Christine Butchko:

Good afternoon to everyone on the East Coast, and good morning to all of the folks dialing in from the West Coast. Welcome to today's session, which should be a good one, about preparing the back office for AI, a readiness checklist for CFOs, heads of finance, and controller. So without further ado, let's go ahead and kick it off. So just to start, we have some standard housekeeping. First things first, you're all muted you're all muted during the duration of the webinar. Um, if you do have any questions, you can ask them in the chat. I'll be moderating that or in the q and a section. It should be on the right of your screen, um, next to, uh, some resources and all of that good stuff. And then you will receive a copy of the webinar about two hours after it's concluded. So rest assured if you have to run off, um, or, you know, you think the content's so great, you wanna share it with a colleague, uh, you'll have access to that. So and finally, any questions that we're not able to get to, um, you can always email me, christine@hubify.com or, uh, Ben and Jason, but I'm sure they'll they'll give their contact details after the fact. So on the agenda, uh, introduction and housekeeping, we're here right now. We'll talk through the four core processes and systems that your team needs to have in place, how to get clean, consistent ARR, MRR, and then what to look for in a vendor to bring to the back office, and then finally, q and a. And without further ado, let me bring on the main attractions, both Ben and Jason. So just to start, uh, Ben, could you introduce yourself to the crowd? 

‍

Ben Murray, the SaaS CFO

Hey. Great. It's everyone. Great to be here. Yeah. It's morning here in Scottsdale still. Uh, but, again, my name is Ben Murray. I'm a SaaS CEO of Hawaii Trade. Got started out in the airline industry, got into software in the early two thousands, came up the ranks of FP and A, uh, to CFO and founder owned and PE backed SaaS companies. And then today, uh, do fractional stuff, have academies, newsletter podcast, so live and breathe, uh, b to b SaaS finance. Great to be here, guys. 

‍

Christine Butchko

Excellent. And, yeah, in the chat, if people wanna share where they're from and what the temperature is, it's currently 20 degrees in New Jersey, which I'm not super thrilled about. Jason, why don't you introduce yourself to the group? 

‍

Jason Berwanger

Yeah. Sounds great. And and for those that haven't checked it out, Ben has an incredible academy on, like, a lot of finance governance, uh, how to calculate MRR, ARR, and just a ton of, uh, fun know how. So if you haven't checked out his academy, definitely recommend it from my seat, and it's some of the most accurate content I've seen from most folks out there on LinkedIn. But, But, yeah, my my name is Jason. Uh, I am a cofounder and CEO here at Hubify, and my background comes from a fractional CFO, been an accountant, and done a ton of ERP implementations, and really did a lot of, like, accounting automation work, uh, on behalf of the office of the CFO and liked it so much and solved so many problems and decided to start Hublify to, you know, focus on the high volume order of the cash accounting automation space. So that's that's me, and that's my background. 

‍

Christine Butchko

Excellent. Well, with that, uh, uh, I'll leave it to you too to to talk through bringing AI to the back office. 

‍

Jason Berwanger

Yeah. Sounds great. Um, so I think, uh, you know, Ben, where we'll start, we I mean, you and I talked about this, but the context for the audience is we we saw a survey where there were 1,750, like, finance and tech leaders talking about how they were leveraging AI, um, and particularly around, like, financial modeling and accounting and finance. And, you know, surprisingly, uh, I think there was a very limited number of folks that were actually using that as a part of their core day to day process. And so the question we get a lot from folks is, like, are are we behind, you know, implementing AI because we're not fully leveraging it in our day to day workflows. And maybe we'll start there with, like, what you're hearing, what you're seeing out there in the market, and then I'll I'll kinda share what I'm hearing and seeing from customers as well and and give some perspective. 

‍

Ben Murray, the SaaS CFO

Yeah. Definitely. And, again, thanks for joining everyone. Uh, Yeah. You know, there's so much You see all those reports out there, Jason, that say, oh, man. You know, 80% of companies or CFOs have adopted AI, and and I think that's completely false because when I actually, when I go to events and SaaS dinners with CFOs and I ask, hey. Raise your hand if you're using AI. You know, consistently in your process, and very few people raise their hand. And, yeah, it's easy to do that one off ad hoc analysis or write something, uh, but what I see anecdotally is is there's, yeah, very little consist like, in our integrated into our process use of AI. 

‍

Jason Berwanger

Yeah. That makes sense. And I I hear something similar, and, though, I think, though, the reality check that often the conversation evolves into with some of the the CFOs and a lot of folks on on the at our front lines between FB and A and accounting, the data point is the there's a lack of context already for the accounting team and the FP and A team. And so what we're supposing is what what makes what makes folks think that they're gonna be able to leverage AI to use something to solve FP and A and accounting if the existing practitioners don't have enough context for the business data today to then, you know, use to apply. Right? And which said more plainly, you know, even if you have the absolute best FP and A analyst, if they have no context or you have the the best, you know, revenue accounting manager, they have no context, then you know? And and they're basically using composite or grand totals of the month and kinda doing a high level reconciliation, uh, or that data is really manual and it's in PDFs and it's not a structured quote to cash process. Well, there's a limit to what they can do, and there's also a limit to what AI can do because of that lack of context and, you know, connectivity, uh, to to the business. And I'd I'd be curious if if that resonates at all with with what you've heard. 

‍

Ben Murray, the SaaS CFO

Yeah. I I I like that analogy because if we're struggling with FP and A and producing great variance analysis, great forecasting, uh, SaaS metrics, or whatever your industry is, uh, yeah, then we're not ready for AI because it needs just like an FP and analyst, we need clean data, structured data, detailed data to do a great job. And if it's that that's there, there's not there's not gonna be anything magical with AI that can do it any better. 

‍

Jason Berwanger

Yeah. Yeah. That makes a lot of sense. Um, you know, I I think from an audience perspective, the way the way that we think about this with customers that have had some success here is is a couple layers. Right? The first layer ends up being this deterministic automation layer, which is connecting to your existing customer financial systems, like your CRMs and your invoicing and your payment platforms, then applying, like, reliable, deterministic automation that basically is replacing the manual reconciliations and the calcs of, you know, the rev rec or an invoicing sub ledger and AR validation, etcetera. Uh, and then once that's done, there I think there are folks that are are doing a great job with MCP and, uh, and, you know, connecting some of those datasets with Cursor. We actually see customers like Cursor, uh, have connected some of their data to their, you know, their their internal AI processes, and they're able to glean a lot of, uh, you know, new insights. But it's all based on that foundation of solve the fully contextualized customer financial data or the operational financial data, like expenses and payroll. You have to solve that first. And if you solve that first, well, then the fun, you know, sexy strategic part at the end, which is the analytics and or the AI application, which are more probabilistic type of solutions, you know, if you apply those to deterministic accounting data, you're in a great place. And I I feel like there's a parallel there, Ben, in what I've heard you talk about this for years, and I've always appreciated and respected this approach of, you know, ARR and MRR is very similar, where if you're basing MRR and ARR off of, like, a CRM calculation without as of date reporting, and you're not basing that off of the actual earned revenue and the p and l's and the earned revenue per customer, well, it's a very similar metaphor in the sense that you're you're not going to be able to answer or tie back or trust the data enough to use it. And it feels like a very similar analogy from my seat in terms of what you've always cautioned folks of having ARR, MRR being separate from the finance and accounting data itself. 

