How AI Agents are Reshaping the Finance Tech Stack with Josh LaSov
In this episode of FP&A Unlocked, Paul Barnhurst is joined by Josh LaSov to explore how FP&A is evolving with data, BI, and AI. They discuss how finance professionals can move beyond traditional reporting by combining financial and operational data, and why understanding data at a deeper level is now essential. Josh shares how AI is changing analysis, forecasting, and workflows, and how FP&A teams can use it to increase productivity and deliver faster insights.
Josh LaSov is an FP&A and data professional with experience in consulting, ERP implementations, and analytics. He has implemented NetSuite and FP&A tools across many companies and previously co-founded Satori Reporting, a BI platform for NetSuite users that was later acquired by private equity. He is now the founder of Causi AI, where he focuses on building AI agents to help finance teams automate analysis, processes, and forecasting.
Expect to Learn:
What makes FP&A professionals stand out today
Why combining financial and operational data is critical
How BI tools expand the role of FP&A
How AI is improving analysis, forecasting, and workflows
Here are a few relevant quotes from the episode:
“AI helps you save a lot of time, but the judgment and analysis still need to be done by you.” - Josh LaSov
“You can’t explain why numbers changed if you don’t understand the data behind them.” - Josh LaSov
Josh highlights that FP&A is no longer just about building reports or models. The role is shifting toward deeper analysis and faster decision-making. While AI can handle large amounts of data and speed up workflows, the real value still comes from understanding the business and communicating insights clearly.
Campfire: AI-First ERP:
Campfire is the AI-first ERP that powers next-gen finance and accounting teams. With integrated solutions for the general ledger, revenue automation, close management, and more, all in one unified platform.
Explore Campfire today:https://campfire.ai/?utm_source=fpaguy_podcast&utm_medium=podcast&utm_campaign=100225_fpaguy
Follow Josh:
LinkedIn: https://www.linkedin.com/in/josh-lasov-7897334/
Website: https://www.talksaasytome.com/
AI Agent Marketplace: https://marketplace.cauzzy.ai/
YouTube Channel: https://www.youtube.com/@cauzzyAI
Follow Glenn:
LinkedIn: linkedin.com/in/glenntsnyder
Earn Your CPE Credit
For CPE credit, please go to earmarkcpe.com, listen to the episode, download the app, answer a few questions, and earn your CPE certification. To earn education credits for the FPAC Certificate, take the quiz on earmark and contact Paul Barnhurst for further details.
In Today’s Episode
[00:00] – Trailer
[02:25] – Introducing Josh LaSov
[04:59] – What Great FP&A Looks Like
[09:52] – Building Satori Reporting
[13:19] – BI Tools and Reporting Challenges
[16:25] – Introduction to Causi AI
[25:05] – AI Agents and Accuracy Challenges
[29:22] – How FP&A Should Use AI
[36:10] – AI Tech Stack & Future Direction
[43:05] – Key Skills for FP&A
[46:53] – Get To Know Function
Full Show Transcript
Host: Paul Barnhurst (00:00):
Welcome to another episode of FP&A Unlocked. Are you tired of being seen as just a spreadsheet person? Well, others get a seat at the table. Well then, welcome to FP&A Unlocked, where finance meets strategy. This podcast was designed to be a practical guide to give you advice from thought leaders, industry experts, and practitioners who are reshaping the role of FP&A in today's business world. I'm your host, Paul Barnhurst, and together we'll uncover the strategies and experiences that separate good FP&A professionals from great FP&A professionals, helping you elevate your career and drive strategic impact. Speaking of strategic impact, our title sponsor for FP&A Unlocked is Campfire, the ERP. That's helping modern finance teams close, fast, and scale faster. This week, I'm thrilled to have with me Josh LaSov on the show. Josh, welcome to the show.
Josh LaSov (01:00):
Thanks, Paul. Glad to be here.
Host: Paul Barnhurst (01:02):
Yeah, excited to have you. I know we connected a few years ago, and then you reached out recently. We had the opportunity to reconnect. So why don't we start with you telling your audience a little bit about your background, just kind of your history and where you're at today?
Josh LaSov (01:16):
So first and foremost, I am an accountant by trade. I have spent many years doing FP&A, so hopefully that resonates with the crowd. But my background largely started out in consulting-type roles, which were primarily FP&A or operations-type consulting. I then moved into the technology side of that work. At one point, I was a NetSuite solution provider. We had our own business where we resold and implemented NetSuite. I’ve implemented NetSuite over a hundred times. We also worked with Adaptive Insights, now Workday Adaptive, implementing it over 50 times. I’ve worked with almost every FP&A tool at this point, from Prophix to Datarails to Anaplan and others like Vena. After building our NetSuite practice and surrounding it with additional tools like close management and AP automation, we sold the business to a larger NetSuite partner and then began our journey into data.
(02:25):
So the next business, which is where you and I got connected, Paul was called Satori reporting. And we had the first prebuilt data warehouse with Power BI on top for NetSuite, for companies running on NetSuite. And certainly, they bring in their other data sources as well. And that was pre-ai. We built that company. That company was bought by private equity back in 2022. And I say all this because we then have taken all that business experience, all that operations experience and consulting and data experience to roll that into Causi AI, which is our latest platform, which is an AI platform to create agents largely for companies running NetSuite. Doesn't have to be, but largely, and that's what I'll certainly talk more about as we go through this. Great,
Host: Paul Barnhurst (03:15):
Well appreciate that background and we'll definitely dig into some of the things. But before we do, I do this on every episode, obviously you've worked in FP&A, you've implemented tools, you've been on the data side, you've been a CEO, so you've seen FP&A from a lot of different perspectives. What makes for great FP&A, what does that look like to you?
