How AI Is Rewriting Product-Market Fit, Pricing, and Go-to-Market

0:00A lot of my use cases with the original Chack GPT mostly involved making fun poems of my friends. You know, the funny thing about this era of AI software and now agents is that product market fit is an evolving concept. You're token maxing. If we want to talk finance and jump scares, then we can start talking our Claude bills. I'm trying to get us a couple of plaques and a handshake from Dario. I've been told that we're one of the largest consumers amongst them. It reminds me of that scene in the big short. I think they're confessing. He's like, I don't think they're confessing. I think they're bragging.

0:31They're not confessing. They're bragging. I think it's safe to say all three of you have found product market fit. I'll call it a product market fit journey. I think similarly, it's not a milestone. It's a lifelong journey. The way that we were tackling the problem around complexity and billing was very centered around what data is required to be able to send an invoice at a complex organization. It's not just CRM data. It's the contract, the PO, the amendment, the renewal. How do we put together a complex knowledge graph of all of that commercial documentation? Is this thing on? Yesterday's price is not today's price.

1:15Hey, are you liking this show right now? Do us a favor and we'll do you a favor. Share a link to the show on Apple or Spotify with your team at work. Drop it in a Slack or God bless you if you have to use Teams, throw it in there too. But send a picture of you dropping the link in to Ben at MostlyMetrics.com. That is Ben at MostlyMetrics.com. Our creative dictator will hook you up with some Run the Numbers swag. What's better than having a friend with a boat? A friend with a rooftop deck in Soho. So the fine folks over at Tabs invited me over to their HQ, where all the work goes down. And we did a panel and we recorded it.

1:56It was kind of like doing a live run the numbers. And so we had co-founder Rebecca Schwartz of Tabs. We had my friend, Fred Havenmeyer, founding engineer over at a fast growing startup called Fleet.ai. And when you're the founding engineer, you really do literally everything in the company. And Raul Recki, the president of Rogo. They're making software for investment bankers. AI native software, I may add. He's been a past guest on the show. We go deep on finding product market fit in the age of AI and scaling go-to-market, which includes how you think about incorporating token costs into the product that you're providing.

2:31So if you're a finance leader out there, your hair's on fire, you don't know what socks you're wearing today, you're working at a startup that's growing super fast, these three people are at the nexus of what's going on with AI and they're building companies that haven't existed before. So this was an amazing conversation. It was like having four people going back and forth on a pod on a roof deck. So I think you're going to really like this one. Please like and subscribe. On to the interview. The only thing better than having a friend with a boat is a Soho rooftop. I'm going to go down the line.

3:05I'm going to ask a couple of warm up questions here. I told you I'd start easy, Rebecca. So first question, we're going to say your name, where you work, what the company actually does, and what you actually do. Okay, we got this. Rebecca, what does the company do? Tabs is an AI-powered billing and revenue platform, helping B2B companies not have to deal with the headaches of AR. Really excited to have so many customers and partners in the room today. So thanks everyone for coming. Today, I met five awesome candidates for our team. I talked to two customers about making sure that they're having an awesome onboarding experience.

3:42I talked to one prospect, and I talked to a bunch of people on our team at lunch and at office hours. Hello, everyone. My name is Fred. I'm from Fleet AI. We're an applied AI research company, which I feel like there's, I don't know, hundreds of us nowadays. So what do we do that's different than everybody else? We build simulation environments and evals that help to push the frontier of what AI agents are able to do in terms of what I describe as economically relevant work. So if, you know, someday you notice that agents are suddenly a lot better at doing things on a computer, using software, fundamental skills of something relevant, perhaps financial modeling and analysis, we might be the ones that really help to make that happen in part.

4:25We also nowadays are doing a lot of work to help enterprises benchmark and evaluate AI models in the workplace so that they can actually get comfortable and understand that agents are doing what they would expect them to do in a way that's reliable, repeatable, and actually useful. Today, I flew back in from SF for this here tonight, so I'm very happy to be here, so I do a little traveling. More so, though, I partner with AI Labs to help them structure, run, and manage research projects as part of what we're kind of describing as like a forward-deployed researching team. I mean, on that, I don't know, ops, everything, payroll provider transitions, hiring, office leasing, whatever needs to get done, we get it done.