‍

Ben Murray, the SaaS CFO

Yeah. Maybe we could talk about that for hours. But you're right. Yeah. It's gotta be transactional. Like, we have to have a bill. We have to have history from day zero ideally. And so, yes, there's, yeah, a bookings way, the P and L way, lots of different ways to do it, but there's a certain data structure we need in both. 

‍

Jason Berwanger

Yeah. Yeah. So that that first principles approach for MR, ARR is the same first principles approach for getting AI in your modeling, in other words. Yep. Cool. Um, so I think pivoting now, so we'll we'll talk about implementing AI, and, um, I'll give a I'll give a perspective because I think there's a ton of promises and great marketing out there in terms of how AI is gonna change the world, particularly in accounting and finance. And, um, you know, I I think the, you know, the the one piece of advice that we have is is you really wanna think about, um, how to validate some of those problem statements that you're going into and then apply them to some of the AI vendors. And so one way I've seen customers successfully do this is think about a very specific use case, like a, I wanna see revenue or, like, a segmented p and l by call centers and operating systems. And, you know, as you're pressure testing that with some of the modeling, if is that data actually available, and is it contextualized for the AI models? And, you know, look at that in a POC with your AI vendors and some of these folks. And, you know, I think there's a ton of promises out there. And, unfortunately, because of there's been so much marketing that it's all kinda commingled with folks that can't help and maybe can't help, and they all sound and look the same. Um, but I think pressure testing a use case, uh, and then working backwards with does does the AI actually have enough context to be able to solve this? Because could you do this manually in a POC even? If the answer is no, then it's very likely vaporware. Um, but having that, like, proof of concept where, like, you can prove out a real use case and think you know there's value, uh, that feels to me like the at least part of the antidote to falling for vaporware with some of the, hey. AI is just gonna replace our whole FP and A team and accounting team tomorrow. 

‍

Ben Murray, the SaaS CFO

Definitely. And I I like that, you know, because, yeah, can you do this systematically in, say, a manual process, and it's very orally step by step, uh, to then put a AI layer over that or automation layer. Kinda what you're saying, Jason, gets me back to thinking about, you know, when companies managing their performance and how they're looking at performance and data as a CFO and for all the CFOs controllers out there. It's well, one question, especially as a fractional CFO, it's like, alright. How how does your board how do how does your c suite want to what do they look at? What data did they need to manage the business? And if you're public or you have third parties, you know, what what are they asking for you? And then that kinda dictates me to me, okay. This is the data structure I need to set up within my company, within my chart of accounts, within my metadata, within all those data sources. You know, it's just so foundational. And then, okay, automation and AI on top of that. 

‍

Jason Berwanger

Yeah. Exactly. Yeah. Makes a ton of sense. Um, well, I think let let's pivot and, um, you know, let's talk about AI readiness. And, um, um, you mentioned board and working backwards. I think that's a great way to think about it. But maybe walk us through, you know, how you think about, um, you know, how to bring in AI, what are the prerequisites before, maybe what are the mistakes or pitfalls to to reach it, uh, to look out for if you're looking to assess both your readiness and how to apply AI to your your workflows for your finance team? 

‍

Ben Murray, the SaaS CFO

Yeah. Definitely. I think one you know, the tech stack is growing within the back office for controllers and CFO. So, alright, what tech do we have in place? How does that data flow? And we'll get to some good slides here just thinking about this process. But it's you know, for me, it's you know, when I work with SaaS companies, AI companies, the four key SaaS data sources, One is HR data, contractor data. Where is that coming? How is it flowing? How is that flowing to the p and l? Uh, two is bookings data. Do we have that sales led process where we close one opportunities? Do we use Salesforce, HubSpot, or some other CRM system? How is that data tracked? How does that flow to the p and l in our quote to cash process? Uh, you know, then it's p of course, financial data. How is that structured? You know? And that's one of, like, always talking to founders and CFOs about how your chart of account's structured because that'll be key for AI. Uh, and then finally, customer slash revenue data. Where are we invoicing? Is that out of Stripe? Is that out of this system, that system? And that's where things get really disorganized, Jason, what I see. And then I get that request, uh, in due diligence. Uh, alright. I need an MRR schedule. I need revenue by customer by month. I need it segmented five different ways, and we can't produce it. You know? So there's that foundation that we have to put in place. And if you're like, yep. We've got that process. We've got from our schedule. We've got all this in place. It's a nice flow. We close in five days or less. You know, then it's like, hey. Are, you know, are we a candidate for for AI? 

‍

Jason Berwanger

Yeah. And and I think what you're highlighting too is that's almost the prerequisite for AI readiness from a financial analytic perspective. Mhmm. Right? Where if you, yeah, if if you have that atomic data set of customer and financial data expenses in a way that you could analyze that's got some of the segments that you need, you you've met some of that criteria. Yep. Now, conversely, I think there's there's likely some operational, uh, applications, like financial operational applications. We've seen some cool stuff with, um, almost like an OCR two point o where, uh, you you, from a finance ops perspective, can, uh, you know, glean information from a PDF and pull that down in a more structured format or flagging anomalies and things like that in your AP process. But, you know, I think, disproportionately, you and I are focused on the, you know and and the the fun part, which is, like, how do we enable FP and A and strategic finance? And, you know, really, it's yeah. Unironically, you mentioned it starts with the chart of accounts just like everything else, which is if you get the chart of accounts right and you get the segments right, well, even if you don't have all of the atomic data, at the very least, you can still trend and analyze your P and L over time. You may not have all of the data you need for AI because it may not, like, trace back from a Lydian general provenance perspective back to the, uh, you know, the customer level. But, um, um, you know and and I think the most common example, which is part of what's on the slide too, is is, like, uh, assuming that the ERP is going to, you know, solve all your problems by dumping that data in. And, you know, from from my perspective, I think it depends. Right? If if your business model is you're using your ERP to manage your contracts and and entering your orders and you're doing your billing in your ERP and your cash application, uh, and you've got this, like, ERP, almost like SLG sales process like you mentioned, Ben. A lot of that detail will live in the ERP, and it's probably well served because you're reporting everything that's tied into it. Most business models, though, nowadays, though, I think are, hey. Stripe has all of our payment processing data. Salesforce has our CRM data. And, you know, the the ERP is really a general ledger. And, um, just by throwing all the data into this very rigid ERP data model isn't actually gonna make you ready for AI. Uh, even if there is a NetSuite MCP, that doesn't actually help you solve the problem is what I that's an example of what I think you're getting at. 

