How AI is Being Used to Make Better Decisions and Build Faster Models with Nicholas Moen

What's great about a tool like Lovable, you can use cloud for this as well. This is what I've challenged my team to use this stuff for is Lovable is great for making that micro app that's completely bespoke to who you are. So if you have that one process that's just uniquely you and you, I wish there was software for this, but no one's going to make it. That's what Lovable can do.

Like, it can make those, like little micro SaaS apps that you've always wanted but never could have because no one's going to build it. There's not a big enough market for it, and an engineer is not going to build it for you internally because it's just so, like, nuanced and bespoke. Welcome to SP&A Unlocked. I'm your host, Paul Barnhurst, and this podcast is all about fpa.

It's where finance meets strategy. Each week we bring you conversations and practical advice from thought leaders, industry experts and practitioners who are reshaping the role of FP&A in today's business world world. Speaking of strategic impact, we have here with us today Nick Moen. Nick, welcome to the show.

Thanks for having me. Very excited to have you. So we're, uh, we're continuing our series for those who are wondering around AI. We got a lot of episodes where we're going to be, uh, talking about people that are leaning in heavily to AI.

So before I, uh, get into that, let's go ahead and introduce Nick, share a little bit of his, his bio and his background. So, Nick Moen is a CMA and Director of Finance at Section, an AI software company that helps enterprises, enterprise organizations deploy AI to their workforces. At Section, Nick runs finance people, ops and operations, actively deploying agentic workflows and an AI native tech stack across all three functions, rethinking how each operates. Before Section, Nick spent nearly six years as a finance operator at Waste Technology company where he built and scaled financial systems and teams across a combined hardware, software and services model.

He has a track record of seeing through operational complexity, identifying bottlenecks and building systems to eliminate them. Love the background and once again, welcome. Thank you. Alrighty, we're going to start with this question.

I always love to see how people answer it. In your opinion or, uh, kind of from your view, what does great FP and A look like? How would you define it? I think nowadays a question, the answer to it is a bit of a moving target.

And what I mean by that is I feel like with AI and how it's rapidly iterating. Because I remember when Opus on, uh, Claude Opus 4.6 came out, I was not using Claude, I was using a different model and I wasn't really doing as much finance. But Opus 4.

6 really transformed how I interact with finances and it's almost completely reoriented how I work. So um, I'm going a bit of a long way around this, but even in the past five years I've worked, I've gotten a little bit more removed, especially with AI from the day to day weeds, maybe Excel spreadsheet building or anything like that, and moving more and more towards essentially the decision framework of FP&A. So when I think FPA, I think this has kind of always been what FPA has been about, but I think it's more about is how can we reach the right decisions and information in the business.

And AI is really pushing us to that. So in my mind, great, uh, FP&A today is what great FPA always been, but just less of the modeling, more of like the analysis, deciding what the decisions are, what are the key levers to the business and just rapidly getting there more quickly and just focusing more on that. And I think it's just going to continue to be pushing teams into that direction as AI just gets more advanced because one, the business is just going to move faster too.

Because finance is not the only, it's not the only part that's getting impacted by AI. And if AI is like velocity, you can do more quicker, that's going to impact the rest of the business. And I mean if you work alongside marketing, sales, you know, the more high velocity teams as finance, you, you kind of always like, I don't know me, sometimes I feel like I'm trying to keep up. They're going to move faster too.

So I think FPA teams will need to move faster to keep up with the rest of the business. But they do, they will be able to. And I think the AI will be a good equalizer between that. So in my mind a, uh, great FP&A is back to what FP&A is on a fundamental life level, which is business partnering and be able to be in lockstep with the rest of the business.

And I, I honestly think that's, AI is going to be the almost catalyst to that. Yeah, I've, I've heard a lot of people say something similar if it really allows us to focus on what we should have been focusing on all along and less of the data prep and the cleaning and all the fun things that we all enjoy. Right? I love it, too.

Yeah, I love doing that stuff. And it's. It's kind of like. It's like a bittersweet.

It's like you're kind of giving it up, but at the same time, you're kind of doing what really drives the value, too. It's exciting, for sure. All right, let's get into AI and Claude. You know, you talked about kind of AI as a velocity tool.

So, you know, when you and I chatted, and I want to start with this question, you mentioned that you haven't used a spreadsheet for a new analysis in a few months. I think you said since, like, January or something. But basically, since you began using Claude. Talk about that.

I would say, again, it was like Opus 4. 6 was the turning point for me. Like, I stopped making spreadsheets, and I didn't stop using them, but I stopped making them. And at a certain point, I just started asking Claude to make them.

Now, I gave it all the information, all the context it needs, but I started to realize that what Claude's like, the. It's the first version it makes, it's actually pretty good, especially if you tell it the right things. You kind of have to treat it like an intern still. Maybe not even the intern.

I think it's gotten a little better than, uh. Like, if you grade these competencies based off, it might be the junior animal or whatever. Yeah, it's more of, like, a junior role now. It's kind of getting a little bit more smart.