Josh LaSov (03:34):
I think it's a really good question and cool question. I think historically a lot of FP&A was I'm going to prepare a really good model in a spreadsheet or I'm going to prepare some really good reports in Excel or in a reporting tool. And I think that that doesn't change with ai, but I think what makes for really good FP&A is that's kind of the core. And then I think you have to layer on two more things to not only be good FP&A, but remain competitive. One is going to be communication and emotional intelligence. AI is not going to replace that. And then the other side is our ability to do more. We've got to do more. We can get into that, but there's just so much more FP&A is capable of that. They will no longer be bogged down doing a lot of manual things. So with this time, they need to do more things to remain relevant and competitive.
Host: Paul Barnhurst (04:32):
So we got the emotional intelligence and do more. And I was going to say, if AI figures out the emotional intelligence, that's when I'm just going to the mountains to live
Josh LaSov (04:40):
Terminator, something like that, that's going to happen.
Host: Paul Barnhurst (04:45):
So as you mentioned, you started your career in consulting. You did a lot of FP&A and operations type stuff, but you ended up moving into data and business analytics. Obviously you ran the company Satori. And so what was it that interested you in data analytics? What kind of steered you toward that path?
Josh LaSov (05:01):
For me, expanding my FP&A skillset into data warehousing, bi, what was really relevant and what I learned was two things. One, I can't just give the financial picture, the F in FP&A, I need to give the operational picture. And a lot of these tools when we work with Workday Adaptive, my biggest complaint from CFOs or customers was, Hey, this is great to build a financial report, but where's my operational data? Where do I look at quantity sold or these other operational metrics or different transaction types that are happening from different data sources? And we're like, go to a BI tool. So I realised that we were kind of stuck in a box if we just work with FP&A tools. Now it's not that box isn't a big box, but can I produce more? Can I do more for the business?
(06:05):
So getting into data and BI allowed me to combine financial and operational data to provide more value to the business, and it required me to learn how these systems recorded data from the input which is process to the output, which is like a table of data. And that made me a lot more relevant and competitive as a resource. Now, a BI tool does not replace an FPNA tool. Enterprise companies have both larger mid-market companies have both. Smaller companies might have to choose because they're not going to spend the money yet. And there's plenty of overlap between the two, but I needed that skillset. So that's why, and I was very attracted to being able to financial plus operational.
Host: Paul Barnhurst (06:53):
It sounds like it really started out a necessity of wanting to understand the operations, and it sounds like it's something you found out you really enjoyed and kind of ended up going more and more toward the data side.
Josh LaSov (07:03):
Yep.
Host: Paul Barnhurst (07:03):
Makes sense. I've seen a lot of people do that, so it's very common. I've always enjoyed the data side of FP&A, so I get it.
Josh LaSov (07:11):
Yeah, I mean, a trial balance only gives you so much data. Transaction level detail and operational systems to me is where the story lies. I can't do root cause analysis. I can't say why my numbers are the way they are, why they change from a trial balance, but I can from the transactions,
Host: Paul Barnhurst (07:29):
No, a hundred percent agree. The trial balance outside of, okay, maybe if you see descriptions on the journal entry, it might occasionally tell you a few things, but outside of that, it's really hard without the detailed forecast you built that comes from operations or operational data itself, you need something more than just what accounting gives you.
Josh LaSov (07:50):
And it's interesting, it's like, well, where is this data to me for FP&A for me, data starts at the ERP. So it's like where is this data? Am I using an ERP where most of my data is in the ERP or am I using an ERP where a little bit of data is my ERP and a lot of bit of my data is in systems around my ERP. There's no right or wrong answer to which approach you go to on the ends of these spectrums. A lot of it depends on the size of your business and the industry you're playing in. Software companies typically had do less in the ERP and more in the ancillary systems. Inventory companies do more on the ERP and less around the ancillary system. So it can just depend, but it is really important to kind of know where that data is and how it's recorded and what it looks like.
Host: Paul Barnhurst (08:40):
No, agreed. And I imagine Satori was something you ended up building because you saw it missing as you worked with clients, as you worked with data, you didn't feel like there was an easy way to build reports in Power bi. So maybe talk a little bit about why you started it and then maybe your key learning from that journey. I know you ultimately sold it to pe. That was what, about two years ago now?
Josh LaSov (09:05):
Four years? Yeah.
Host: Paul Barnhurst (09:06):
You're making me feel old. Thanks. So yeah, I'd love to hear a little bit kind of that journey.
Josh LaSov (09:12):
So the thesis there is basically everyone wants to report out of their ERP and every ERP tool has reporting features and functions, but they only take it so far. And a lot of times it's an afterthought. Most time it's an afterthought because it's all about how come I record transactions quicker and better, which is business process. So basically what we said is, Hey, this is not a knock on NetSuite. Every ERP falls into the same camp, but we had a footprint there. So let's start there. So hey, why don't we enable people to have all their NetSuite data in a prebuilt data warehouse with Power BI on top so they can have advanced reporting features and functionalities is that they desire, which was in Power bi, which could be financial and or operational reporting. Some people in this will probably say, well, you can't build a financial report in Power bi.