5:02Raul, over to you. All right. Really nice to meet everyone here. Thanks to Taz for hosting. Thanks to CJ for moderating. Rahul Reiki, president at Rogo. Rogo is based here in New York. We provide AI for Wall Street, for global finance, for the work of bankers and dealmakers, both in terms of accelerating the work of the professionals, making models and decks and doing research, also accelerating the AI transformation of their respective firm. How do I spend my day and what do I do? Well, most of the time, I want to make sure that I'm adding value above what Claude tells our staff to do, and that increasingly gets harder.

5:41Beyond that, I oversee the finance function, so I actually really enjoyed the last panel where there was lots of discussion about usage-based and outcomes-based pricing, very top of mind, quite a lot on partnerships, and especially embedding really deeply with the large enterprise partners we work with, the likes of JP Morgan and Bank of America and others on transforming each and every part of their firm to be AI native. So we're going to talk about finance. We won't talk about deferred revenue, Doreen. Fred, I've known you for, I think, going on seven, eight years now, back when you were a sell-side analyst at Macquarie.

6:15I remember when you were thinking about going to Fleet. It was a different idea. I want to talk to each of you about finding product market fit, because I think you need that before you start to scale a finance company. Fred, what was the journey to where you are now and the actual product that you're providing? I could spend a long time talking about the actual journey because it involved a bunch of stuff. My wife and I quitting our jobs, traveling the world, backpacking, and finding out that CJ and I overlapped or missed each other, I guess, by two days in, where was that, Sorrento one time.

6:47Prior to joining Fleet, I was head of AI and software research at Macquarie, the Australian bank, where I ran the essentially enterprise software focused equity research team. Huge amount of fun. Like I got to meet and speak with everyone, but it was a less applied role. And I was looking for something where I could get in the weeds, build products and have real ownership. We just ended up meeting the co-founders of Fleet, Andrew and Nick, when they pitched me on a product, a prior version of Fleet, which back then was a financial services co-pilot. Didn't see as much uptake as Rogo. We had some customers.

7:22We were doing pretty well. But we also noticed that the deal cycles were rather long. And it's great when you have really exceptional talent to be able to own and run that. But when we were a scrappy team of three people, it took us a bit of time to come around and understand that fundamentally what we were doing is we were building what we call simulation environments to tune agents for financial services work. And rather than try to structure this into something that we could explain as a value add to Fortune 500 financial services companies across all domains of investing, etc., into insurance, we thought, why not go to the labs instead?

8:00Flex our muscles, do more work on building simulation environments for a wider variety of domains. And so that ended up taking us to AGI House in SF. That was a unique period of time. My wife was very tolerant of me flying out to the West Coast with a couple of 20-year-olds to go live in a house together. We ended up building a couple of demos based on our initial experience and what we were doing in the prior version of Fleet. And we found immediate product market fit. Ever since then, we have had more demand than we're able to fill in any given day. And I think the moment that we knew we had something was when, you know, we stopped having free time.

8:36So I can keep going, but I'll cut it there. That's great. And we were joking before the panel, and this would have been an awkward conversation if you were sitting next to Raul and you stuck with the original idea. But Raul, how did it develop for you? Because you were in investment banking before at Lazard. Did you see the product and just know it? Or was it a leap of faith and they figured it out after you joined? You know, the funny thing about this era of AI software and now agents is that product market fit is an evolving concept. I would say early on, you know, I was at Lazard. Lazard was actually Rogo's first enterprise customer or enterprise client.

9:14This was shortly after the advent of GPT-3 and then GPT-3.5. I don't know about all of you, but a lot of my use cases with the original instances of ChackGPT mostly involved making fun poems of my friends, occasionally helping with email writing. And, you know, the product market fit of Rogo originally was, hey, for leveraging GPT-5, chat GPT or GPT-3 for financial work, do the tough work of data integrations and citations and auditability so that you can do research and not have to worry about hallucinations. And that in and of itself was a breakthrough at the time, right? Now, of course, that's table stakes in lots of ways.

9:57And the real evolution of that for me in our most recent inflection of product market fit is making our platform, making those agents useful not only for the junior professionals that are conducting research tasks, but for partners and managing directors across Wall Street, including at firms like Lazard, that need a real thought partner and that see these agents as a kind of agentic workforce or agentic colleague. And there are lots of, you know, changes and shifts in both the underlying intelligence and the product later, the application later to achieve that. So I've sort of seen the evolution of that over the years.

