Autonomous Finance and the CFO's Next Frontier

Read a transcript of the conversation below:

Jen Rogers: All right, gentlemen. So Miriam also had this great analogy. She talks about how not everyone used electricity when it first came out. And we all fly in planes now because we have this framework that's all around them. And we have faith in it. So as we're at the beginning of this transformation, really, what are you finding? It's a learning curve, right, for people in finance to get their hands around AI? So when it comes to that what is the use case that you see?

Michael Heric: Yeah, so I think the first thing I would say, Jen, is we say, AI's exciting for the future. I think the real question is, what can it do for us now? And I think the first thing I would say about AI in finance, in the finance function specifically, is AI is not one thing. There are different forms of AI. So from traditional machine learning, generative AI or agentic AI, each one's different levels of maturity. So first of all, within the finance function, many finance functions are already using AI in some form similar to what was mentioned before.

So for example, if you take accounts payable and intelligent document processing where people are processing supplier invoices, most companies are using this solution with AI already embedded within it. Then the question becomes other use cases really around generative AI and AI agents, which is a bit newer. And I think there's a whole range of use cases that we could go into from financial planning and analysis all the way to, say, tax and treasury.

A data point, for example, around forecasting about 20% or so of finance organizations are already using AI in forecasting as an example around revenue cost or cash flow as an example.

Rogers: So, Steve, one of the issues is adoption, really, getting it out there. So when you've seen adoption and this start to take off, what has the transformation been like? Is it difficult? Is it easy? What are we looking at?

Steve Beam: I think it's difficult. I think there's a bad track record in corporate finance of adopting new technology in general. I mean, you look at a lot of our clients, and I mean, honestly, your tech stacks are probably being utilized for finance at about 20% today. And so you introduce a new concept, you introduce a new approach, advanced analytics and AI. I think it's about being very specific and prescriptive about what you want it to do.

And so there's successes when you say, OK, we're going to put guardrails on this. We're going to be very, very protective of what we want this agent to go out and do for us. And then you do see results. I think the flip side of that is you open up your doors and you release these agents into every element of your financials, and you're not really sure what they're going to come back with. And so it's hard to find those situations where you've got analysts and people on your team that say, oh, it did what I needed versus it generated a lot more questions.

And I think that's the adoption piece.

Rogers: So you're inside lots of companies and you talk to lots of CFOs. I'm sure you know, they're pretty risk averse. I'm going to say I think sometimes it's a little bit part of the DNA that's there. And just listening to that answer, there's a reason to possibly be risk averse. So how do you help CFOs navigate that?

Heric: Yeah, no, I mean, it's totally understandable. I think also a lot of the things around AI, the technology has just not been as mature. So within finance functions, if it's sort of like, hey, I can use this generative AI copilot, none of the information's auditable. Have no idea where these numbers come from. Oh, by the way, generative AI doesn't really do math well, but I'm going to give it a shot. And I'm going to start replacing stuff that works with stuff that's really expensive. It's just kind of stupid.

And to be honest with you, a lot of the early stage things around AI within finance didn't have a good return on investment. And if you look at the return on investment from basic copilots, for example, out there, the return has been pretty poor. The only thing CFOs get is kind of a bill. They don't actually see a lot of improvement. They don't see a smaller finance function as an example. So I think some of that, just like anything, creates a lot of concern around, is this real?

And so I think it's really overcoming that. And so what we see with companies that we work with is just the ability to sort of overcome that. To overcome just like basic experiments to really use AI to modernize the entire finance function. Not just, hey, I have five random use cases here that seem kind of cool. I've got five people that use it. They tell me it's good. But it doesn't actually move the needle on finance overall. It's just like interesting experiments. And so where you see the big traction is when companies really fully modernize.

Beam: It's a tool. It's an enabler more than anything else when used the right way. I would also say we keep coming back to this idea across industries that normally within corporate finance, as corporate finance goes on a transformation, about a third of the existing work should stop. Reports, KPIs, overproduction of materials, of insight, it's just a lot. About a third of the work is really, really primed for AI to help in terms of value stream mapping and process mining. And then the final third is technology, speeding up those cycles.

We've had a client recently in consumer products, 180 countries, two weeks to do a forecast. They leaned the process, applied advanced analytics and AI. And they're doing it in an hour every week versus two weeks, every month with better accuracy in 180 countries. There are use cases, but a lot of times, you have to prune or release a lot of capacity so your teams can go and focus on the right stuff.

Heric: I think also, Steve, one other thing to add right on that example is it wasn't just, I built an AI model, but actually fundamentally changed how the company did forecasting. Moving from bottom up, forecasting huge, manual production to more of a top down approach, where you use AI, and then the entire process around forecasting actually changes. And to drive that cultural change is really hard because one thing that inhibits adoption is, I use AI, but it's kind of on the side.

And it's an interesting check, but it doesn't actually fundamentally change how work is done. You've got to do both to really get the full benefits of AI. And I think that's the example that—

Beam: We have a flip side to that, which is another client that has 15,000 AI generated models, and they're budgeting, forecasting, and planning cycles for as long as ever.

 It can really, really help. But let's be specific on what you want it to go and do.

Rogers: I've heard you say that it is also a draw for employees. You actually have to do it to retain people out there. As we're wrapping up, what is the exciting part of this that you are hearing from clients when you're in there implementing that you're not hearing other people talk about?

Heric: Yeah. Well, I do think, you know, people are always worried about job loss and that sort of thing. But if you go into a finance department, let's say it's like FP&A, and you look at how much people spend their time on, for example, doing data queries, reprocessing data, putting in different formats. I mean, it could be anywhere from 60% to 80%. And if you go to those people, they're like, I want to do something cool and fun. AI is more exciting than like running another SQL query to, pull out a bunch of information to drop it into a spreadsheet.

So they want to be energized. There was a study in mid-September from Anthropic and OpenAI where they shared all their information on how people are actually using those applications. If you look at Anthropic in particular around enterprise, you can drill down almost like every state and every role. And if you look down at accountants, you see widespread adoption. So what that says is, even though companies are saying you might not be able to use this technology, people are actually using it anyway and they're just using it with these consumer applications anyway.

So I think that that's what they're looking for to be energized around AI.

Rogers: All right, well, thanks for getting us a little bit excited about it.