It's. It's. It's done with college. It kind of has a few reps under its belt, and it's like, you can probably give it a task and can give some fairly coherent.

Yeah, it's. I think it's at that point now. Now I don't think it actually nails it right out of the gate. Usually, uh, I have to take the last mile.

It gets me 80% of the way there. And then I bring it into a spreadsheet, and we use Google Sheets. So I bring it to Google Sheets and modify it, challenge the assumptions, maybe tweak the assumptions, maybe fix some kind of dumb formulas that it did. But I think for the most part, it does a lot of the bones and the framing quite well.

And that's time. Like, if you open up a new spreadsheet, you have to build that. And, like, I'm assuming you've done a lot of modeling, but it's, uh. And I'm sure a Lot of people listening have done a lot of modeling.

Those framings are not like exciting work or anything like that. You're building out your assumption rows and your tabs and all that. And especially if you can give Claude kind of your way of modeling, you can especially replicate that. It can build that like version one for you, and then you can take it and model it from there.

And it's a meaningful amount of time. And especially when I'm building and I'm, um. I kind of have an idea of what I need to model, but it's not fully formed. So it helps me talk it out with Claude.

That's kind of how like, I kind of like go back and forth. It gives me kind of like a, a mock up and I'm like, no, that's not it. That's not it. That's not it.

I need to. What if we change this variable? Like, uh, no, we need to rethink how we waterfall this out. Uh, I was actually doing that today and it can reiterate and even rebuild an entire model a lot quicker than I could rebuild an entire model.

So if I just start off a bad wave, my assumptions were just all wrong. And then when we built it out, it's like, no, this is not working. We need to go. It can just take it apart and rebuild it a lot faster than I can.

And then when I feel like I actually have something good, then I bring it over and then I own the spreadsheet especially. It's like a reoccurring model. Then I refine it. I get the formatting correctly.

I will say cloud is still bad on the formatting. It's really bad in formatting actually. But that's okay. That's an easy fix.

I can do that part and I'm still doing that. Like, I still haven't had a reason to change that method. It's. It allows me to build things a lot more quickly.

And how much time do you spend validating? Because we all know Claude or any AI makes mistakes. How do you think about that? With all of it?

Yeah, I think that's a good question. I spend a good amount time validating, but I, I put it kind of on a sliding scale of like who is using this information and how. What decisions is it driving. If it's pretty low stakes, yeah, I'll go look through the assumptions and make sure the formulas tie on everything like that.

But like, if this is for like, let's do an extreme example. If this is for like reporting numbers to the, to the board or CEO. Yeah, I'm probably going to date pretty, pretty hard and just check everything else out. But I'd be doing the same thing with any employee as well.

I kind of see that the same thing. And it's really like, it's a balancing act of like, how high stakes is this information versus how thorough you need to be. Because one thing I also do is I'll take the model and have Claude, or you could use ChatGPT or anything and I'll tell it, okay, now go check your work. Like go look at everything.

Go look at the formulas and actually does come back. It's like, oh, I made a mistake. Let me go fix that real quick. And then you do that a few more times and say, okay, now go check your mistakes.

Go review your work. You can even copy and paste the prompt and it's going to keep doing it over and over again. Eventually it'll say, yeah, everything looks good. And then you can take that and then you could maybe save yourself some.

You could still check everything. You're still going to want to check everything, but it's going to be pretty clear, especially if you tell it to design it. Like in a. More.

I'm very much a fan of the KISS principle. I want to keep things as simple as possible because that makes it as easy to audit. So if you have a design, it's also a lot easier to validate everything as well, especially if you have the assumptions flow instead of having one behemoth of a formula. Makes sense.

And yeah, no, I've heard someone else say to me, like, if you're using, you know, AI, make sure you ask it to check itself at least three times for model building. Kind of goes in line with what you said and it will find mistakes and get better. We asked one to figure out, you know, early on, none of them could tell me why the balance sheet didn't balance and we'd always ask them to correct it. Favorite answer I got along the way was, your balance sheet is out by 1.

2 million. That's only, uh, you know, 30. 30 basis points. That's an acceptable error rate.

No, you don't understand how balance sheet works. That is not acceptable. You're no longer the balance sheet. You are fired.

Yeah, actually, I think it was early. They may have been using Claude. It wasn't Claude's Excel, but they were using one of their early models. And this was six months ago now, but it even a little longer and it was like, oh boy, we got a ways to go.

If this is the Type of answers we're getting. Yeah. Oh, one, one thing is actually I got this from our cto and I quite love this prompt is, um, he just simply says, what would you do if you were me? Simple thing.

And especially since these tools are kind of understanding who you are as a person, the more they use it, they kind of understand you. They know, hey, I'm Nick, Director of Finance. And it kind of gets my analytical profile and it'll just run through the same things, but maybe it'll take it from my perspective. Kind of like that prompting method where you assign a Persona.