10:35And it's only really been in the last few months that that last milestone, as it were, was crossed. Hey, thanks for listening. We'll be right back after a word from our sponsors. Remember when pricing was simple? One product, one subscription, one invoice every month. Those days are over. The AI economy is changing how everything gets bought and sold. We're seeing usage-based pricing, hybrid contracts, and something called a credit. The finance systems built for vanilla subscriptions weren't designed for this. This isn't just a RevRec problem anymore, because outdated revenue architecture can kill a great pricing idea before it ever reaches the market.

11:12RightRev lets you recognize revenue in whatever shape it comes in. Whether you're launching consumption credits or testing entirely new commercial models, RightRev gives finance the flexibility to support whatever pricing model comes next. So product teams can keep on innovating and finance teams can keep pace. If you're architecting revenue for the AI economy, learn how RightRev can help. Visit rightrev.com slash CJ. That's me. That's rightrev.com slash CJ. Hey founders and finance folks, you know how early on everything is fast and scrappy and then suddenly equity gets messy? Spreadsheets break, grant docs pile up, lawyers are sending new forms and collecting fees.

11:53It all shows up right about when you're already swamped. Well, if managing your cap table feels like one more frustrating thing on a very long to-do list, you need to know about Pulley. Pulley makes equity management simple and stress-free. You can issue options, model dilution, complete 409As, and more. All in one place with support from real experts when you need it. So if you're raising, hiring, or scaling, Pulley keeps your cap table clean so you can stay focused on building. Not wrangling spreadsheets. Learn more, request a demo at pulley.com slash mostlymetrics. That is P-U-L-L-E-Y dot com slash mostlymetrics.

12:30I got news for you. The ERP category is finally getting disrupted. And if you haven't heard of Rilla yet, please pay attention. It's the AI native ERP built specifically to replace NetSuite. and it's already won over hundreds of finance teams. Their mission is to make these zero-day close a reality, and they're actually doing it. We're talking teams closing the books at 1.35 p.m. on the first day of the month. Companies like Windsurf, Mercore, and hundreds of others run their entire finance stack on really. Revenue recognition, close management, multi-entity, native Stripe and Salesforce integrations, woo, everything a scaling company needs.

13:05They've got 5.0 stars, 5.0, wow, 5.0 stars on G2. They're backed by A16Z and Sequoia. Heard of them. And CPA-led implementations that get you live in 45 days. That is simply unheard of in the ERP space. If your books aren't running as fast as your business, check out Rillet. Book a demo at Rillet.com slash CJ. That is R-I-L-L-E-T dot com slash CJ. That's me. Rebecca, talk to me about product market fit. I was talking to Kurt earlier today, And I didn't realize that you helped incubate tabs when you were at primary. I think it was a, I would say, slightly less ambitious idea going after just robotics at the time.

13:46Like, that's incredible. I think we started with a very operational focus. And building was always central to that. And usage-based building was always central to that. And actually, I think the, I'll call it a product market fit journey. I think similarly, it's not a milestone. It's a lifelong journey. And in that, we realized that the way that we were tackling the problem around complexity and billing was very centered around what data is required to be able to send an invoice at a complex organization. It's not just CRM data. It's the contract, the PO, the amendment, the renewal. How do we put together a complex knowledge graph of all of that commercial documentation and use that to power billing?

14:24Certainly, it's definitely still part of what we do, but even more powerfully, revenue and reporting that is much hairier to do requires a lot more domain expertise and has allowed us to go really deep with accounting teams far beyond the operational side of billing tasks. And so I think our journey has taken us deeper into the finance and accounting organization, deeper into sort of our respect for that domain. And as a result, to larger companies, more complex companies and upmarket quite a bit. I think it's safe to say all three of you have found product market fit. It is an evolving target, as you've all pointed out.

14:58Would you be comfortable with just sharing the directionality of the business? Like how fast are things growing? Like, Fred, I don't think you've slept in a couple of weeks. You can give me anything here. I just want to know the size of the bread box. Could be employees, just year of year growth. How are you feeling about things? So I'll start with you, Fred. Three of us last March, April, May. Now fast forwarding one year. We're about 39 of us today with offices across New York and San Francisco. I can't comment about revenue at this point. I know that there's some recent bits and pieces in the press that I just, I appreciate people are getting excited about fleet, but when we're ready to say, we'll disclose it.

15:33I think what gets me excited though is that, you know, I used to work at a fairly large bank in a division that did a lot. And I think that we're comfortably well above that in terms of run rate at this point. So it's very exciting. I did not leak those numbers to the information last month. Raul, how about you? So one way to conceptualize it is around this time last year, we were just shy of 40 employees. I think it was like around 35 or so. We just crossed 140. If we can find enough engineers, as discussed on the last panel and among some other profiles, we should be well above 250 by the end of the year.