‍

Ben Murray, the SaaS CFO

Yeah. I mean, I see it all the time, say, with with companies that use Stripe, and then then we just you know, there's some batch entry to the p and l uh, for revenue, and it's just so hard to decipher. It's so hard to reconcile, uh, that and and, you know, employees and contractors and how they're getting paid, and it's just yeah. And we have no visibility in the p and l. Uh, one, we can do better. Right? At least let's get those everything to the right spot in the general ledger. And then, you know, we've gotta work on that underlying data structure because that's what we really need to operate at a high level. 

‍

Jason Berwanger

Yeah. So it's really chart of accounts designs the overall structure. Getting it into the chart of accounts, even if it's a high level, is a good next step, but doesn't get you anything special because there's still a little bit summer level data, but it's still better than not having a well designed chart of accounts. The last bit is now all of the details are at least available on the backup of what makes up the general ledger. That's really where the fun ability to, you know, do slicing and dicing comes in. Because if you wanna slice and dice, it's probably by a bunch of segments that aren't in your chart of accounts. So by having this essentially, you know, the ability to analyze a subledger with all the atomic records, that that's really the last phase, which is the full unlock of of true analytics and and AI readiness. 

‍

Ben Murray, the SaaS CFO

Yeah. I mean, that that's what's so important because say say you post a batch rev rec entry and it's, alright, it's a million of of MRR in that month. Okay. That's fine. But then backing up, where did that come from? Does does can that produce an MRR schedule? Does that have the detail of the sub ledger? You know, and that's what we need because there's always the balance. How much detail detail do we have in the GL versus these other submodules or systems that connect to our GL? But, yeah, backing up from there, we have posting. Alright. We have all the detail, and it goes alright. That connects to invoice data, and that's really that utopia. And then, you know, that that this third tiles box. Right? All that metadata associated with it. 

‍

Jason Berwanger

Yeah. Exactly. Well, and I I think common examples is like customer segmentation. Right? Yep. And, well, it's not a GL segment, but there's a customer where you have a product type or a location or a marketing channel that they came in on or maybe some combination of those things. If that's in that metadata and you can slice and dice your your ARR and your MRR and your earned revenue by those segments, that's where it becomes really fun because you trust it because it ties back to your p and l. But then you also have the benefit of both of the metadata to slice and dice to form whatever hypothesis that you're looking for is that and I think that's what you're getting at, which as an analyst, that that was always the thing that made me really excited was, alright, we're gonna segment the p and l and profitability and customer revenue metrics. But the question always became, if it didn't tie back to the actual p and l, well, then no one trusted it. You never took any action on it. Debit's the whole purposes of doing it in the first place. 

‍

Ben Murray, the SaaS CFO

Yeah. Exactly. I mean, it's this is the data funding I mean, it's so important. You know? And if you don't have a number of schedule that ties with, you know, then that's not gonna be trusted and useful. Uh, you know, so there's that step by step process that's repeatable. Uh, you know, so it's just just so important. And, uh, and I know hey, guys. There are couple polls in the chat too if you wanna answer those. But, uh, yeah, I I live and breathe this, Jason, every day with the SAS and AI companies that I help, but just that foundation. And and also, right, here's an MRR schedule, but then you have segment. Right? Buy major product lines. Do you have SMB versus enterprise? How do you slice and dice the firmographics? 

‍

Jason Berwanger

Exactly. 

‍

Ben Murray, the SaaS CFO

And that's where FP and A has to be closely associated to, you know, to the accounting side, uh, you know, that we're all, you know, trying to go for that that same goal. So it just cannot be siloed anymore. 

‍

Jason Berwanger

Yeah. And I I think, you know, from a from a marketing and SaaS perspective, I think folks really wanted to think about FP and A as this, like, separate tool set and separate data and separate platform. But what you're highlighting is, you know, great FP and A starts with great accounting and a great chart of account and all that data. And finance and accounting really should operate off of the same data set. And, you know, they they haven't been in recent years because there's a lot of cool tools that came around. But then the efficacy and the use the usability of those tools is questionable if it's not connected back to the accounting team because then if you've got, you know, an ARR schedule that doesn't match the P and L, then are you really gonna take any action on that that you trust? And the answer that you're pointing out is no, and I feel the same way. Yep. Cool. Um, well, let's let's pivot a bit. Um, let's let's think more of, like, the the CEO board level. Um, maybe walk us through you know, if you get to ask a question like this, Ben, of, like, the CEO wants to use AI to predict churn from payments. How do you know you're ready? Uh, how do you know that AI is the right solution to apply to that problem? And maybe maybe walk us through your thinking here. 

‍

Ben Murray, the SaaS CFO

Yeah. Yeah. I mean, this this gets back to kinda what we're discussing is, you know, if you just have generic data, uh, you know, if you just you know, and if you're a larger company and just have stuff at a higher level, like, how the heck is AI gonna predict churn? Right? There's just not enough data points in history to do anything. So you, like, you look at some of these bullets. Uh, you know, billing cycles. Right? That's a big deal if you invoice monthly versus annually. Those usually, the retention profiles are much different. And then the firmographics, we talked about metadata. Like, is this a you know, you know, if you target different verticals for your software, like, that can show a pattern. You know? So that's where a lot of companies fall show fall so short is, alright. We've got this invoice. It creates MRR, uh, but then we have nothing else associated with it. You know, no firmographics whatsoever. No better data than that. Well, we've got some, you know, some fields in HubSpot. You know? But then the customer name is different than the how the customer name sits in the invoicing system, and you can't compare that data. You know? And I see this this over and over. You know? So I think this is where we really have that metadata so important for this to predict any sort of churn. And then I think kind of the utopia is like, yeah. Can we you know, is there any sort of usage data that we can attach to these customers as well? And that's kind of that dual threat, not just finance, but also usage data. Uh, so I think, you know, gets back to the metadata and just the flow of data and what data points we have to predict churn. 

‍

Jason Berwanger

Yeah. Makes sense. Uh, you know, I I think the only thing I would add in is also maybe some of the firmographics of the business. Uh, even if you can identify those billing cycles, but you're an SLG business with not a ton of volume, you really need a lot of volume to have AI or any type of predictive model to be useful. There's gotta be a a pretty deep density of data there to to to be to have, like, past performance be indicative of of of the future. Um, and so I think in context, it depends on the business model and super high volume PLG customers. Like, we like, Hublify has a forecast mechanism, and, you know, we've got a good sense of, like, when things are gonna roll off and and out of the backlog and how they're gonna earn out in the future. But we can do that because we've got, you know, customers that have tens of thousands and up to hundreds of millions of transactions, and that's enough data that we can use to predict and so can the customer. But, uh, for folks that are maybe on high dollar transactions, like, where every contract's very variable, like, yeah, I I don't I don't think you're gonna see a lot of value in AI trying to predict churn. You'd you'd probably be more, uh, uh, you'd be more useful to really understand, you know, uh, both from an external data perspective, uh, you know, what are some of those benchmarks and what are what are other businesses like yours experiencing in terms of churn and then comparing that to your churn. But, yeah, I I I think, uh, you you know, not not having enough of that historical customer data is a is a prerequisite to being able to use it for predictive and and and AI functions. 