You're kind of doing that in a way of assigning your Persona to the cloud or ChatGPT. And then it's going to take that approach and think of it from what it thinks my perspective is. And then again, copy and paste that three more times and it's going to keep running it through and running through. It's actually quite effective.

It's one of those classes cases. It's a very simple thing. You can say that's very powerful. I haven't tried that exact one, so I'll have to give that a try.

I was already thinking in my head, I'm wondering what it would say about this. So thank you for that. It's a good, good one. The other one is just, you know, I'm reminded to keep checking and so I think, you know, we'll keep going here.

But it sounds like in the spreadsheet your advice to people is just start using AI to help you build. I think that really what it comes down to is like, don't try to boil the ocean. Don't take your most complicated thing and try to have AI do it. Keep that to yourself.

That's okay. Make, make that your special sauce. But like, try to eliminate all those little things that you don't maybe are not as value add. It doesn't even need to be FP&A.

It could be your emails. Get an agent to run through your emails. It could be having it summarize. If you, if you have Slack ah, and you integrate with Slack, have it summarize your Slack messages for the day.

Maybe teams, anything that's not so to speak to that special sauce. Like, try to get rid of that so you can focus on those things that are really valuable. Especially if going back to like what great FP&A looks like, like anything that gets you closer to business partnering. I think, I think that's really how.

And a great way to start using AI is try to eliminate those and it Might be a few five minutes here, might be 10 minutes there, maybe 30 minutes there. But all those, all those minutes add up to hours over the course of the month, which gives you more time to actually put your brain power to something that's actually valuable and that requires brain power. That's good advice. I know you mentioned you've been using Claude code and Lovable quite a bit.

Had you ever written code before you started using them? I was fine. Decent fba. Not going to say I'm the best enough to kind of hack together solutions, struggled through SQL, SQL, um, enough to accomplish a few projects.

But I'm not going to say I could. I could probably look at SQL code and be like, that's. That looks like gibberish nowadays. Other than that.

No, I don't code. I don't. I never really had strong desire to learn. Yeah.

So with vibe coding, like, I, I know it's making code, I don't know what it kind of means. I could probably read some variables, especially with this, like string text. You could probably work your way backwards. And I've, I've modified string text.

But, yeah, I'm not, I'm not a coder by any means and nor would I ever claim to be one. So what made you decide, hey, I'm going to try to code some stuff. Was there a need where you thought, hey, this is supposed to be good enough, I can solve a problem, or it was just like, hey, this looks interesting. So how did you decide, hey, coding, you know, using lovable others could potentially help me in, in my work?

Yeah, I was given a lovable subscription, so I decided to try to use it. I mean, the real story is it was some. It was an AI summit and they had like a Vibe coding, like, here's, here's lovable, and we're going to Vibe code. And it's like, okay, well, I'm here, I'm going to do it.

And I was like, blown away that I could have a mockup app in five minutes of prompting. Like, I'm talking like an actual, like, SaaS app now. It was, it was a mock up. It was not like, nothing production.

It was, I believe what it is. Uh, it was an inventory tracker for my sister's, uh, nonprofit where she collects baby supplies for families that need it. And she has like a bunch of supplies and just random bits and end. But she doesn't, like, have a way to track it.

So it was in a Google sheet and it was kind of a mess. I'm like, uh, hey, this is A great use case. It's pretty low stakes and let's see if I can get some inventory tracking. It did it so well and I was like, this is crazy.

Like it was so easy. And then to iterate on it. If, if you're using Chat, GPT or Claude, you can use Lovable. It's the exact same method and it's just this exact same prompting mythology.

And just because it's writing code is not really any way to be intimidated by it. It's doing that for you. You don't even have to worry about it. I'm not going to say the code's great.

Uh, I've heard from, you know, engineers I've talked to, it's not good code. So if you're trying to make it like a app to sell on, like to the world, you may want to engineer to take that code and make it not breakable. Sure. You might want to treat it as an mvp.

Exactly. Or, or a wireframe or. Yeah, whatever you want to call it. But not production ready, so to speak.

Right. Exactly. And uh, I, I would. It.

It m. There may be a time when it gets there. I don't think it's that time right now, especially if people are going to pay you money for the app. But it's pretty simple and pretty quick to use.

It's just, I think for me it was starting because it felt intimidating. It's like, I can't build an app. Like, that's not my skill set. And then when I did like one in five minutes, I'm like, oh, I can build an app.

This is like, this is. This is different. This is way different. Yeah, no make.

Makes a lot of sense. Uh, I know you. So you ended up building a commission software for the company. How long that take you?

Less than 60 minutes, honestly. And how often are you using it? Uh, we use it every commission cycle now. It is completely calculates our commissions.

And did an engineer have to do any coding with her? You did a hundred percent of it. Is it sitting in the cloud? Talk a little bit about that.

Yeah. So what we did and we're. We were iterating on it. Uh, uh, but it's.