16:06Okay. Rebecca, how are you feeling about things? I need a few more margin read us. We are hiring for engineers, product managers, forward deployed folks, customer success, implementation, sales, marketing, BDRs, partnerships, people ops, talent. I want to dig into that because I feel like a lot of people in the audience are having constraints around hiring. And from the outside, you may say, oh, we just need more engineers. We need more people who know about AI, who can code or at least use the AI to code. I actually am hearing from people that the constraint is onboarding and the ability to get customers set up.

16:43I'm not sure if you're feeling that pain. Someone told me once it's like good when the problems are growth problems. Especially for us, we are hiring incredibly thoughtful domain experts for the most part who have real depth of experience in systems migrations and ERPs and accounting. And this is not a dime and dozen SaaS generalist role anymore. And so as our business gets more complex and our customers get more complex, the type of profile we're hiring for gets more complex, too. And we have lofty, ambitious, and we want to keep the bar really, really high. We have to be incredibly, incredibly dynamic in market to find that talent.

17:21What is a forward-deployed engineer? In the traditional sense, I feel like it's just a rebranding of a consultant in a lot of ways and places. But I think in like a purist sense, it's someone that should really be spending time embedded with a customer, working with them, understanding exactly what their needs are, exactly what will actually get whatever their objective is across the line and do it using your product. Like we're figuring out exactly what the term should be, but we are considering ourselves again, like forward deployed researchers. We'll go in and actually sit down with researchers at the AI labs, partner with them on projects, structure it, and then figure out how to execute it.

17:55It means that we're doing a lot of everything from engineering to data work to our own internal research initiatives. And day to day, it just means we're spending a lot of time with our customers. Just to go a little bit deeper on that, because I think that's the team that you run in addition to doing the payroll and cleaning dishes in the office. How does this differ from traditional customer success? It's a lot less account manager focus. It's a lot less account management focus, rather. If a researcher comes up with an idea or we go to them proposing an idea or we show them a benchmark that says, hey, your models have some place that they can improve upon, we can then build the software, populate it with synthetic data, represent an entire enterprise workplace environment in our simulation environments, and then develop a training regime or eval regime that can actually push the models forward or determine exactly where their capabilities begin to fail.

18:42I tend to think that goes a little further than a traditional kind of customer success or engagement role. And that's, I think, where the magic is. It's actually getting in and doing the engineering work and doing the research work with the client. Raul, what's been the constraint for you other than everything? There is a little bit of everything in terms of the normal growing pains of a startup, so I'll corroborate that. But for a long time, and this ties to our conversation about PMF as well, the blocker for what AI can do in a workplace, including a financial firm, was the model intelligence and the adjacencies around that model intelligence.

19:17For example, something as simple as tool complexity or context complexity for a model, as recently as six, seven months ago, was a major constraint on the kind of work and the kind of ROI, therefore, an application, AI application could deliver. The model intelligence is no longer a blocker. And then the blockers for growth, including our growth, growth across the application layer is everything else. It's you show up at the likes of a JP Morgan to pick on them. It's the myriad PNL and workforce issues they have to sort through and thinking about tokens in addition to headcount or as a substitute for headcount.

19:55There are blockers around workforce enablement and transformation and getting these, you know, crusty 50 and 60 something partners who are used to doing something a very specific way and getting them to move to a new regime. Those are the blockers. And so that's as much about how we organize ourselves externally as a business, because if we can help get there more quickly, we're going to sell a lot more agents. We're going to sell a lot more of our product. How do you think the traditional go-to-market motion has changed when it comes to hiring salespeople? Are you looking at this like you can just throw bodies at it in quota or is it completely different?

20:30You need the conventional enterprise SaaS playbook instincts and skills, how to navigate these large complex organizations, knowing when to do qualification and discovery, all of that. Combine that with deep product fluency and intelligence at a time when the products themselves are changing by the week. And also, frankly, the ability to sell partnership. And that's hard. It's hard to have a set of reps that spike on all three, much less have the ability to show up and deliver a suite of capabilities like we were talking about. So I think we're all figuring that out. But I think that becomes the evolution of the playbook.