(10:06):
Yes, you're still going to use Excel. If you want an income statement, fine, you can, but it's going to take a lot longer. But a lot of the financial metrics and operational reporting, which is really what's most important, and the ERPs aren't as strong app yet, you're going to do that in a BI tool. But then there's the second piece of the pie of not only just building these beautiful reports, but then the second part of it is you don't have one data source. So it doesn't matter how strong the ERPs reporting tool is or isn't. You don't have one data source. You have tonnes of data sources. So what about your Salesforce data? What about your Shopify data? And the list goes on and on that you're not keeping in these ERP systems for good reason. It's not a data warehouse. So we were really solving two problems, giving people a centralised location to do reporting across the variety of their data sources with a prebuilt idea of a data warehouse and a BI tool because oh man, is that an expensive complicated tool?
(11:03):
And the second thing was these more complete reporting functionality everybody wants, everyone wants Power bi, everyone's already using it. Everyone has a Microsoft licence. So without this idea of a pre-built warehouse and pre-built instance of Power BI with reports and KPIs in it, what people had to do is they had to go call a co consulting firm, have a team internally try to figure out how to get data out of their data sources and build a data warehouse and what's the data model, but how do they know what the data model for something like a NetSuite is? Unless they talk to the functional folks and the functional folks can talk to the data folks, you're talking hundreds of thousands of dollars and you're talking a year plus to build a good data warehouse from these data sources. And most of those projects failed. Well, we could get 80 90% of your data in a data warehouse and in a prebuilt reporting tool in two weeks or less, and then the last 10 20% was your customizations. We'll work through that. Not hard. That's the easier part. Surprisingly, once you have that foundational semantic model or data model, it's a lot easier. So we were offering something that didn't exist and people pounced on it.
Host: Paul Barnhurst (12:10):
And anytime you can solve reporting problems and provide real value, people will want, it'd love to get your thoughts. I'm not sure if you noticed about they had Fab Con this week. Can you throw a total surprise question at you? It wasn't something we talked about before, but I'd love to get your take because you've implemented a lot of tools. You have the BI side, I'm sure you're familiar with Fabric and all they're doing. So they released this week Fabric Planning, which is just a usage component within Fabric, and it is a white label. They partnered with a planning tool similar to Power bi, works with Power BI and works on its own. And I'm curious, what's your thoughts of that in the marketplace, something like that, having dealt with all this because now you can have your warehouse there, there's some benefits with the way Fabric brings data together. How do you think that will play in the planning market? Yeah,
Josh LaSov (13:03):
I think it's really smart. I've said for years, why can't we combine BI and EPM tools? And I know the answer why on the EPM tool side, most of them's backbone is your chart of accounts. So it's only going to work at a trial balance level. Largely. They'll all advertise, they can do a lot more with operational data, but there's limits. There's limits when you have a lot of transaction level detail. Some of them have taken better steps forward to have a data warehouse component, like a plan full or solver BI 360. I've seen some cool stuff, but if somebody can pull it off to combine those two, that's what every CFO wants. I don't want to be in two tools. Where am I loading my budget? Which tool? How are they talking back and forth? So I think it's cool and it comes down to usability.
(13:53):
That's the other thing I always found interesting about the EPM tools. I've talked about BI tool usability in a second, but EPM tools, we would end up just building these beautiful reports and an Excel add-in Workday Adaptive as Office Connect, and they all have their own versions of that. But when it came to the planning, building, the models, the budget planning models, and those tools wasn't always as good as Excel. People would wind up in Excel. They're really using it as like, Hey, go fill in a budget so I can collect the information back to aggregate it. But a lot of times people are just like, give me an Excel spreadsheet to fill in. So it's kind of funny. So usability is big, but on the BI tool side, usability is tough. Those are complicated tools. Way more complicated. EPM tools because you've got to arguably not how to code a little bit. So if somebody can put these together and make it usable for a non-technical user, I think that's a huge win. And I think they'll try to buy with AI and they'll say, write a prompt. It'll write your formulas, it'll help you. I'm very skittish on that because when you talk about writing a formula using something like ai, I think it's very challenging when your data sources unique and when was it trained on that and how does it handle customizations in your data source and stuff like that. But hopefully they get there.
Host: Paul Barnhurst (15:17):
No, I appreciate that. I was just kind of curious your take, because I know it was a big announcement this week. I know the company that's doing the white labelling, the guy that runs their product team, and he was pretty excited to tell me about it. So fun to see as we continue to see the two kind of merge in different ways.
Josh LaSov (15:32):
How long has it been, I know you know this, but in the Power BI Space act out of Australia,
Host: Paul Barnhurst (15:39):
There's Aris power on moon mail or Info River AIM plan. Those are probably the four biggest. There's a few other smaller ones. Terrace just got bought by Vena.
Josh LaSov (15:49):
Yeah,
Host: Paul Barnhurst (15:50):
I did a guide on these last year. We wrote a guide and we covered also defacto Global is the other one we covered. We covered five tools in detail in the guide
Josh LaSov (15:58):
Smart blend of EPM and bi. Yeah,
Host: Paul Barnhurst (16:02):
I mean ideally everybody wants one source, sometimes easier said than done obviously because there's a reason we have a huge EPM market. If it was easy, they would all be combined. But let's move past that. We're going to talk a little bit about ai, everybody's favourite subject nowadays. I think you've heard of the term, if I'm not mistaken. So you even started your own company called Kasi ai. It's an AI agent. I know primarily for NetSuite gate. I know you could support other ERPs, but that's kind of your background experience. You started there. So tell us about the business. This is your moment to just tell us a little bit about why you started it and what you're doing. It's
Josh LaSov (16:39):
A wild world out there right now in ai, and there's lots of experts, if you will, but I can tell you what we know from our experience with our customers. We are revenue generating, right? This is real. This isn't some prompt engineering studio. But I'll tell you, let me start with why we got into it and where we are today. I think it tells the interesting story. So we got into it when AI first came out, November, 2022 I think is when Chachi BT was released. I was so interested in, can we explain data? Can we tell a story? Because when we were in our reporting tool days, we would build these beautiful reports, whether it's a formatting perspective or getting the right KPIs from the right data sources all joining and combined blah, blah, blah, right? But at the end of the day, I get the same question from FP&A , from A CFO, from a CEO.