You're not going to claim it's perfect. No, it's not perfect. The calculations are perfect. I'm going to make that clear.

Our reps are being paid accurately. But that being said, I would say the app is not. Clarification point. I'm reps listening right now going, what the.

The. The apps are like. The apps are good for me to use. I would not give it to somebody else because it's a bit more of like my brain as an app and no one wants that.

But that being said, how I actually went about it is we have, you know, I, I don't want to say it's a simple commission structure, but it's. It's a very nuanced one and there's multiple triggers. I gave, I actually did this a two part mostly because I, uh. Lovable has more expensive to use.

But I gave Claude our commission agreement and basically pull the calculations out and the triggers, analyze it, understand it, then give me a prompt that I can give to Lovable. I did actually say lovable. It gave me a prompt, took that prompt, copy paste in the lovable and lovable, gave me basically something that was accurate in the first prompt, but it wasn't quite what it needed to be because I couldn't finesse it as I need. Like I hard coded the rates.

I didn't want to hard code rates. I want to be able to change those. So this would have been a very complicated workbook to make. And that's actually what kind of started it.

As I started making the workbook and I'm like, I don't want to do this. This is, this is so much uh, I would have to do this per rep. And there's just so much I'd have to build out. This is at least 300 rows.

So I'm like, you know what, let's try it. At worst, I'm 30 minutes out of my day if it doesn't work. But if it does work, I am sitting pretty. So I gave it to Claude and gave it to Lovable and I'm like, this is way easier.

No, I'm already thinking of. I've been meaning to. There's some. There's two things.

I'm wanting to vibe code and I just need to sit down and do it. But what are those things? They're. They're not FP and A things.

They're just some stuff for my business. Some. Yeah, process things that some, some I want to get on my website. You know, basically some questions you'd fill out.

It would give you different options, type of things. See how kind of questionnaires to give you options. So yeah, that's a great use case for it as well. Yeah, no, I think it would be good.

I have the data now. I've done a lot of the work on all the back end stuff and I'm like, okay, I'm sure it could Help write some of the logic and just work through it all and see what I can build. It's just time. You know how it is.

You just gotta sit down and do it. We all pick where we spend our time and I haven't spent enough time on it. Well, I think that the touch is a good point. Like, what's great about a tool like Lovable?

You can use Claude for this as well. This is what I've challenged my team to use this stuff for is Lovable is great for making that micro app that's completely bespoke to who you are. So if you have that one process that's just uniquely you and you wish. I wish there was software for this, but no one's going to make it.

That's what loadable can do. Like it can make those like little micro SaaS apps that you've always wanted but never could have. Um, because no one's going to build it. There's not a big enough market for it and an engineer is not going to build it for you internally because it's just so, like nuanced and bespoke.

That's where I'm finding a lot of power in it. Just something that's. I need this automated, especially with agents nowadays. Agents also kind of filling that gap as well.

All right, so you, uh, when you and I chatted a few weeks ago, you mentioned you finished an analysis for the CEO Lou using clot on your phone while putting your son to bed. Tell us the story. Yeah, yeah. So this was like classic end of day request.

It's like, it's okay. I was gonna log off and go make dinner. Obviously I realized that this needs to be done. Yep.

They wanted it yesterday, right? Yep. So what I did was it's. It wasn't like one of those things where it's like, okay, I'm going to just get this done in 30 minutes.

This is probably, this is like a pretty massive deep dive in our model assumptions. So I gave Claude like all the documents it would need. That's on my computer. I gave her a model, I gave it other supporting documents.

Anything that it would need to do, need to have for this analysis had it as a conversation. And then I closed my laptop, made dinner, played with the kids for the evening. My youngest is two. Uh, so we still put him to bed and like any two year old, he lays there awake for an hour, just not going to sleep.

So I knew that, okay. I had all the context I needed in the chat. So I start, I pull out my phone And I just start prompting like it has the context. So now let's work through the analysis.

So I'm just telling it, this is how you analyze it. This is how you should look at it. And okay, now build the spreadsheet. And I'm like reviewing the spreadsheet on my phone.

That was painful. But I wasn't typing in it at least and I was understanding the assumptions, the model, redoing it. And essentially we were supposed to, we wanted to view some cost assumptions in buckets. We didn't really have it modeled out and viewed.

It basically just described everything it needs to think through and how it can pull the assumptions, where it can pull the assumptions and, and how it can organize those assumptions better based off like the vendor or the supposed vendor. Then structure that into an output that I can give back to the CEO. That's bit more to the point. Cut all the fluff and gets get direct to the point.

And I did that over the course of just the night, just putting my son to bed. I didn't need my laptop or anything. I think that to me was like this is a completely different way to work. I never thought I could do a full financial analysis on my phone.

Yeah, you gotta get creative when you have kids. Oh, what can I say? I do. I do have a daughter, but a little older now.