21:07Hey, thanks for listening. We'll be right back after a word from our sponsors. Maximor, get your Maximor here. Sorry, I just came from selling Cracker Jacks at the ballpark. Everyone in finance is adopting AI right now. The problem, it's a dozen disconnected tools. Yeah, one for revenue, one for AP, one for the clothes. and each one needs training and documentation. And the expertise to run them lives in your head, not the systems. MaxiMor takes the opposite approach. One autonomous finance platform for the whole operation. Order to cash, procure to pay, the close, cash management, and reporting.

21:43It all runs on a unified finance context that pulls from your ERP, billing, banks, even email and Slack. No rip and replace. MaxiMor runs on self-learning agents that learn how your team already works. No prompts to wire, no workflows to build, nothing to document. Every output is audit ready, and when a call needs human judgment, it escalates for review. Because the agents optimize, not just execute, and your KPIs keep on improving. One PE-backed customer posted 98 % of transactions straight to their ERP. No error rate. The other 2 % go to a human before posting. You pay only for real outcomes, not seats.

22:23See it at Maximor.ai. That is M-A-X-I-M-O-R dot A-I. Listen, I'm no mathematician, although that is what my mother-in-law tells people I do for work. My son-in-law is CJ. He's a mathematician. If your finance team expands linearly with your headcount, your leverage is broken. Too many CFOs waste incredible talent on low-value maintenance, manual reconciliation, chasing missing receipts, and policing$30 software subscriptions. It's the expense police, run! That's a systems problem. And that's why you need Brex. Brex is an intelligent finance platform powered by AI workflows that handles the heavy lifting automatically, which is a fancy word for without you.

23:08Instead of burning elite finance minds on administrative maintenance, Brex gives your team the leverage to focus on momentum and growth. Thousands of companies, including Anthropic, Coinbase, and DoorDash, already run on Brex. Stop asking your A-level finance talent to do B-level admin work. It's time to get Brex. Head to brex.com slash metrics. That is brex.com slash metrics. Today's episode is brought to you by Anrock, the sales tax platform behind companies like Anthropic, Notion, and Vanta. Here's a fun way to totally ruin a Tuesday. Open a letter from a state you've never set foot in telling you that you owe back taxes you didn't know existed.

23:47Happen to me. because the rules never stop moving. States are now racing to tax AI, digital ads, streaming, really anything new, and they're doing it faster than a spreadsheet can keep up. Anarok handles all of it. One platform that watches your exposure everywhere automates compliance and flags risk before it turns into that nasty gram. That's why thousands of finance leaders trust Anarok to stay ahead. Talk to an Anarok sales tax expert for a personalized exposure estimate at anarok.com slash RTN. That is A-N-R-O-K dot com slash RTN. Rebecca, how has your go-to-market motion changed in the last year?

24:26The composition of the team, playbooks that you're using, is it different than what we were seeing three, five years ago, or even different than you're seeing three months ago? I think ours has changed in a couple of material ways. As we've gotten more sophisticated, our product has gotten more sophisticated, and our buyer has too. I think all of that has required a much deeper motion than we were used to, maybe a year ago, even six months ago. So we've built out an exceptional solutions consulting team to pair with our salespeople to provide that product knowledge, domain depth, really customized solutions that are highly fit for purpose for that individual customer and all of their quirks so that we can support edge cases and be really good partners to them, making sure that the AI works for their use cases, but also all of the integrations and workflows that need to be possible such that the AI can do its thing.

25:15That has created more complexity and part of that's going up market, but part of that is also about meeting the moment where the market is asking us to come with product expertise and domain expertise and really consult on where their systems and their processes should go. We're going to go down the line here. So the question is, how would you describe your pricing model today? The second question is, how many times has it changed in the last year? Raul, I'll start with you. We've historically been classic success. Classic subscription per seat, you know, typically annual, sometimes multi-year.

25:47And I think we all recognize that that's changing. And so it's less in my mind about how many times it's changed in the last year and more how many times it's changed in the last two or three weeks. So stay tuned. Take us into that. This is both a moment of first principles thinking in each of our sectors and application areas for how do you price and package? And there's a little bit of the pattern matching or reasoning by analogy with where we've seen usage or consumption-based models, coding obviously being one, cloud services being another. Startups that take those models wholesale into sectors like invoicing, billing, or finance, or legal, or what have you, are very unlikely to be successful, or at a minimum are very unlikely to be the leading firms in their segment.