‍

Ben Murray, the SaaS CFO

Yeah. That that's a great point. And I'm always talking to founders and CFOs is, like, how much we need enough data flowing through your business to be able to put metrics frameworks in place, calculate different things. And, yeah, if your 10 enterprise customers paying you a million year per each, yeah, like, I don't need AI yet. I don't need, you know, anything too fancy. You know? But if I have 10,000 customers, I'm not paying me $50 and 10 different pricing plans, then, yeah, there's a lot of data to work with there. Yeah. 

‍

Jason Berwanger

And then you can predict how often someone will will renew after their first year term versus renew after the first month term. And, yeah, there's enough data that it's predictive, and it's actually useful to an AI model just like it's useful to an analyst's model if they're doing it manually. Yeah. Makes sense. Cool. Well, let's keep going. Um, you know, let's get into the, you know, the the prerequisites and, you know, the the the specific processes that we think that folks need to have in place, you know, uh, let's say, uh, prior as a prerequisite to being able to deliver value with AI. Cool. Uh, well, Ben, Ben, I'll let you go first, and then I've I have some thoughts. But, um, if you wanna frame it up and kinda walk us through the prerequisites, and then I'll we can go a bit deeper. 

‍

Ben Murray, the SaaS CFO

Yeah. And I'll I'll kinda this first box, I'll I have some thoughts on continuous revenue close, but love to hear your thoughts there. But, you know, kinda almost going backwards here with, uh, with the audit trail. Uh, you know, I think this is great where, you know, you're almost, like, thinking like an auditor. You know, can we trace, uh, you know, data back, uh, you know, from alright. It hits the p and l, then it came through the RevRec system. It was an invoice at some point, and then it then we had a contract. You know, and that's big. Do we have a systematic process? And I like to call, like, that quote cash utopia. So, you know, the data starts here with our prospect and then it's here and ultimately collect cash. And is that a nice system? Have we mapped out the tech stack? You know? So I think that that's a big one, a big prerequisite for this. You know? If it's all disjointed and manual journal entries and true ups and prior period adjustments, uh, we're we're not there yet. You know? And then looking at the, uh, consistent recognition logic, that's a big thing. Do you have you know, Jason, like you mentioned, maybe we have an SLG motion. Maybe we also have a PLG motion. Maybe we have a variable. Maybe we have usage revenue plus fixed ARR contracts. Uh, you know, so maybe those are coming from different systems. Uh, but we've got a that the framework that we put around that still still applies. You know? And we're gonna create MRR schedules. We're gonna create it by the different revenue type. Then we need to segment it with metadata. You know? So, you know, that's where I see just crazy revenue postings, and, you know, there's really not much we can do with that. And then going back to then the source system integration, uh, Yeah. Like, are we connected to to tech stacks where the data is flowing nicely? Are we, like yeah. Or is it just CSV uploads? Are these batch entries, and we have some backup sitting in a in our closed folder for the month? Uh, you know, so, again, following that data, following the tech stack progression is important because we need a systematic process to this. And then, you know, let you know, because these revenue close, I don't I I feel like, Jay, this is a big one up for debate. It's like, yeah. Is it no longer, you know, hey. You know, we hit day minus two, start some close, and day zero and all that stuff. Uh, you know, and can we do a a day zero close or a day one close? But, you know, definitely on the revenue cycle, um, right, we we need like I said, if we're posting that batch entry or posting multiple entries, we have all the source data that backs it up. 

‍

Jason Berwanger

Yeah. No. I I I agree with your skepticism. I I actually don't I don't like framing or thinking about it like continuous revenue close because that's almost implying you're you're closing repeatedly throughout the month. And, like, who wants their close to last longer than it needs to for the month or day? Um, but I do think when we think about, uh, rather than continuous close, but, like, maybe having revenue be automatically closed daily, uh, when I think that is important, um, and it will enable a day zero month close because it does two things. If there's exceptions, you know about them during the month rather than having a pile of work waiting for you and a problem at month end. Two so there's a compliance benefit and a workload benefit, I would say. Two, uh, yeah, to to to be the strategic partners for the business and to help with those board level questions and really to elevate finance and accounting, you need to have data more often. And so being able to understand, like, how, you know, how promo code, uh, improved backlog and what you're getting out of when you did a new product launch. Or there's, like, a pricing issue. We've seen this with a lot of customers where there's a pricing issue when you launch a new product and, you know, you're you're are you mispricing it in you know, for a month and a half because you didn't catch it until close? Or because you were actually connected to the business, you can go, woah. Woah. Woah. Like, our unit economics are are way off on this product launch, and I've seen finance teams catch both fraud and, you know, the, uh, you know, promo codes getting applied or, like, double coupon clipping where there's, like, a free sign up, but then they keep getting free, uh, you know, essentially free service over a period of time. And so, yeah, no. I don't like the idea of continuous revenue close, but I think the idea of having daily, uh, you know, revenue numbers that are automatically closed is a key part to AI being, you know, useful because that that, you know, contextualized atomic customer data is a prerequisite. Um, and so because of that, you you need that more often, and that gets you the volume of data you're looking for and the detailed level of customer data. So I I think we mostly agree there that, yeah, the continuous close sounds like a nightmare, but, you know, having daily revenue you can trust feels like the prerequisite here. 

‍

Ben Murray, the SaaS CFO

Yeah. Yeah. There's a question in the chat or in the q and a. Uh, you know, how clean does our historical data need to be before we start using a for protections? Do we need to go back and fix everything? And that, you know, that's always the the question I'm even just putting in the right foundation, uh, is well, either you have a choice. Either you fix it day from day today going forward and time goes so fast, and soon you'll have enough historical data, or maybe do clean up the current fiscal year, you know, so we can have that a good year over year comparisons next year. But that's like, how much work is that? What data needs to be fixed? And and I've seen it. Like, I tried I'm, you know, using AI on a lot of different things. It's great. Like, you know, summarize this podcast transcript. Perfect. No problem. Done. You know? But how often have I tried to alright. Alright. Summarize this data for my board or an investor update. It gives you 10 different answers or short answers, and that always gets back to, does the AI have clean data? And can it get to the source of all these all the data points? 

‍

Jason Berwanger

Yeah. I think that makes sense. And, uh, the the thing I'd add too too is it depends on the use case of what you're predicting. But more often than not, what we see is that you're you you need to have the auditability, so, like, the provenance and lineage of the data as you're applying AI. That then protects the output so it's actually useful. Because let's assume that you don't have, you know, that auditability that then you were talking about. Well, you're gonna then get an output from a model, and you don't know how to prove whether it's right or wrong, which means that it's it's virtually useless in terms of making decisions and taking action. Versus if your historical data isn't perfect, but you can at least prove why the model concluded what it concluded because you've kept this lineage of provenance, well, then that that that means you're it's clean enough to start iterating on models. It's kinda our our day to day take on it. 