(17:36):
Cool, great. You got me my report. Better, faster, prettier. What the heck do these numbers mean? Can you tell me that? And we would be like, no, I can't actually. It's not our job. No. And it was almost like a slap in the face, like we thought we were doing this magic of putting in a reporting tool, a BI tool. And this crazy joins all these modelling between these data sources to get you the numbers you need subjective KPIs in hindsight. But with ai, I said, wait, can we take the data and explain why the numbers are the way they are and why they're changing? And in 2022, the answer was no.
(18:21):
Fast forward. And we kind of followed this along and AI got better and better and better. And now the answer is an outstanding yes, we can explain the numbers better than humans. And that is a scary thought, but a cool thought. It's a productivity thought. It's not a competition, it's just letting people do something better than they historically could. But it doesn't stop there. Basically what we have is a platform where you can build agents. So it could be FP&A, could be the CFO, could be COO, could be staff accountant, controller, right? Whomever it is can build agents to multiply themselves. And yes, we do a lot of marketing and targeting to NetSuite because we have a big footprint there, but it can work with other ERPs and certain we pull in all types of other data sources because you need that for an agent to take action.
(19:17):
This is not a chatbot. Chatbots are highly inaccurate. Most of them, like MCP tools pull like a thousand rows of data. What make it a thousand rows of data? I need 10 million rows of data. So we've been able to build a platform that does three things. We started with our first bucket focused on analysis. Why did my numbers change? Tell me why my revenue changed. Drill into the transaction level detail. Is that 3 million rows of data? Is it 10 million rows of data? It doesn't matter to us. We work with big data and cannot do that from an LLM website, nor would I from a security perspective. So analysis is big revenue analysis, gross margin analysis, inventory optimization, churn analysis, whatever it is, depending on your industry, it's not replacing a reporting tool. You still need that, right? Maybe it's reporting for a board or compliance, but it's doing what's next.
(20:13):
It's doing the narratives. Why did the numbers change? What do I do about it? What are the risks? What are the next steps? The whole point of a report is taking that raw transaction line, big data down to something shrunk, summarised, aggregated into something a human can digest, sit down with, pull out the insights, the recommendations, the risks, the next steps, and put it in a presentation layer. We just do the whole freaking thing and one fell swoop. But humans can't look at the raw transaction level and we just go straight across with it. So when I tell you why your revenue versus marketing changed by item, I'm not looking at your top 10 items, your bottom 10 items, I'm looking at all 10,000 skews and I'm telling you everything. Is it a rate impact? Is it a price impact? Is it a mix? What is the mix?
Host: Paul Barnhurst (21:04):
And now, a brief message from our sponsor, Campfire. Build a best-in-class AI-native ERP designed for modern finance and accounting teams. With Campfire, you can automate revenue, streamline your close, and centralize accounting, all in one unified platform. And now they've hosted Finance Forward, a first-of-its-kind AI summit in San Francisco. They brought together the sharpest minds in finance and operations, all focused on how to actually use AI to move the needle. You don't even have to travel to learn the tactical skills to put your team ahead. You can stream the whole thing on demand, totally free, anytime after the event. So if you're ready to get tactical about AI and finance, head to campfire.ai and register for the on-demand content for free. That's Campfire, where finance teams go to think forward.
Josh LaSov (22:03):
So analysis has been a huge one where we're materially changing people's way of getting insights and answers quicker automated and better than they could ever do before. The second bucket is process automation. So it could be month end close. We can automate journals, we can automate reconciliations. You get to tailor this. It's not some rigid tool. You get to tailor this to how your business actually works. It could be reconciling data from the e-commerce tools and fixing orders and payments and your returns. It could be cash application. And there's a point for me saying all this could be cash application. So there's all types of business process automation that's happening as well. And the third bucket's forecasting, right? Getting back a little bit more at FP&A, we're creating revenue forecast, expense forecast, cashflow forecast. Again using that transaction level deco. So we have a customer there today who we built a revenue forecast for.
(22:59):
They got about 5,000 SKUs and they want to build a revenue forecast using the last two years of all transaction history, looking at inventory balances on hand too, and build a revenue forecast to say, of my 5,000 SKUs, what's the quantity and price that I should be selling going forward? Factors in seasonality, any assumptions the business might be making. And it built the forecast for them in a beautiful Excel workbook across all 5,000 SKUs who plans across all 5,000 SKUs. And then the next stage, it picks that up and turns into supply plan and breaks down those finished goods into components looking at inventory in hand. I can't do these things as humans at this level of granularity in any reasonable time or of any reasonable accuracy. So this platform is automating and improving. What we're doing today is where we start analysis business process and forecasting all things FP&A needs to own.