26:38So there's a bit of R &D, quote unquote, that's a commercial R &D, not just engineering R &D. Fred, how the hell do you price what you sell today? There's a lot of bespoke work, so everything kind of gets a bespoke price tag, depending on what it is. We sell software, we sell infrastructure, we sell hosted evals. Traditional software typically is driving what we're doing. We have data that's part of all this as well. We have infrastructure for anyone who wants to run evals or reinforcement learning training runs. They can do it with us. If you're running evals and you want to run everything through Fleet, we'll sell you the tokens.

27:13We can do usage-based pricing as well. So we have a bit of everything. I think that we'll have to come back to this conversation as we continue to scale this up. It's an interesting point in time where we are trying to make sure we have a very nicely structured playbook for everything. Because at the end of the day, we want to make sure that we're very easy and reliable for all of our customers to work with. and that we have a quick answer for everything. So that's one of the things that we were working on in San Francisco. But just to stay with you for a second, we used to nerd out about how SaaS companies would price in the enterprise versus mid-market when you were covering cybersecurity back in the day.

27:45This is a completely different world. We have dozens and dozens and dozens of different types of simulation environments we built. Each of them today carries a pretty standardized price. It's a software-based sale. I think the difference from what I've seen is it's not like a SaaS sale where we're able to go through and say, this is three different SKUs and we're going to have like some or some value metric price. We tend to be a bit more like outcome oriented. And that's how we're beginning to think about it at this point of if you're able to see some sort of major gain in your model, we'll be able to figure out the value that is associated with that.

28:18Rebecca, what's your pricing model today? How many times has it changed in the last year? I don't know if we're on V3, V4, a lot of Vs. I think in that, we're trying to accomplish a couple of things. One is predictability for finance leaders who hate anything that's not predictable. So I think we are trying to speak the language of our buyer. I think we are trying to accomplish that at the same time, making sure that the key places where there is either deep complexity or where we are driving deep value, that we are finding the right way to commercialize for the right customers. So that might be in certain cases driven by integration complexity, custom builds, unique workflows, or other really complex parts of the platform.

29:01So I believe we are really the best solution for anyone who is doing revenue recognition to 606 based on usage commitments. And so we are really trying to find always the right pricing models to accommodate what I think is really best in class software in our space. And then beyond that, I think trying to figure out how we commercialize our forward deployed motion. I think something that I heard from all three people was that you're trying to do right by your customer and meet them where they want to transact. You brought up the finance persona. You brought up the people running these large language models.

29:31You're dealing with investment bankers. You have to crawl, walk, run and package it in a way that they're used to and what they're willing to pay. I want to dig into the finance function. Maybe I'll start with you, Fred. What does the finance and accounting function look like today? And how many people is it? Should be more people. It's been a bit of me, a bit of another one of my colleagues with a background in finance and a whole lot of QuickBooks, Excel, Stripe, and everything you use to run a small business. But I'd argue, I think we've graduated above that and we're in the midst of upgrading everything.

30:04For payroll, we just transitioned from Gusto to Sequoia, checked out Rippling and others. I think I can keep going from there, but it's a small business stack still at this point. If you need a deal on cash to collections, go to tabs.com slash CJ for 20 % off. Raul, what does the team look like today? You know, it's basically four of us that touch the function in an active way, albeit, you know, heavily leveraging outsourcing and kind of third party support. We have a head of strategic finance where we are hiring for strategic finance leads if there are profiles out there or in network. We have a controller, which is a tough one to get.

30:37I can share lots of PTSD about that process for those who are interested. We have a modern software stack. AI for software stack tabs prominently in that. I think what we're finding is while the outsourcing is always a bit, you know, check or hit or miss, even in SaaS era, in the AI era, it's even more difficult to do it with high fidelity. So there's a kind of gradual insourcing process underway. The record for largest company in terms of revenue still on QuickBooks, Fitbit made it through 300 million in revenue, still on QuickBooks. I met another finance leader. They made it through$75 million in ARR before moving off a fractional firm.

31:16I thought those were interesting factoids. I don't know if anybody has an opinion on this. How far do you think you can take it with a fractional outsource accounting team? I was alluding to this, and I guess maybe I won't say the firm so as to not lambaste them. In our experience, the outsource firms to handle the complexity, I mean, even in the commercial models we were just talking about here, maybe not, or maybe not quite at fleets level in terms of the range of the menu, but To think about now with more usage-based models, that transition, but some customers still on legacy per seat. And you have these like four deployed elements, SKU proliferation.