‍

Ben Murray, the SaaS CFO

Yeah. And I I'm starting to experiment with some of my own AI stuff in in my financial process and and creating metrics and explaining variances. And and it's it's promising so far, but it's giving so much detail that a lot of numbers this went up, this went down, and that's why CAC payback changed. It's like, wow. Like, it would like, to write this would probably take me eight hours, you know, just to get to that minutia. So it's like, there's gotta be a lot of trust there, and I think that's why it's you know, it's like, well, are these numbers correct? It just explained three reasons why CAC payback went up or down. Well, is that really right? You know? And it can do so much more, but it's like it's trying to get to that like, I found, like, this yeah. That data structure, the data access so important. And then, Jason, there's a question here when you're talking about revenue close. When you say, do you mean daily revenue recognition when you say daily revenue close? 

‍

Jason Berwanger

Yeah. I I I think that's part of it to a degree. Now there's there's a nuance here where there's there's daily revenue recognition where it's a pro rata daily rev rec. Yep. And not everybody's on pro rata daily rev rec. And I'm not saying you have to be on pro rata daily rev rec for you to take advantage of that or need their daily revenue close. But I think what we're saying is revenue recognition, the invoicing subledger, the customer financials, all of that needs to be recognized and, you know, fully correct and closed on a daily basis to the point where it's both reconciled and, you know, it it's fully automated into the journal entries because that then gives the baseline of analysis that you know you can trust, and it'll tie back to your mid month p and l. So, like, I yeah. I think we're basically saying, yeah, daily daily rev rec in order to cash accounting, uh, which is is really the prerequisite to to getting the outcome that you want for for analytics purposes. Cool. The only other quick, uh, real quick. Let's go back one slide. I was gonna add, uh, one ticket context. Uh, the source in that system integration, I I think this is a key differentiator for, like, getting the right outcome. Uh, Uh, you know, being able to have finance and accounting connected to the operational or customer systems where your customers are transacting, this is a key prerequisite. Because if not, you're dealing with either composite, like, batch shuttles or you're trying to manually move files from the source system. They're not codified into financial transactions. So you're basically asking AI to interpret, you know, the what are those transactions because they're not really modeled into financial records yet. They're still just the raw customer transactions, and the interpretation of those is limitless. So I think that source system integration, you know, outside of the, you know, audit trail itself and, you know, having consistent, you know, rev rec in a policy, that that feels to me like one of the more important bits here. 

‍

Ben Murray, the SaaS CFO

Yeah. I think I don't know, Jay. Like this, there's, you know, great bullets on on this slide, these four questions they answer. But I think the the first bullet is a big one. Pick an invoice at random for six months ago. Can you reconstruct the full revenue recognition path? And, you know, that's what I'm always doing when I'm creating MR schedules, you know, for operate you know, just to have in place and then also when it's requested due diligence is is that just that trail. Alright. So we post this revenue. Alright. That came from some rev rec engine. That came from some invoice. It came from a contract. And can we tie all that together? So we believe, one, in our revenue postings, do we believe in our revenue in our MR schedules? And MR schedules create MR waterfalls, which creates gross revenue retention, net revenue retention, always requested in due diligence, of course. You know, so that's a big one. You know, is that a circuitous path? Is it a manual path? Is it an ambiguous path? Uh, are we posting rev rec correctly? Uh, so, yeah, I think I think that's a big one. 

‍

Jason Berwanger

Yeah. Agreed. Yeah. And then the second one is, like, can your controller understand the rev rec methodology? To me, this is really a policy and a governance question. Right? If you if you don't have a common definition in of how you're applying your policy, uh, to the actual methodology of rev rec, then, you know, you're you're probably not at the sophistication needed to then somehow automate and and use AI to improve that because there's still some very foundational governance aspects and and policy aspects of what needs to be done. So I I think that second question is pretty key because it it forces you to get the governance and and business and accounting context and policies correct first before then trying to skip to automation and kind of the sexy fun stuff at the end. 

‍

Ben Murray, the SaaS CFO

Yeah. And and the third bullet, recon's so important. You know? And and I was like, alright. I follow, you know, step one, step two, step three to recon my books, to recon deferred ledger. Uh, you know, so I kinda run this report, run this analysis, boom, ties, I move on. Or does it take you days to recon? You know, that's gonna be a problem, and that'll be a problem for AI. Uh, and then, Jason, there's a question in chat for you. I think it's interesting. Uh, have you seen any of your customers find creative ways to connect nonfinancial kind of, you know, quote, unquote, atomic data to tools used frequently by early stage companies, uh, say, using, uh, QBO? Do they use custom fields in QBO? Do they use snake, Snowflake, or other data solutions? You know, pull those together. 

‍

Jason Berwanger

Yeah. That's a great question. Uh, I would say disproportionately, that's that's how most folks leverage, like, Hublify and a number of these automation tools. More often than not, we're connecting to nonfinancial atomic data like Stripe or a lot of folks have, like, a bespoke internal billing or order system. We're connecting to that plus the payment processors, and then allowing folks to apply their policies, and then we're creating the journal entries into NetSuite, QuickBooks, or basically any GL that folks have. And so there's a there's a last mile for that as well, which is we're posting some of these transactions. Sometimes they're composite totals into QBO, but sometimes we also push a bunch of detail into the general ledger, though we don't recommend that because, you know, general ledger is only as useful, you know, as as reporting that you're pulling out of it. Um, but then we also have integrations where that data can then be shared back to things like Snowflake, which then are are basically hydrating the BI tool layers in the analysis, which Ben, that that's kinda like the the pipe dream that you and I have talked about, which is, okay. Now you you've connected to your source systems. You've automated all the calps and the financial data into the ledger with enough detail you can fully reconcile, but not so much detail that it breaks your general ledger. And then that data is now shared back to all the business teams where revenue is actually governed by finance and accounting, and it ties back to the p and l, and everyone's making decisions off the same dataset. And to me, that feels like a very scalable way to run, like, FP and A and analytics, and it it ensures that the folks that are making, like, P and L ownership decisions like marketing and operations are actually using the same revenue number that accounting and finance are. 

‍

Ben Murray, the SaaS CFO

Yep. Same AR number, same revenue number. Uh, Yeah. Definitely. It always comes yeah. That well, what's that source of truth? 

‍

Jason Berwanger

Yeah. Makes sense. Cool. Alright. We can pivot to the next slide at this point. So, you know, when we think about, um, you know, our our road map and, like, how I think this line of questioning, Ben, is really thinking about how, you know, how how does a team take their existing, often manual process, and there's nothing wrong with that. How do they take that, and how do they get from a to b to c, to d, to e in terms of the layers that they need. Right? Like, we we've been clear that there's a first principle approach, but there's clearly hard prerequisites that we've laid out. So how does how does a team go from current state into, you know, some of the, you know, more fun later stages once they've made those first principles investments? 