Host: Paul Barnhurst (23:54):
Well give me a couple thoughts, but first one I want to go to is what is it about agents that really make it a core feature of FP&A? Everybody hears about agents, we all know ai, especially if it's probabilistic being generated, can make hallucinations or errors. And so how do you think about all that? What would you say to FP&A professionals as they keep hearing all this talk about agents? Let's kind of bring it back to the average FP&A person. Yeah,
Josh LaSov (24:20):
So I want to start with getting over the hallucination error thing. 99.9% of these tools out there are prompt chatbot. You're going to have hallucinations and errors and it just makes sense because what's happening is you are saying to an agent, analyse my gross margin and why it changed this month. And that agent is now going back to the data it's connected to and saying, okay, I go look at gross margin. What the heck is gross margin revenue cogs? What do I include? Exclude for revenue? Am I looking at sales orders? Am I looking at cash sales? Am I looking at anything that just hit a certain which GL accounts of revenue? Is it posting or not? Do I include journal entries? These chat bots have no idea. So people are getting all types of hallucinations and errors because of that approach. And then these chat bots are like, well, I can only pull back a thousand rows of data, so I'm going to summarise stuff.
(25:23):
So I can't even tell you the why, even if I could get the metrics side of it accurate. So they're setting themselves up for failure. And lemme say this, we did that for way too long in our business and we had to pivot. And the chat bot, a working chat bot that can do those things is the A GI of FP&A, we're not there yet. AI is not here. I don't know when it's going to be here. I don't know if it will be hearing time soon. It hasn't improved in my opinion, even 1% since November, 2022. So we pivoted. So what we do differently is we pre-organized the data before AI ever runs through it with questions. So it's accurate. That agent has the data. So if I'm saying why did my growers margin change? You already know the data is right, inclusive and accurate, and at that transaction level detail, you're not going back after the chat bot gives you results.
(26:16):
You're like, how the heck did you get these numbers? No, no, no. You already know the data's right. That is what you have to do to not have hallucinations and errors. You can't do that in an LM website. And Claude, I loaded 30 megabyte file, that's not enough. So I need to be able to do this where I can have gigabytes of data or 800 megabytes of data at a time kind of stuff. So when I think about agents with FP&A, it's a couple things, and this is not necessarily unique to FP&A, but I think fp a has a unique opportunity to capitalise on this because of the skill sets. And first and foremost, it's about multiplying yourself. If you're working 40 hours a week, which you should be, right? That's probably what you're contracted to do. What if you could produce 50 or 60 hours of productivity in a week?
(27:04):
But what if then you do that not in 40 hours, but in 30 hours, no one ever got in trouble for being of value to the business or providing additional value to the business. So if you're an FP&A and the CFO says, why do my revenue change? And you say, give me two days, I'll be back or give me a week, I'll be back. Or if you could say, give me five minutes and I'm going to tell it to you at a level that you've never seen before, wouldn't you want to try it? So the FP&A can deploy agents for everything they're doing today and then some, and they're going to have more time to spend with the results and the next steps. So they're going to be more productive and more intelligent in what they were already doing. And now with the time back and focus on other things, execution, big picture, we talked about the integration of finance and strategy and that becomes real. And why I'm saying it's not just for FP&A, it's for everybody. We all have to remain competitive with every new technology improvement. Salespeople went from a Rolodex to A CRM, they're more productive, they're more intelligent. I don't think this is necessarily any different.
Host: Paul Barnhurst (28:17):
Yeah, no, definitely a productivity tool. So I'm curious, how should FP&A professionals be using AI today? So let's ignore any particular tool per se, but just in general, where should they focus on to be adding value? How should they be thinking about things? And now, a brief message from our sponsor. Is your AI thinking like a generalist or like a finance leader? Here's the thing: most large language models weren't built for finance. They're trained on everything from Reddit threads to recipe blogs. So when you plug them into your accounting workflows, the results can be, let's just say, not balance sheet accurate. That's why Campfire, the company pioneering AI-native ERP, just made a huge move. They've raised $65 million in Series B funding, co-led by Excel and Ribbit, bringing their total to $100 million raised in just 12 weeks. And they're putting that momentum to work with LAM, the large accounting model. It's a proprietary AI trained specifically for finance and accounting, already hitting 95%+ accuracy on accounting tasks. That's redefining what ERP can do for modern finance teams. Unicorn companies like PostHog and Ripple are already using Campfire to automate their finance operations and seeing real results. You can see it for yourself, sign up for a live demo of LAM at campfire.ai. That's campfire.ai.
Josh LaSov (29:58):
I think step one is shifting from their traditional reporting. Don't stop doing their traditional reporting, but I think they need to add in the automation of the analysis side step one, right? So again, that raw data to the agents to be able to tell you why the numbers are changing, it'll do a faster and better job than humans. I know that's so hard to believe. I know people are probably listening saying they're full of smoke, our CFOs are bouncing off the walls over this stuff. I mean, we show them one agent to analyse their business and their hook. This isn't smoke and mirrors anymore, you just have to be on the platform that actually can do it and pull it off. But I think that analysis bucket is a huge one. And you're doing all types of analysis today. Start with what you're doing, automate and improvement, getting it down to that raw detailed level of the data.
(30:48):
Nothing rolled up. I think step two is forecasting and AI cannot replace your models, right? You're better at building financial models than AI is, but it's going to get you a large portion of that model started or done and save you a tonne of time. So you can still run scenarios, you can still build in all your assumptions, all that's still possible. And to me it's like if I can get you 80, 90% away there and then you can finish it up with all of the Excel wizardry that we all love, go for it. But it's going to do that in ity split. I'll tell you a story. We have a CFO, and by the way, all the F FDA folks probably are looking to be a CFO one day. So we have a CFO. She went do a couple things and she's like, I need to build a revenue plan for 2026 down to the SKU level, down to the sales channel level, and I need to do that and I can't do it down to the SKU level, but I can do it down to the sales channel level.