31:49It's very challenging for any outsourced vendor to handle that in an enterprise context where the tolerance for errors is not particularly high. If I can wave my wand and have basically all of it be in source tomorrow, I would. We're all beholden to our book of SKUs. Rebecca, what does the finance and accounting team look like today? Team of one. Part of that is by setting up the right system stack early that makes that possible. We have customers on the QuickBooks revenue front that blow that out of the water, north of$500 million in revenue. It's not really about QuickBooks. It's what you're doing in your subledgers that you can then dump into QuickBooks and make that a brain in the way that an ERP is effectively a database.

32:28really powerful workflow automation around it makes that possible. So we see companies in our network who either go all in on fractional and outsource support to make those systems run, or vice versa, very early on, trying to do as much as possible with teams of one and two. So all of you have AI in your products. I'm curious how you wrap your head around the cost and the gross margins associated with it. It's really cool being an AI company. It's also really expensive being an AI company. How do you manage that? That's the reason why the panel before this was VCs. I think everyone's on a growth journey and prioritizing that.

33:01And I think the economic model of all these businesses, everyone's trying to figure out what that's going to look like in time. I know because I see it in our customers pricing models, which started out with a lot of people eating the cost of all the token consumption. And now people are both trying to pass through that token consumption and figure out ways to add margin on top as the models change their pricing frequently. And they're constantly trying to figure out the best way to put that on the customer. And we're seeing a lot of people experiment with commitments that can get burned down across seats and usage and tokens that create a slightly more customer friendly experience with controls and commitment, but at the same time, flexibility in how that gets used.

33:37Fred, does gross margin matter and why? Oh, deeply matters. I think from day one, we were actually thinking about how to price these wildly bespoke products with a really solid software gross margin. And we've actually stuck to that. On tokens, we are massive users of all things, all types of tokens. You're token maxing? If we want to talk finance and jump scares, then we can start talking our Claude bills. I'm trying to get us a couple of plaques and some handshake from Dario, and I think it might happen soon. I've been told that we're one of the largest consumers amongst them. We have traditional software R &D work, which is invested and actually goes towards the product that we're selling.

34:12That's all great, very easy to price, very easy for us to consider in terms of the costs of goods sold. Where it gets tricky is with hosted evals, where we are reselling tokens. That's a place for us where we're not trying to extract software-like gross margins necessarily. It's a product where everyone knows the cost of a token from the vendors. That's where we'll think about that more as like a cost-plus kind of model. We see this with a couple of vendors in the market too. Like if you're looking at an open router type of a vendor, often you'll see that they'll host and then charge you for tokens plus some margin on top of that.

34:44I think that's how we're thinking about your token consumption. But for everything else, we're doing pretty well so far managing to a margin in the midst of token maxing. It reminds me of that scene in the big short when he's like, I think they're confessing. He's like, I don't think they're confessing. I think they're bragging. That's like what you see with these AI companies putting up the leaderboards. It's like that system engineer at Duolingo consumed that many tokens. It's crazy. Raul, how are you thinking about costs? Yeah, maybe this is the inveterate banker in me. So don't try this at home.

35:12When I think about our North Star, I mean, I agree with others that, you know, kind of conventional best in class SaaS metrics at a point in time are not super helpful. What is the terminal value of a business like Rogo being three to five years out? And what do we look like in that world? The weights on gross margin today, the components, the major components. It's inference we were just talking about. Content, you know, for us especially, you know, servicing all these financial data sources to our customers. Embedded support, our four deployed bankers in our parlance who show up. Bankers expect to get paid.

35:47But you think about really those as being the three big drivers. And you kind of tick down the list and you think, wow, in three to five years, our market positioning and ability to manage those costs looks very different. I would be surprised if in three to five years you have one model provider that has premium pricing and a kind of commanding market position over the others. On content, as we get bigger, we're able to amortize our content costs over a larger base. We're able to have leverage over our content suppliers. So that looks different as well. And then you'd like to think that a lot of the embedded customer support advisory type work also scales sublinearly.

36:25So when I added all that up, I think, all right, well, I'm managing to something that in three to five years looks maybe not quite SaaS level, but a lot more like SaaS. And that's the North Star. In the meantime, invest for growth. I do think a lot of the SaaS companies of yesteryear, call it five, eight years ago, they had what I would call big arms and skinny legs. They had amazing gross margins, but their OpEx was nuts. It was off the charts and they were burning a lot of money. I'm wondering if it becomes the opposite where companies start doing leg day, but the biceps aren't as big as before because it's just a different profile.