She's 13. So she puts herself to bed. What's your thoughts on the whole idea of AI is going to take our jobs? I think the answer is nuance and I, I feel like that's just a bad framing in general.

It's gonna, I'm not gonna say it's not gonna eliminate jobs. Of course it's gonna eliminate jobs. But I think it's not as simple as. It's just gonna completely decimate work.

And to be clear, I think any company right now that's saying oh we eliminated so many jobs, so we're layoffs because of AI. That's not, that's, that's false. They, they just over hired. That's not the case.

They're just using. They just found a, they found a serendipitous moment where they could have a reasonable excuse to just lay off 20,000 people. I said that was my view on uh, when um, Square did that. Mhm.

A lot of people disagreed with me. Some agreed. I, I could believe there's some that are due to AI, but when you're making huge cuts and saying it's all AI, you over hired. Yeah.

So I, I'm mostly with you. I think it's a, it's nuanced. But yes, there's definitely an over. Higher component.

No, I, I do think they have AI gains. Sure. And maybe some of those rules were actually cut because of AI, but not 50%. No, I don't think so.

I would like, I'd be very interested how they've utilized AI. Uh, if they cut 50% of their work, I would as well. And I agree with you. I mean I talked to somebody who is very close in the investment banking industry and he's talking to the banks all the time and he said right now they can't eliminate a single position because of AI.

Now they have people that are all being more efficient, but they're not at a point where they could actually cut people that will they get there? I'd, uh, say yes. I think most, most areas will get where there could be cuts. But this idea that, you know, huge layoffs are all due to AI.

I'm with you. I don't, Yeah, I don't buy it. I think that's a. Messaging versus reality for the most part.

I think where the answer gets more nuanced is I think it's going to have a reckoning on entry level work. Yes, I've had a lot of discussion around that. Yeah. And I, I think it's going to be a question, I think collectively as people, how we figure that one out.

Because obviously if you don't have Internet entry level workers coming in, you don't have people to train up to be back into middle career and senior career. So you basically severed the top of the funnel. So there's nothing going in the funnel. I don't think I have an answer to that.

Very hard answer to find. Because what a lot of AI is really good at nowadays is the entry level work. Yeah, no, I, yeah, I've said, okay, there has to be some changes the way we do education. There's a question of how much time you have to invest in somebody.

And fundamentals, what does that look like? You know, if you think of a lot of skill trades, right. Uh, you do an apprenticeship for several years, is there almost going to be some kind of apprenticeship type thing where you're working along AI so you get those good fundamentals. It'll be interesting to see.

But I'm with you. Something has to change. I don't know the answer either, but it's a fascinating thing to discuss and to think about because I think education has to change. I think for many companies that entry hiring also has to change.

How much of each what does that final look like? I don't know. I don't think anyone knows yet. I, uh, think to the point, entry level work exists.

It you just raise the bar, raise the floor to be more complicated. Of course, then you have the problem of having people, the skill sets to do that. Uh, the apprenticeship is a very interesting concept and I'd be willing to bet that companies would be more and more interested that as a recruiting tactic, it's not like, oh, uh, we're going to just take these unskilled people and get no value from them. It's, it's very much an intentional investment approach.

And I think you see that with bigger companies that have like, rotational job programs for new workers and stuff like that, they exist. The. It's probably just going to have to be more of it. I mean, it also creates a little more risk in that, you know, the younger workforce churns jobs a lot more than we used to historically.

And if that first year you're getting less value, do you really want to. You're taking more risk in the sense of knowing they could leave early. So it'll be interesting to see how they kind of work through all that. Yeah.

Uh, and at this point we're getting out of finance, talking more into talent retention. But I think that's to your point, though. It's like it's, it becomes a function of how do you retain talent. Now you got to be a bit more competitive if.

Right. But at the end of the day, I mean, it's an FPA show and many of us lead people, so these are things we have to start thinking about, you know, FPA as well. But definitely we've got a little out of the typical lane, but I'm trying to bring it back. Yeah.

Yeah. Well, uh, I think even for my team, I do think very consciously in how to keep them retained and keep them engaged and like a lot of work to recruit and retain very exceptional people. Hiring people is expensive and it's easy enough to make mistakes. You don't want it to be.

You don't want it to compound. And so you got to think about these things. Yeah. And I think that's what AI is going to push us more into as well, especially as senior leaders, is we're going to have to think more about our teams too.

I mean, we, uh, already do, but I think we're going to have to do more of it. Especially in the early days where we especially have to. If people are good at AI, we wanted to make sure we want to keep Them because it's harder to replace that maybe over time that will normalize a little bit. But sure, right, right now it's like anything early computer, early spreadsheet, uh, there's a new technology, you want to keep that person that knows it well because it's going to be hard to replace them.

Five years from now could be a very different story. Chaotic time, got to change the frameworks every six months. It's a time of incredible change. It's, uh, Part of me is like, man, I would love to be working for somebody and you know, kind of automating and seeing all this.