‍

Ben Murray, the SaaS CFO

Yeah. I mean, this one is one is kind of, alright. What's the landscape of your tech stack? Can you consolidate systems? You know, when I've seen, alright, we invoice from Stripe. We invoice actually, we have a few customers we're invoicing out of bill.com, and then we're invoicing out of QBO. It's like, alright. Well, can you know, it's so much easier when you have one source of truth. And sometimes that doesn't you know, maybe we'll always use Stripe and maybe we'll invoice, you know, enterprise invoices out of here, uh, but it flows to one system and, you know, we can combine this data. You know, and that's that's so, you know, looking at our tech stack, like, is this the right tech stack for us? Is what's that flow? Uh, and like you say here, document manual intervention. You know? And, again, that quote obviously, as a CFO, the quote to cash process is utopia. All the systems tied together. It pushes data from here over to here, the right structure is in place. And and so, yeah, first, what's our tech stack? And how does all the data flow into the GL? And then, also, what data do we need to support sometimes journal entries? Maybe that are cool. You know, it comes from a different system or Snowflake or whatever it might be. You know? So it's it's really documenting that that tech and data flow first because, yeah, yeah, if we don't understand that, like, there's there's no point moving past this step. 

‍

Jason Berwanger

Yeah. That that's really the manifestation of the how are your customers transacting in your business model and how does that get into your GL. And if that isn't clear, then there's no reason to move forward is what I think we're we're hearing you say because, you know, you really have to understand those inputs in terms of how the operational data manifest with financial reporting. And if that's not done, then you wanna start there. 

‍

Ben Murray, the SaaS CFO

Mhmm. Yeah. Yeah. Definitely. If you don't know what that batch entry means for the revenue posting, then, you know, we we've gotta take a take a pause. Yeah. Then, yeah, get to continuous close for your largest revenue stream first. Yeah. I mean, that's I mean, the the eighty twenty rule eighty twenty rule of forecasting is where where should we focus? Uh, you know, if we have 95% of our revenue subscription and 5% professional services and onboarding, Well, yeah, focus on subscriptions. We have usage. What are and that's always what I'm going through is our p and l. What are our revenue categories? Are those clear and distinct? Uh, you know, do we have just a subscription? Is it monthly or annual? Do we have usage revenue? Are we taking a cut of credit card processing revenue? Where is this coming from? Uh, so, yeah, thinking about just that automation of invoicing. You know, again, it's the same thing, invoicing to rev rec, how that post to the p and l. Uh, you know, so we have a nice process, and that's not not a manual process. I mean, really, for me, that should you know, I should be able to run that process in an hour or less. 

‍

Jason Berwanger

You're really observing and using the output of the process versus actually managing and reconciling the process. And that that's the key distinction. I think if you're if you're still performing the actual calculations and the reconciliations, that's gonna end up being the blocker of you both your time and the ability to use the data for analytics and 

‍

Ben Murray, the SaaS CFO

AI. Yeah. Definitely. Yeah. And this yeah. I mean, this you know, it's a right of passage for all FP and A analysts, you know, to to write variance explanations and understand variances. And that's where we catch a lot of things. That's where we catch missed journal entries or wrong journal entries or incorrect things. So there's actually a lot of audit going on the FP and A side when we look at variances. And, again, it gets into that detail. Alright. We have variance here getting into the core data, what GL transactions made up that that number that I'm seeing on the p and l or in that GL account. So this is where yeah. Again, that source entry or source data, the GL data, uh, is what we need because if we just have one just one top level number, uh, and we're comparing that to budget, I really can't do a great job digging into that or explain variances. So, uh, yeah, and and doing this in a timely way where it's not you know, I don't spend an hour just trying to explain one thing. 

‍

Jason Berwanger

Yeah. And it's it's really there's there's both a analytic, and there's also a compliance aspect to that where the compliance aspect, you question if if you have the, you know, the grand total entry for revenue. Well, unless you're keeping a subledger that proves out what you made your entry, like, that's actually not a great way to set yourself up for the audit because then they're gonna ask you for samplings and all the records, and, hopefully, it's the right point in time file that you you pulled down to substantiate what was a composite journal entry. On the analytics side, same thing. Without that level of detail to your point, you can't segment it. You also can't explain why there's variances month per month because well, you know there's a variance because you compare the two totals, but you don't know why and you don't know what makes up the variances to compare, like, the individual characteristics of the two, you know, datasets between the July and the August comparison. So I think that makes a lot of sense of of, uh, you know, both from a compliance and an analytics perspective, how to how to use that variance analysis and the atomic data to explain what's happening. 

‍

Ben Murray, the SaaS CFO

Yeah. I don't know, JC, your thoughts of it. Like, lower risk stuff, if we think. Right? It's AI is not used consistently in the back office yet. Uh, but why is that? Because we have to be so accurate down to the penny. Uh, but, you know, you think about this when I've been reconciling my deferred sub ledger and I reconcile usually, you know, to the penny. If it was off a $100, a thousand dollars, I just I had to find it. Like, what is that variance driving it? And, again, now if I look back, if I had a very structured dataset, you think about AI, the more structure there, like, AI could have helped. Like, what went wrong? Why can I not reconcile by this thousand dollars? Can you look for your patterns in the data that would explain that? And if if we have that data, we have the structure of the core data, that's gonna make it so much easier to explain the variance. Like, what stands out in this variance? You know, look at the data. Uh, so I think that's, you know, where it's like, yeah, it's not in some of your piece now, maybe having AI prepared journal entries, but we can start with lower risk things that maybe we're not posting a journal entry. It's not doing the forecast for us yet. Uh, but we can have it assist us in searching through volumes of data that's just gonna do so much faster. 

‍

Jason Berwanger

Yeah. I I strongly agree with that. And and we that that's a use case that we've seen customers start to adopt with HubSpot five recently, particularly around things like your, uh, your dunning process and and where your where your AR and some of your your short term collectibility risk is. We've seen a lot of customers be able to ask, like, hey. Where should I spend my time? Like, what are the customers that are they still are, like, getting service on Stripe Billing and their subscriptions, but haven't yet paid or maybe had a chargeback? And that quickly is a way to then, you know, have AI comb through a bunch of that data and then surface out what are some exceptions for you to spend your time on. And there's a lot of analytical value in that. Not high risk because you're not you're not pivoting how you're running your business, but it's giving the finance and accounting team focus of where to spend our time and leveraging, uh, you know, some of those exceptions by coming through hundreds of thousands of records and, you know, pointing out where there's problematic customers where, yeah, you're probably getting the 

‍

Ben Murray, the SaaS CFO

And that's where Amazon delivered. Yeah. Yeah. I know. And that's where, like, when companies move past that point from, like, visual inspect inspection of your data, maybe I just have a 100 customers. I can review them or schedule visually. What if you have thousands? Like, then you have to rely on your system, your data, your structure. You just have to have faith in the system. And this is even without AI, and now AI can help with that. Uh, but, yeah, you always move past that point where it's like, yeah, I can't review 5,000 rows of data, uh, and inspect and see what went wrong. Uh, you know, so I think there's, you know, just that slow infusion of AI where it can just help us with fact checking and and risk analysis. 

‍

Jason Berwanger

Yeah. Exactly. And it's less about, I think, predictions at that at that stage and more about, you know, how to comb through and find patterns and thousands of rows of data, but it's based on prerequisite that the subledger that AI is scanning is correct. If it is correct, then the risk goes down, and you're you're starting to spend your time more wisely and on things that actually improve the bottom line or our compliance concerns. Um, so I I think when when we talk about, like, the lower risk items, it's really surfacing data to folks that we trust the human in the loop to then spend more of their time of things of analytical value. 