(31:43):
It's going to take me like a week. I said, give me a couple hours. So we built them a forecast. We had an agent build them a forecast down to the SKU and the sales channel level, which they couldn't do historically. And she goes, this is the start of my planning cycle. You just saved me weeks of time and have more granularity. And that was it. So that started their plan. We had another CFO who built the forecast using an agent and then said, I'm going to triangulate this against my own FP&A team and my sales team's forecast to see if they're sandbagging it. So there's many use cases for this stuff, but I think if I'm an FP&A, what am I doing today? Let me improve and automate it and then we can worry about the future stuff. We don't need to land on Mars, but we are driving to work every day. So let's just improve driving to work.
Host: Paul Barnhurst (32:29):
I want to go to Mars.
Josh LaSov (32:30):
If it's one way, it could be tricky.
Host: Paul Barnhurst (32:33):
Yeah, I'd like a return trip. Not going to lie.
Josh LaSov (32:36):
So
Host: Paul Barnhurst (32:37):
I'll wait. So if you could give a piece of advice to FP&A professionals what they should be doing today to future proof their career, what would that be?
Josh LaSov (32:46):
Well, I think it's to embrace new technology. No matter what it is, use it as an accelerator, not be scared of change management. A lot of the reasons companies don't move forward is a change management perspective. It's like people have, I do things my way. I don't want to change. It's with any tool, any implementation, you have to deal with that. But this is new, right? This isn't like, oh, we know we need to implement A CRM or an ERP. We've gone in prior businesses. This is like, I never implemented this. So I think FP&A is a unique position to embrace change. Other positions are not as reluctant to embrace change, but FP&A is forward thinking, forward looking. So I think if FP&A wants to accelerate their career, accelerate their value, they need to do it. And if they don't, I do worry like a controller can now build agents to do analysis, but isn't that what a fan you're supposed to do?
(33:39):
So I think to really bulletproof your career, it's embracing technology that's going to help you do your job faster and better. And when I say better, it's truly add a more detailed level to get more insights and intelligence from the business. I also think FP&A is a unique ability to take more on, I know we don't want to do journal entry creation or bank racks, but set up an agent to do it for yourself. And we have fp a folks doing that. So we have FP&A folks who are like, we do allocations every month, and I hand it off to the controller who takes my spreadsheets and builds the journals and pops 'em back to the ERP. And we have FPA folks who are like, I do that now. I don't do anything really. I just told 'em my allocation methodology and it runs through and it does it really nicely and then it creates the journals for me.
(34:26):
So I think there's more handholding or hand helping that FPD can do for the business. Don't just focus on analysis and forecasting, do those things and do them better, but there's more that we can do to help the team and no one ever got in trouble for being proactive as A CEO. I hate nothing more. I should know, hate the wrong word, dislike nothing more than when I asked someone to do something and they put something on my desk that I have to go do more work for if they can get it done, great. And it's not adding more hours to your workload each week with these agents, build it, set it, forget it largely, and the value you're providing is increasing your value.
Host: Paul Barnhurst (35:06):
That makes sense. If you were a CFO and you had a small company, you had freedom, how would you think about implementing an AI tech stack? Not necessarily tools, but how would you go about it?
Josh LaSov (35:18):
I think the word tech stack is getting minimised with ai. And here's where, I mean, historically a business and A CFO, and I'm guilty of this. I've had it and I've sold it and I've implemented it, but historically, A CFO said, okay, I need a different tool for a different function. Some industries will have more, some industries will have less because their ERP will do more and less. Some size of companies there's a spectrum, but I need a month end closed tool. I need an AP automation tool, I need a reporting tool, I need a BI tool, I need all these other tools. I might need a bank rec tool. Look, the list goes on and on. I need a rev rec tool, I need an invoicing tool. So I need all these tools that sit around the ERP again, more or less depending on your company size and industry, which ERP you're using, and why.
(36:11):
But what's so interesting is our customers, what they're saying to us is they're saying like, wait a second, I deployed agents to automate my fixed assets. I don't need a fixed assets tool. I deployed agents to capture my expense reports. I don't necessarily need an expense report tool. I deployed agents to build my forecast, do my analysis. I don't need an EPM tool. I am not telling you that we do as good of a job as those tools. Not yet we're going to skate there or month end close tool. We're skating there. So when I think of a tech stack that our agent platform or other agent platforms that are being smart are encroaching and encroaching and encroaching on the tech stack and making CFOs say, wait a minute, if I put in one platform that can do all this different functionality, so I'll have to manage all these different vendors and platforms and integrations, it's compelling, right? It becomes very compelling. So if I'm a CFO, what I'm going to look for is an agentic platform that could start to do these other functions. I'm going to lean into it. Again, I'm not suggesting you still don't need bill.com you or something like that, but how much of those things you need is getting smaller with these agentic platforms?
Host: Paul Barnhurst (37:30):
Got it. So it sounds like to you that it's really okay, we got to have my core ERP, I mean your agent platform's not going to do that. You still got to have all that kind of data storage, but from there, it's really how much can I do with an ag agentic platform versus best of breed approach?
Josh LaSov (37:49):
That's right. And if you think about this too, the best of breeds are all going to add chat bots and some agent functionality and they'll make their tools better. But I think if I was a best of breed tool, I would go try to find how I can get out of my swim lane and go bolt on a series of other tools. Like if I'm an AP automation tool, can I bolt on an EPM tool? Can I bolt on a closed management tool? And so forth and so on. I think, and I'm talking to other founders in the same space as us and we're all thinking, we're all seeing the same thing. This isn't like our thinking. The only way I get thinking is from customers. So literally I have a CFO who's like, I don't know if I'm going to renew my Workday adaptive contract.