And there's another part of me was like, no, I wouldn't, you know, like I was saying that the other day because I have someone explaining to me the stuff he's doing with Claude. He, he has to be in the, uh, 1/10 of 1% of finance people, if not even higher. I mean the, the way he's thinking about the stuff he's doing is just off the charts. And it's like, that would be really cool.

And then I'm like, that'd be a ton of work. But, you know, I'm figuring out more and more use cases. So kind of speaking about use cases, I want to ask a question here on that. And then we're going to go back to training for a minute, but in a little different aspect of talking about the whole, you know, new employees.

So what's the one workflow that you think every FP and a person should automate first? Where would you say they should start? I mean, obviously I know it's not going to apply to everybody, but what's kind of that low hanging fruit that you say, if you're not doing this, try it? This might be a little bit of a different answer, but for me, context is everything, especially when I'm building assumptions, trying to understand things.

So getting a lot of the business, partnering. So one, one thing that I utilize a lot is a software called Granola. And I mean, I think we've all been on calls, sales calls or any type of call where like a bot comes in and just records everything. Yeah, I join every call now and there's usually some note taker before there's a person.

I'm like, I'm here with your three AI tools. Thanks. Joining. Yeah, they're five minutes late.

So you're just kind of sitting awkwardly with the robot. I'm saying things to it. I hate to see if they notice you have to test their, um, recap notes, but Granola is A little bit different. Uh, it sits on your machine and it transcribes.

So it doesn't actually need access to the meeting. It just, I'm assuming it takes in the audio for the meeting and just transcribes it. It's actually pretty good. It's actually really accurate.

And one thing I like about it is completely agnostic what you use. So if you need to do a teams call, it's not on your calendar, it'll pick it up and start recording. Why this is like probably one of the most key linchpins pieces of software for me lately is I take all the information I get from granola and feed it back into Claude. Because they have an MCP they can connect.

I have it query a lot of the context that I pull from meetings. So I don't need to really focus on note taking. I could just be engaged with the conversation. I mean, today I was building out assumption model and I already had a lot of conversations about this assumption model.

So now that I'm actually building it, I'm querying Claude that's pulling from my granola, get all the conversations and put everything in a neat, tidy format so I can just jog my memory. I think that's something like that is incredibly helpful for me to be more effective at picking smart assumptions, building out things that are actually impactful to business, and more importantly, just gets everything grounded back to reality. Because I can think of all the assumptions and how things should be done.

But it's, uh, Mike Tyson quote, everybody has a plan till you get punched in the face. It's kind of the same with my plan's perfect until it hits operational reality and then all the assumptions go topsy turvy. That to me has probably been the biggest lift for finances, even if it's not a finance workflow. So I appreciate that.

Thanks for sharing that. Let's talk training. How are you training your team? Any tips?

I mean, obviously you guys have been using this pretty heavily, have several people kind of rolling into you. So how are you going about training and making sure your finance team is using AI and that you're consistent across the org? Yeah, well, uh, I would, I want to like, qualify this as I'm very fortunate that the company I work, uh, at, I know everybody has different security policies, especially depending how large they are. We're AI transformation company, so naturally we're going to use any AI tool we have because it's a strategic advantage.

So I have access to a lot of AI, uh, tools. That being said, how I Get my team to use it more often. I do have kind of advantage that my team is motivated to use it because it's kind of baked into our culture. But tactically, I think one big thing is hackathons, like team hackathons, just to sit down, carve out some time, and like, let's pick Lovable as example.

Uh, it's like you pick a use case that you need. Start by coding Lovable. We'll all do this together. We'll do our own thing and do a show and tell at the end, see which one people like.

And I like this approach for two reasons. One's it gets people actually using it. And I think a lot of part of AI is just people not familiar with it. It's a tool, it's a software, but it's not like traditional SaaS.

It's a completely different kind of software. So the only way to get used to it is to use it. So that gets people used in it. It gets past the point I had with Lovable of like this, like the starter gun, like, like at the.

At the beginning. Beginning of the race, it's like, okay, now when I start running, I can finish the race. Gets people started on it so they feel less intimidated. And then it has.

The second thing is it gets people to show what people have made. And then you can ask questions, well, how. Why did you do it that way? Or how did you even do that?

And then people start knowledge sharing. Because I think that's the other. That's. I think that's the other component of getting your teams trained is you want them to share knowledge.

There is no. There is no progress in training if people are, like, hoarding AI knowledge themselves, whether intentionally or unintentionally. So knowledge sharing is quite important. It's like, hey, I did this.

Or I noticed that AI can do this now. Or like. And that has almost like a compounding effect on learning. Because now people, uh, I mean, this.

This is also just kind of basic learning principles in general. But when people can work off of each other, they can. They can move a lot more quickly, a lot more fluidly instead of. Because now let's say you have a team of five, and they're all just in isolation using AI.