‍

Ben Murray, the SaaS CFO

Yep. And then, you know, it's it's you have tech stack data structure and then, uh, yeah, then getting real help from AI in this process. So I think that's why it's so important. We always do in the back office CFOs, we need technology. We can't just scale our our finance and accounting function manually. We've gotta have the right tech in place. Uh, you know, so, you know, thinking about, alright. I've got a systematic process. You know, I can reconcile the books really fast. And, you know, just think about, alright, what processes are still manual, you know, or just take a lot of time where I can overlay AI? So, uh, you know, that's a great example of, you know, AR analysis, uh, you know, different ways we can we can infuse it now and into our practice, into our monthly process. 

‍

Jason Berwanger

Yeah. Well said. Yeah. This is really the the chart of accounts is in place. The policies are in place. You're connected to the source systems in your business, and you have this level of detail that you need. You can slice and dice. You've applied AI to places in the sub ledger that you trust, and you're becoming more efficient. And now you're taking some bigger bites out of the apple, which are where, you know, you know, where do we wanna double down investments with marketing and and and, you know, from a selling expense perspective, you know, where where could we focus our time on, you know, collectibility or, like, a seasonal analysis and risk based on, again, that deterministic sub ledger. This becomes the fun part at the end where, you know, you can trust the data, but you're still not asking it to, you know, create journal entries or to do deterministic type of accounting things. You're using that data to then apply AI to then get a a a business outcome from an analytical perspective. 

‍

Ben Murray, the SaaS CFO

And and, Jason, Scott has a good question here. What AR number or perhaps number of customers would you recommend a team begin with this audit process, you know, and then scale the process? And and, you know, that's it's a great point. And I think for me, I kinda mentioned before is when you can no longer like, if something's wrong with your number, something's wrong, you can't reconcile subledger, Deferred is not tying where you can't where you go from, well, I can just look through the data and then see, oh, yeah. Look at this. The dates are wrong on this invoice. Two, there's too much data. Right? When there's just too much data where I can't visually inspect and run that quality control process, then definitely that's we've gotta do the audit because now we're relying on our process. We're relying on systems. We're relying on how that invoice data got entered. You know? And that's when you know? And that could be a 100 customers. It could be a thousand. Well, a thousand, definitely. But we're we're we're making that at least that's my opinion. You know, that transition where where when I post that entry, we're really relying on all the logic, the systems, everything to do its thing, you know, to have things correct. And that's where, yeah, we definitely need to audit our process. 

‍

Jason Berwanger

Yeah. Well said. I think I agree. I I think that there's a big difference between a $5,000,000 ARR business with, you know, less than a 100 customers and one that's, you know, 5,000,000 ARR, but then has a 100,000 customers. And disproportionate of what we see is that, like, 3 to 5,000,000 ARR range for folks that have a PLG motion. That's a that's a good time to start investing in this. If you're SLG and you're 3 to 5,000,000 and, you know, you've got, you know, a couple $100,000 on average contracts, that's probably a bit too early as a this is not a science, but, uh, I think that's that's my take in terms of where are you actually gonna see benefit from starting this process. You you gotta have some degree of transactional scale, uh, on on your revenue accounting process. 

‍

Ben Murray, the SaaS CFO

Yeah. Definitely. Cool. 

‍

Jason Berwanger

And let let's take this one, and then we'll pause for questions. And I know we're about out of time at this point, so we'll we'll pause after this for final questions. 

‍

Ben Murray, the SaaS CFO

Oh, man. You know, this, I always say, you know, you gotta treat your MR schedule like gold. Right? You need it to run a business. You need it for due diligence. You need it for your board, investors, potential investors, buyers. And why is this so hard? And I Jason, I live this every week, every day, uh, with with clients. And one, we're don't really have a formal invoice process or what systems we're using. You know, we started out using Stripe for just credit card transactions. Now we're actually in invoicing enterprise invoicing out of Stripe. Uh, but then we've invoiced out of QBO too. Uh, and then also we don't capture all the metadata. We invoice that customer. We know it's an annual invoice, but nothing. No data shows that that's an annual invoice. So how do we create an MR schedule? How do we do a rev rec? Uh, you know, do we have it are we bundling revenue together? We invoice them for 10,000. Oh, that includes services too. You know? So it gets back to really the core invoice data, the metadata start and end dates, SKUs, how SKUs tie, you know, uh, cross reference to the what GL that they're posting to. So there's a lot here. But if you if, guys, if you're struggling with this, this is a common struggle. 

‍

Jason Berwanger

Yeah. Well said. Yeah. The only thing I add to that is the I I think part of the reason why the AMRR and ARR is so challenging is typically I see this, uh, as a metric outside of finance or that finance is performing on operational data like CRMs and or the data teams forming the analysis, and it's completely decoupled from the actual earned revenue and rev rec. And, Ben, you and I talked about this ad nauseam, but I I really think when you think about ARR, the recurring revenue for monthly and annual being tied to actual rev rec, that to me is the unlock because that's what makes everyone be able to trust the numbers. And I think it's really tough, though, with early founders, especially because you wanna take advantage of your ARR the minute that customer has signed, but you may not have your earned revenue yet. And so there's there's a tendency from the business to pull all that forward, like, the minute it hits the CRM is closed. But, uh, I think your approach is the right one, which is ARR and MRR are based on the actual p and l earned revenue, and that keeps that metric completely trusted and gold, and it keeps finance and accounting and management on one source of truth. And, uh, I think that's often, though, a hard lesson where it starts as two different numbers, like, once for bookings that doesn't matter in your QuickBooks, and then the other one's kinda like how you talk to investors. But that quickly catches up to folks where, you know, your your CRM data, when you rerun MRR for January and it's now a different number than the last time you talked to an investor, that's usually when people panic, and they're like, oh. Or walking it back to your p and l, or, like, how does your schedule tie to your p and l? Well, it doesn't, and I've seen a lot of diligence get blown up because of that, and I know you have too. 

‍

Ben Murray, the SaaS CFO

Oh, yeah. Yeah. And then opening up the books from three months, four months ago. You know, sometimes that happens only worst case scenario, but where prior numbers are changing just drives me nuts. Uh, so, uh, and, you know, there is there's a question here. If we do this foundational work in q one, what's realistic timeline of CROI from AI implementation? And, right, alright, let's say we have perfect data structure in q one. You know, it depends where we're gonna deploy this. Uh, you know, one, as an FP and A guy, gosh, I love it. You know, write the core variance explanations, write the board update, all that stuff that can save you so much time and then you can fine tune it, huge savings on the FP and A side. Uh, you know, so it it just depends. Are we gonna, you know, tactically in in, you know, maybe fraud analysis or journal entry analysis or posting analysis on the accounting side? But, yeah, like, on the FP and A side, like, the huge, huge ROI there from that. 