Host: Paul Barnhurst (38:32):
From what you're saying, I think what you're going to see from the tools in general, whether they're an EPM tool or they're an whatever they are today, and you've been seeing it happening more and more of many of these tools take the platform approach where, hey, we're trying to be a platform. AI makes that easier. I think you can get a better platform than you used to with a tonne of Boltons with the Gentech ai. Does it get you all the way there? Is it going to be the best, probably not five years from now, we'll revisit it 18 months from now, we'll revisit it every six months, we'll revisit that. But today it's not, I still think you're going to get more functionality and more features from the individual tools. Now the question is, is that little extra functionality worth the additional cost? And that's where we're going to see more and more analysis being done and saying, are there benefits of having one agent with more data, more things?
(39:28):
Is there ways we can implement that across the rest of the businesses? Because one of the biggest things with agents, and one of the biggest concerns we have right now is businesses are siloing themselves with agents, much like we did with SaaS. And to get the best out of data, to get the best out of ai, to get the best out of our tools, you have to minimise that silo approach. Just like to get best out of your employees, you have to minimise the silo approach. It's no different. And so I think that's the challenge we're going to see. And I think the real issue isn't even at the CFO level, right? It's at the CEO in that entire C-suite to make sure do we have the proper strategy so that AI can really help us make better decisions, not just be more efficient. Efficiency is going to happen no matter what. If you don't get efficiency out of AI in the next couple of years, there's a problem. Now you can question how much that is versus your ROI and is it worth it? Water struggling today, that's a whole separate, I'm not going to talk the economics, but you can get value out of it. You can get efficiency out of it. The question is how do you make better decisions because of it?
Josh LaSov (40:32):
Couldn't agree with you more, Paul. I mean, and no one has a crystal ball, so it's hard to predict the future.
Host: Paul Barnhurst (40:38):
You don't, lemme go get mine.
Josh LaSov (40:40):
Yeah, I carry it. I carry it with me at all times. It's changing quickly. It's all changing quickly, but I'm with you. I think the more that we can go back to one platform to do more thing or a couple platforms to do one thing. I am not a spectrum guy where I have this big qualm with the battles of the internet right now of use this spectrum of the ERP or this spectrum of the ERP of the legacy tools have more of a suite approach where it's like all this function I built in and the other side is like, no, just a GL and get all the point, best point solutions. It's like most companies head up in the middle and they should, but again, depends on your industry and company size. It's kind of the same thing. It's like I'm not sitting here saying one agent platform is going to replace all your other tools. I think it's encroaching and minimising the value to your point. And some people look at that from a cost savings perspective, but most people kind of see the value of like, wow, I can do a lot more. And everyone's going to land on a different spot on that as this continues to evolve and solves those requirements specific to their business and their industry and their size and stuff like that. Makes
Host: Paul Barnhurst (41:47):
A lot of sense. I think we have a similar view. Alright, we're going to move into our, this is what I call a more standardised section, our fp a section where I ask some similar questions. So looking for just some quick short answers here to get your first thoughts. So what is the number one technical guild that fp a professionals should master
Josh LaSov (42:06):
Understanding table data from their source systems?
Host: Paul Barnhurst (42:11):
So kind of basically understanding data. Really understanding the data. I've heard that before. I think the way you said it is a little different than others have explained it. I understand why you said the table and from your background it, I've heard different versions of that answer before. What about soft skills
Josh LaSov (42:26):
Communication and emotional intelligence? I
Host: Paul Barnhurst (42:29):
Had a feeling emotional intelligence would be in there. Surprising. Alright, so I know you've used Excel a lot over your career, obviously working in FP&A. I mean I think everybody is a fan of a spreadsheet. So if Excel removed one feature tomorrow, if one went completely away, which one would cause you the most panic
Josh LaSov (42:48):
Format painter? The shortcut code. Is it alt HFP, is that right?
Host: Paul Barnhurst (42:54):
I think so. Not like that.
Josh LaSov (42:56):
Why isn't it alt, h, FP&A. I mean I think I use that more than anything else. I mean there's many, but that's one of my favourite tools.
Host: Paul Barnhurst (43:05):
See, I was just going to say with an agent now you can just tell it to do your formatting for you.
Josh LaSov (43:10):
Probably. I mean our agent produce these crazy beautiful Excel workbooks. It's wild. But again, I'm not going to sit here and tell you it's perfect. You still might want to tweak it.
Host: Paul Barnhurst (43:19):
Yeah, totally kidding. So that's your one. I get different ones every time. So it's fun to see what people say. Don't think I've got that one before.
Josh LaSov (43:26):
What's yours?
Host: Paul Barnhurst (43:27):
Excel. Remove one feature tomorrow. Power Query.
Josh LaSov (43:31):
Okay. I feel
Host: Paul Barnhurst (43:32):
Like I'd go back five years.
Josh LaSov (43:35):
Preacher Fire. But that's interesting and that's why I went back to data tables and the prior answer is because for me, power Query is about building that data model. That's where Power Query comes in. Power Query is a game changer. Oh, that's so cool.
Host: Paul Barnhurst (43:47):
I can't wait till the day. And I know Microsoft from people I've talked to struggling to figure it out. I want that agent that I can talk with sitting in Power Query. That's why they put Power Query on the web. It'll be much easier to add it on the web than desktop. And
Josh LaSov (43:59):
It's interesting too, it's like with Power Query, do I still need data models for building reports? I do for analysis, not really. The agent can just look at table data and you can say, here's the join, use this as a common denominator. But I love Power Query. It is a powerful tool. I'll
Host: Paul Barnhurst (44:17):
Give this example, and I've given this many times somebody, I was talking to a group and they're like, well, I have this process every month. It's kind of the same. Should I just give it to ai? And I was like, can you easily do it with Power Query? Yeah, then no, I wouldn't create that. You just need a deterministic rule. It's like if you can do it simply in Power Query, why figure out how to build it with an agent.