Every discovery is just them learning more versus if they learn different discoveries from the four other teammates. Now you have five other potential events of discovery on how to use AI. And I think that's the big thing about AI too, is I don't think everyone's quite even fully realized what it's capable of. Yeah, I, I would agree.

I mean, we're still all learning. I don't think anyone, anyone who claims they fully understand the capability is lying. That would be my opinion. But, uh, uh, there's.

People are quite good at it, I'll give them that. But, yeah, I don't. I think we're seeing where it's an iceberg, where we're just seeing the top of it. Well, it's a lot like Excel.

If anyone tells me they understand all the formulas, all the functions, all the languages, they're an expert in everything in Excel. I'm going to call BS. Now, multiply that times 10 for AI. I mean, the best in the world is maybe using 20% of Excel.

The best in the world might be using 2 to 3% of probably what all the use cases and capabilities are for AI. I thought it was pretty good Excel. And then I watched the, the Excel championships. I'm like, I don't think I'm good at Excel.

Yeah, I've, I've interviewed several of the world champions. And you want to be humbled on your Excel skills, Watch them for about two minutes. Yeah, I felt that way when I tuned into that. Uh, it was a few years ago.

I was like, wow, you want a great. I'll share this to anybody. Um, uh, so last year's world champion, he's been a world champion before. Dim, early, dermied early.

He's originally from Ireland, he lives in New York, and he's a Microsoft mvp. And he recorded a video with Nadella, right, the CEO, where he's showing him m about the formula he did. He's teaching Nadella Excel. And he's looking at.

He has this look on his face like, wow, you could write something like that. Like you could do that. I mean, just a classic, like, what better, uh, promotion video can you have? Then you're teaching the CEO of Microsoft how to use Excel almost, you know, like you could see him just having this look like, wow, you're doing that in Excel.

So that'd be, that'd be a great moment. Yeah, it's a really good go out on you. He just put it up this week. Go out on LinkedIn and you'll see.

Yeah, it's worth watching. It's just two minutes long, but it's a, it's a good little video. I know, I know the guy. I've interviewed him.

Great guy. But we digress there. So there's an example. All right, so we're getting close to our time, so we're I'm going to pick a few more questions here.

First I like to ask a couple FP and a question. So I'm going to ask two and then we're going to get through a couple get to know you questions. What's the number one soft skill FPA professionals should master? Yeah, I'm going to give you two answers and try to make them quick.

I think the one is kind of obvious is partnerships. AI is going to really stress that we're going to have to be very good partners because that's the one thing AI can't do. I'm still on the belief that people want to talk to people generally. So I, I am as well.

I agree with you. Yes. So I'm, I'm banking on that. But I think the other soft skill is learn how to be a product manager because the way, and I've, I've been fortunate to work with product managers and if you're not in tech, it's just basically the person that helps design the product and gives the schematics to engineering.

They take what a customer wants, understands what they need and then has engineering to build. So a good product manager can build, suss out what a customer needs. AI is very structured to think that way. At least right now it is.

It's being able to understand a question behind the question or like a need behind the need and essentially ask really smart questions, really think through things in a very nuanced, even first principles way and describe things in an articulate, clear manner. You can almost think of your LLM as the engineer. So you need to be able to translate a vision, a thought into something that Claude or ChatGPT can understand and build correctly. And I think that type of skill will extremely benefit anybody in finance as they're trying to build things.

Especially having a model being built into a spreadsheet. Prompting is, well, live or die, how good that model will be. What about technical skill? I still bank that understanding data structure is probably one of the biggest technical skills.

Probably even before AI, I went pretty hard, um, I don't know data scientist, but I got pretty good at least competent enough to navigate data schemas to at least understand it. And I think that's just compounded pretty massively just being able to understand how data's structured, especially with AI and be able to structure it so it's contextable for AI, it can understand it a bit more clearly. Messy data added, garbage in, garbage out. So you know to structure it.

When I'm interviewing people that are using AI A lot that is becoming a more and more common act or some kind of data thinking, systems thinking, you know, understanding data structure, what. They all answered a different way. But that theme, I might have heard it once in the last year. I've now had it three times in the last week.

And that's an area I'd always prided myself I was a little bit more of a data person in FP&A. Yeah. And I'm seeing now. I've always said that's a very important skill, and it's just getting raised up the.

Up the list. It's gonna be a nonprofit. We're hearing Excel is the number one technical skill. It's really interesting to watch this transition.

Yeah. All right. We're gonna get to know you a little bit, and I might. I might tweak some of the questions I have here.

I might have a little bit of fun. Are you a music person? I. I love to listen to music.

I've tried and failed to learn an instrument. I do get and do enjoy music theory, even if I can't quite grasp it fully. We'll go with the music question then. You could only listen to one album for the rest of your life.

What album are you picking? This is like pick my favorite child. Yes. What is your favorite child?

I'm just kidding. Go on. Let's get that on recording this to your wife. Yeah, I told my kids I'm gonna be on YouTube.

And. Yeah, let's. Let's get that on recording. Uh, yeah.