‍

Jason Berwanger

Yeah. Yeah. I I I'll take a different spin on the the answer to this question, but I think, you know, in terms of foundational work, more often than not, we see customers get more ROI more quickly from the foundational automation work than they actually do from AI. And Mhmm. We've seen this on our our last six implementations we've done with customers of, you know, when you tie the, you know, the accounting team to the core customer financial systems and you get that data in their hands, the accounting team does a killer job translating into, like, well, where are their customers that are getting service that are should be charged but aren't being charged and finding revenue leakage? We We find a lot of revenue leakage or, you know, bad debt or all the type of use cases the folks just never knew about because they only just had high level numbers. I think if if you start this work, it's realistic. You're gonna get value out of that in weeks, not by the end of the quarter. And I think a disproportionate amount of that value will actually be done in the foundation work, not just from the AI outcome at the end. So, um, that's the benefit of taking as the first principles approach there is that here's the three or four problems yourself along the way, and I think they all have a ton of business value, and you'll you'll be repaid in kind. Yeah. Uh, another question. Do we need to hire someone with AI expertise to manage these tools, or can our current controller handle it? Ben, lead the way, and then I'll I'll follow you. 

‍

Ben Murray, the SaaS CFO

Uh, no. I don't think you need someone with AI experience to manage these tools. Right? I mean, like you said, Jason, it comes like you need automation first. Right? We need that nice quote cash process. We need systematic ways to close our books and then look at AI. Now now you're seeing a lot of jobs. Yeah. Yeah. AI experience is super helpful, but I think in the back office, it's still we're on the frontier, uh, versus, say, like, uh, you know, marketing or go to market function where, yeah, it's highly ingrained. So I think someone yeah. Definitely, someone has to, one that's like no tech stacks. Where is the tech stack going? Where is AI going? And not just, oh, yeah. I can do debits and credits. But I think it has to be awareness around that the function is changing, and we we do have to see you know, we have to follow that, you know, so we're not left behind. Yeah. 

‍

Jason Berwanger

Yeah. I think the thing the piece I'd add to this is I I think there's a a strong opinion that the accounting team so the controller in this example should own of the output of all of these, which is to say that they they probably don't, and we shouldn't have someone else, like a technical AI person that isn't the controller that doesn't have that responsibility of finance accounting, should not be who who owns these things. That's I think that's a that's a key distinction. Now if you need a tech partner that you wanna get enable the outcomes faster and you're dealing with internal data and the integrations, like, I think it's probably fine to have a technical partner, but the person who's gonna be held responsible is the the controller in this example, and they're the ones that gonna need to own the outputs of those tools. 

‍

Ben Murray, the SaaS CFO

Yeah. Yeah. Definitely. And this is a good one, I think, good one for you, Jason, because you're building, you know, building software, but advice for evaluating all the AI native tools out there. 

‍

Jason Berwanger

Yeah. I I think more often than not, what we've seen from marketing is there's a ton of, like, AI native everything, um, AI native ERPs, AI native clothes. And the thing about it that we've seen is they're neither AI native nor early ERP, and that also isn't the right thing to ask. Right? Like, no one no one has a problem that what they're the problem is not that they need an AI ERP or an AI close or an AI anything. There's a real business problem that they're trying to solve, and I think, you know, first principles, um, you know, you you think about solving that problem and those source systems and the automation, the reconciliations, and then you evaluate the tool against the business value that could be created by by, uh, having AI slash automation help with those things. So, um, Yeah. I think that that's a that's a part of, uh, the way to think about the AI native is, like, AI native is a great way for, uh, you know, to talk to investors if you're a SaaS company, and you'll get a good valuation. But that isn't what customers are thinking and wondering, and that's what the problem of the business is. So definitely dig deeper into what is the first principles problem that you need, and then can this still help me solve it? And whether the AI native or not is actually, in many ways, irrelevant. Mhmm. And and and and it's more about how how how important is that problem to get solved, and is this the right tool for that? Alright. We'll take one last question. Uh, when you're describing AI, are you referring to to using tools like FloQast and close analysis? Do you mean, uh, staff person building AI analysis, like financial modeling and JetGPT, like an AI platform? Ben, if you wanna take that first, great, or I can take the first step up to you. 

‍

Ben Murray, the SaaS CFO

Yeah. Yeah. Take the the first stab at that. Yeah. 

‍

Jason Berwanger

Yeah. No. Don't don't mind at all. Um, Yeah. I I think, Lonnie, the answer is we're talking conceptually, but trying to ground with specific use cases. And I think we've hit on a bit of both throughout the presentation today. There are things like FloQast where a FloQast is really connected to your GL, and your GL has a ton of atomic contextualized data and a clean chart of accounts. I think FloQast has a lot of value they can add with that close analysis tool, and that is absolutely an application of AI. But there's a lot of work that had to happen very foundational to what Ben and I talked about today for that to be valuable because what you can't do is just go by by FloQast, but then you have none of the foundational things in place. That is an AI tool. It'll just get you nowhere like we've pointed out. Uh, and then, certainly, there are, I think, tons of examples of, you know, using, uh, things like CHBT or perplexity to then try to build an MRR ARR schedule. And I think those are good helpers to get you started, and that is absolutely AI. But those are more like day to day efficiencies of an individual practitioner and not necessarily application of AI to an organizational level problem. This is probably the way to think about them. But they're they're definitely both AI tools, so you're you're totally right in describing that. 

‍

Ben Murray, the SaaS CFO

Yeah. Definitely. And I know I know we're out of time here, but it's like, you know, I think it's it's so hard if you've tried to do just analysis with CHAP GPT. It's so hard. Uh, you know, that's where you see now MCPs come out and using structured data that just instantly produces, you know, clean looking reports, board reports, etcetera. Uh, so I I think first, yeah, I don't I don't know if we wanna create our own, but for, like, you know, what you know, if we have our tech stack right now, what are they doing? You know? And, like, whether you call it or not, you know, like you said, Jason, does it make our life easier? Does it make us more efficient? I don't care if you call it or not. You know, it just doesn't make my life easier. 

‍

Jason Berwanger

That's right. Yeah. Exactly. Awesome. Awesome. Christine, I think we're a minute over, so we can we'll close-up. And thanks, everybody. 

‍

Christine Butchko

Yeah. Absolutely. Well, thank you all for participating in today's webinar, and thank you, Ben and Jason, for talking about a topic that seems like has a lot of interest. Um, rest assured, the beauty of today's webinar is any questions that weren't answered or come to you late at night, um, you can always reach out to us to answer. Uh, Ben and Jason, do you mind sharing where folks can find you, um, if they have questions? 

‍

Ben Murray, the SaaS CFO

Yeah. Sure. I'll I'll put my email, ben@thesasscfo.com. So, yeah, reach out with any questions. Yeah. 

‍

Christine Butchko

Awesome. Perfect. And Jason has added his. And, yeah, as we mentioned, this webinar will be available about two hours after the session. We'll send out a copy of the recording. So thank you all, and we hope to see you at the next one. 

‍

Ben Murray, the SaaS CFO

Alright. Thanks, guys. Appreciate it. 

‍

Jason Berwanger

Thanks, all. Take care. Bye. Bye.

‍