasingly, finance is not just about finance. It’s about everything else that happens in a company, from HR to sales to marketing. They have been historically the owners of financial data, but there’s a realization that all data eventually falls down to the bottom line somewhere.

The other one is storytelling. One of the things that finance people feel pretty deeply is what I call the Cassandra Effect. Cassandra is a Greek mythological figure who can foretell the future, but is cursed to have no one ever believe her. That is the experience of many finance people. They are the owners of the numbers. The numbers are accurate. They have these conversations where they make recommendations to the executive team and leadership team, and the frustration is they don’t get listened to. A lot of that is about understanding the difference between a story and something that compels emotionally, and broad numbers. Bridging that gap is one of the other opportunities I see as they own more of the data. 

How do you see AI continuing to change how the finance department works?

We’re going to climb the capability ladder. We’re going to start with the low leverage work, but increasingly it’ll be more and more human augmented or AI augmented.

There’s two ways to think about this. The first path is the agentic path, where you’re having a conversation with this creature on the other side of the screen, and this creature is doing things. That’s sort of like, ‘I have this junior employee doing things.’ As it becomes more and more capable, what’s going to increase in value in the marketplace of skills is going to be experience, intuition and judgment. The bet there is that at some point you are going to have to review or approve some course of action, some plan, some suggestion from an AI tool. The quality of what you approve really depends on your understanding of it: your experience, your intuition and judgment. How do you develop this intuition, experience and judgment? You have to still go in and deeply understand how the business works. 

These tools for thinking and understanding—in the form of spreadsheets, or whatever other tool that you choose to use—cannot really be replaced. Even in the AI driven world, the AGI [artificial general intelligence] type of world, humans are going to have to approve these decisions. 

The other branch of this is how AI can help you clarify your understanding. If you think about the evolution of information technology, what the whiteboard did is allow you to think about more complicated thoughts you can’t hold all in your head. You can write it down. And then, because it’s a whiteboard, other people can also understand the context you have in your head, and you can collaborate together. A spreadsheet is a similar thing, where very complicated calculations that you can’t do in your head very quickly, now a computer will do for you. It makes sense of this complexity, that allows you to offload complexity.

AI also is increasingly allowing very complex business changes and structures to be understandable to finance people and people outside of finance. Making sense of complexity is the other area in which software tools are going to design around. It’s not just a conversation with a chatbot or an agent. It’s more like I’m looking at a spreadsheet, my model, a number I can instantly understand, and AI will tell me what the important variations are, how this works, what this page does. That’s the thing that helps us develop intuition.

What advice would you give a CFO trying to figure out how to go forward with using AI to make an impact in their department?

I would actually say don’t believe the hype yet. Simultaneously, AI is the most under-hyped of technologies and the most over-hyped of technologies. 

Over the long run, it is the most under-hyped technology. It is the most transformative technology ever. It is the greatest thing since fire. The issue is that there is a lot of overpromising in finance today, based on what the capabilities of models are. But on the other hand, today is the worst it’ll ever be. 

One of the most important things is just understanding and being curious about what these AIs can actually do. The wonderful thing about AI is it democratized learning a year or two ago. Now, it’s increasingly democratizing building. One of our sales people shipped 6,000 lines of code last week. You can vibe code things through AI and non-technical people contribute to the software product for the first time ever. 

Now, it’s the rise of the “idea guy.” It used to be the idea guy wasn’t valuable and you had to find a technical founder. Now it is reversed. The idea guy can just build things now, which is remarkable. I think that understanding is still not deeply embedded today, because the mental model is that software is really hard. 

The reason I’m bringing that up is if you just play with the raw models and try to build things with it, try to have it teach you things, you get an increasingly good intuitive understanding of what these models can or can’t do. That’s probably the most useful thing to evaluate the hype around vendors saying, we have this matchable thing. You can actually have a much better judgment around if that’s actually true or not.