So what's my favorite album? I. I will say I'm gonna. I'm gonna do a cheap thing.

I would say one of my favorite albums right now that I could listen frequently because I just started. Started getting. I just got a record player and started collecting records. Um, the one.

The one I'm listening to a lot. It's. It's, uh, Black Holes and Revelations by museums. Looks phenomenal on record.

I've been listening to Muse for quite some time, and it's just. It's a great end to end album. You can listen from beginning to end, and it's just the story that built itself. Nice.

All right. This is going to be another kind of fun one, but it's much easier answer. Even though you haven't prepped Claude a he, she, or it, I think I say it more than anything. And have you named any of your agents?

Oh, okay. So this is lame to admit. I, uh, have an. I have an agent that helps me fill out bank forms because I got sick of writing bank forms.

For vendors or, excuse me, customers. So I made an agent that could fill out the forms for me. Phyllis. As in Phyllis form.

Nice. Oh, fun. And then last versatile one and then we'll wrap up here. If you could go anywhere in the world tomorrow, you're taking a two week vacation.

Where are you going? Maui. Maui. Have you been before?

I've lived there for about five years. Ah, uh, got it. I know it on a local level and it just, it's one of those places that will probably never be replaced as my top destination point. Alrighty.

Uh, as we wrap up, first one, any parting advice you want to give our audience around being a better business partner? As we know that's becoming more and more important. Any advice you give? Yeah, I would say learn some skills from your sales team.

Good ones. Good. And uh, let me. Not the diva ones, in case anyone wants.

Not the diva ones. No, the ones that are really good at selling and you'll, you'll know why they're good at selling is they're very good listeners. They really try to understand what your problem is and understand like the pain that you're. They're trying to solve and they really.

They. A let's. Someone would say this to me all the time is um, listen to understand, not to respond. So being able to uh, really pay attention and understand what people are saying, understanding their issues or what they're thinking or their mind is.

And not trying to sound smart or anything, just really trying to understand and listen is going to probably do more things than any other skill or technique you could ever do when it comes to partying. What you said there, the way I sum it up is, uh, on the listening, one of the best ways I've heard is. Is. Stephen R.

Covey, 7 Habits of Highly Effective People seek first to understand, then to be understood. Yeah, I just, that that's what I always think of when I hear that. It's just so critical and I forget myself doing it. But you want to learn to get better at listening.

Do 400 podcast episodes. It's a, It's a skill too. It's a skill that you have to develop. It's not something, I mean some people probably come across naturally, but it is a skill that you have developed and can develop.

It's. There's techniques out there, you can look it up, but, um, on how to, how to kind of develop that. But it is something very, anyone can do. 100% agree.

It is a skill that can be learned. Different people are better at certain aspects of it than others. Naturally. Just like some people are better athletes in certain ways than others.

We all have our strengths and weaknesses, but most everything can be learned in life, I believe. Yeah, not everything. There are exceptions, but. All right, well, we'll go ahead and wrap up here.

I think this is a good place to stop. So if anyone wants to get in touch with you, learn more about you, LinkedIn, the best way to do that or how should they reach out? LinkedIn's always great. Just tell me that if you message me or connect with me, just message that you heard me from the podcast.

Or else I'm just. You're just gonna probably go into my BDR purgatory and, uh, say you heard. You heard about Nick. It.

You're excited because you saw him talk to some guy with a big beard and. Yeah, yeah, because we'll say it'll be like, wait, big wow, I better read this. Just, just send hashtag, hashtag big beard. I don't know who you.

Where you heard me from. Well, you do know, uh, I'm not sure if you're a basketball person at all, but you know who James Harden is. I don't. I do not know.

Okay, so he's a huge basketball player and his Twitter handle is the beard. There you go. Go look him up online. He has a huge beard.

Oh, uh, that's great branding name. So I'm like. Because people have said you need that and he already has it. Yeah.

And take his handle. So someone's like, you need to do a commercial with, you know, they joke James Harden. Like, you know, great. And I just gotta laugh.

So anyway, well, I think we've covered, We've covered beards, we've covered education system, we've even covered some FP and A. We've covered coding. Do you ever think you'd be on an FP and a podcast and cover those topics? I don't even think I was going to be on any podcast, so.

Well, there you go. Well, thank you for joining me. I've really enjoyed it and, uh, good luck with things. Uh, continue to build out AI and exciting times.

So thank you again. Thank you and thank you for having me. That's it for Today's episode of FP&A Unlocked. If you enjoy FP&A Unlocked, please take a moment to leave a five star rating and review.

It's the best way to support the FP&A guy and help more FP&A professionals discover the show. Remember, you can earn CPE credit for this episode by visiting earmarkcpe.com downloading the app and completing the quiz. If you need continuing education credit credits for the FPAC certification, complete the quiz and reach out to me directly.

Thanks for listening. I'm Paul Barnhurst, the FPA guy, and I'll see you next time.