Josh LaSov (44:42):
Automate and improve what you're doing the easiest way for yourself. I
Host: Paul Barnhurst (44:45):
Talked to a guy this week who it was funny. He goes, we just streamlined our hire bor deck process and we were talking about Claude and getting ready to do an interview of how he's using Claude and Finance. He goes, I'll be honest, most of the benefit came from Power Query, not Claude. And I was kind of like, okay, you're preaching my language in that yes, AI is great, but let's start with the data and the foundation and the tools we have that are built to solve things already versus trying to find a whole new solution to a problem that's already been solved.
Josh LaSov (45:12):
Totally. And that goes back to our conversation earlier too. It's like chat bots are not primetime and that's why we start with the data and heck, a lot of our customers', data starts in sql, starts in Blob, starts in S3, it starts in a database which was generated from PowerPoint. So anyway, I'm with you. AI is awesome. If you get the data right first, the data needs, you need to know your data, have validated, signed off on it, then use ai, not use AI to go grab the data. I just haven't seen that work enough.
Host: Paul Barnhurst (45:46):
Yeah, I think we're on the same page there. Alright, so now we're going to move into the Get to Know You section.
Josh LaSov (45:51):
Okay. Lemme
Host: Paul Barnhurst (45:52):
Ask this one first. I'm going to switch my order up a little bit. If you had to pick a movie or theme song to best represent your life, what you going with and why?
Josh LaSov (46:02):
Can I ask these questions back? I react back to you so you get the answer to.
Host: Paul Barnhurst (46:05):
I'll give it a try. I'll have to think about it for a minute, but sure. I already got one in mind. You can do it.
Josh LaSov (46:10):
I couldn't think of a movie or a theme song, so I'll tell a short but funny story. We went into name Causey, we wanted to name the business Grateful Data. We like the Grateful Dead, so we thought Grateful Beta would be a cool company, but some guy owns the domain name and wanted $14,000 for it.
Host: Paul Barnhurst (46:31):
I like it. All right, well I'll answer that question, so please. One, I'm a fan of Weird Al Yankovich. I have a weird sense of humour. Anyone who knows me knows I'm a little sarcastic. I can be snarky. It's been hard for my daughter to figure out my humour and so I'm going to go with one of weird Al's most popular songs is my theme song. White and Nerdy, I mean, come on. I attend the Excel modelling World Cup. I have red hair and I burn Easy. I'm pretty much white and nerdy. There's my theme song for my life.
Josh LaSov (47:00):
Perfect. Have you heard of a band called Big Data?
Host: Paul Barnhurst (47:04):
I have not.
Josh LaSov (47:05):
Okay. Give them a listen. It's kind of like a, I don't know what you call it, but it's cool. I mean it's a little bit, I don't want to say it's house music. You would never hear it like a club, but it's kind. I could see jam into it. It's kind of fun. Check it out. Big data.
Host: Paul Barnhurst (47:20):
All right, well I'll have to go on YouTube and search for big data. There you go, everybody. But make sure you put big data, music band, not just big data because you'll get a whole different set of answers from the algorithm. Alright, what do you like to do in your free time when you're not building a business? What do you do?
Josh LaSov (47:37):
Is there life outside of that?
Host: Paul Barnhurst (47:40):
You have a two and a 4-year-old? No, there's not family and business kids
Josh LaSov (47:45):
Work. I enjoy being outside. I live in Los Angeles, so anytime I can be outside enjoying the elements, if you will, that is a win. I like to exercise, so if I can combine those two, perfect. But yeah, mostly with my kids these days, which is awesome.
Host: Paul Barnhurst (48:02):
Got it. All right. Last one here. If you had to pick a new profession tomorrow, what would you be doing and why?
Josh LaSov (48:11):
I've thought about that a long time. When we sold the last business, I did take a little time off. It's hard to turn it off. So when the right idea presented itself, I did jump back in, but I thought about it even since then. And I know what I'll do after this and I want to do something more altruistic and I want to get back to young founders, ideally ones that are bootstrapping, not doing the VC route, and I want to just offer time and resources. I'm not saying I'm an expert, I'm not saying I'm the best, but if they can use me to triangulate decisions across of other experts they talk to, I want to do it and I want to give people those, I had a mentor or two that really shaped me and my career and I just don't think people have access to that on a free basis. Well,
Host: Paul Barnhurst (48:59):
I'm not young, but if you start that, let me know. I'll come to you for advice. I am good
Josh LaSov (49:04):
To you for advice.
Host: Paul Barnhurst (49:07):
Right. Well thank you so much for carving some time out today. Josh, I enjoyed chatting. Good luck building. Cosy. Last question. What's the best way, if someone from our audience wants to reach out, maybe get in touch, what's the best way for them to do that?
Josh LaSov (49:20):
You can go to our website, czi.ai, CAU ZZ y.ai, and there's a little bottom, I think it's like request demo or tryout. Just fill out the form and it'll get to us, or you can ping me on LinkedIn or we're pretty casual. So one of those ways is great and we will reply very quickly.
Host: Paul Barnhurst (49:40):
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 credits for the FPAC certification, complete the quiz and reach out to me directly. Thanks for listening. I'm Paul Barnhurst, the FP&A guy, and I'll see you next time.