36:56But at the end of the day, like you said, the terminal value is predicated on cashflow. So how you get to it may not matter as much. I don't know if you have thoughts on that, Raul. I think there's a real possibility. I mean, I try to have a certain amount of epistemic humility. One way to frame exactly what you're saying, CJ, is there are probably users of the likes of FactSet and CapIQ in this audience. And maybe I just triggered them by invoking those pieces of software. But nonetheless, if you look at like classic per seat costs for those tools, it was in the few thousand dollars a year, depending on the size of your firm, etc.

37:28It's probable to me that the equivalent token spend, you know, and we may have a different concept of seats and headcount in a couple of years, but the equivalent probably looks more like 10, 20 ,000, maybe more for a period of time. Like, let's see, even if you have a lower gross margin and your gross margin is I'm making this up 30 percent on 10 ,000 or 20 ,000, you know, you have maybe a third to a half of the contribution margin coming in. But it's on a base that's, you know, three to five times higher and you actually come out ahead. In the end, if it's all about cash flow, you have a higher value, you know, if you have a higher equity value business.

38:05Brad, do you think that over time we'll be spending the same or potentially more on tokens per employee than what their OT is? This is going to be so subjective. I think that from what I see amongst my peers and amongst software developers, it's so easy to just start spending and spending and spending and spending. And then you don't even notice it. And then all of a sudden you spend north of your salary. There has to be a rationalization of that at some point. I have to think CFOs are going to start looking at the bills. And I tend to think that we're probably going to see some sort of a rationalization and digestion here of some of the token spend.

38:38But on the other hand, if there is real value, like I for one see tremendous value being realized from investing money into tokens right now, I can make the case that token consumption will continue to rise, perhaps not uniformly at every single business. But amongst those businesses where there's a real use case, I would certainly expect it to rise. Rebecca, how are you thinking about investing in costs? Because in order to compete, you have to still build a lot of the stuff that people love and need in their workflows, but then you're adding AI on top of that. How are you managing that? Yeah, absolutely.

39:08I mean, I think it's both something that we think about in terms of where do we invest people resources, like at the company? How do we split up product and engineering focus? How do we split up post-sale focus? Those are like internal investment decisions that we're making on how much do we internally need to focus on core product and integrations and security and all the things that people care about even before we get to the massive investments that we're making in AI. What we're going to do next is hop over to a segment that I like to call asking better questions. And this is brought to you by our friends at EY.

39:43There have been some spicy debates going on on X lately and LinkedIn. I'm terminally online and people are arguing over what counts as ARR. We have context on all three of your pricing models, your business models, how you monetize. Rebecca, I'll start with you. When you report your revenue figures, do you call it ARR? What are you reporting it as? Yeah, I mean, I think we are in a lucky position in that we are getting contractual commitments to year or multi-year deals with structure around it. So in that sense, I think we have a lot of defensibility on how we think about ARR. But I've seen some of the asks from our customers on how they want to do ARR in tabs.

40:21And so I know well how... You've seen some stuff? I've seen some stuff. It's crazy out there. You can have an underwater basket weaving company and just call it ARR if you saw enough of them. Fred, how are you thinking about revenue and how you report it? I have a lot of thoughts about how you can report revenue. I was a software analyst in a past life and I've seen a lot. CJ, you were never a bad offender. When we at Fleet get to a point where we report revenue, I'll have a lot more to share about what we're doing. Philosophically, if something's usage based, if there's actually contracted revenue or backlog, call it like it is.

40:52Emily was saying on the last panel, somebody just took$100 ,000 from one day and multiplied it by, I don't know if it was$365 ,000 or something, but that company, I think, just raised a big round. Well, of course, it's always the case, as we were just talking about, the real both checks and enablers of this are the capital suppliers here, our VC friends, if the last panel is still around, hold that crew accountable. We try to self-police, to think that we're pretty responsible about what's truly recurring, what's a pilot, what's a paid pilot, and how to distinguish between these. You do get gray areas.

41:25A, we try to make sure finance holds the pen on what counts as ARR or ARR. That helps a lot. You can't have sales define that concept, needless to say. Try to keep it consistent, write it down and hold us to it and don't make exceptions. I have a friend who's a VC and they were trying to get into this really hot deal. It was like five people in a garage, but they were at, call it, 50 million already. It just kept showing up over and over. What can we do to help? What can we do to add value? And finally, the founder turned to him, handed him a box or the shoebox, said, we have all these checks from our customers and we've just never cashed them.

41:57I say this because you're all making money now. It's working. You have product market fit. What's the hardest part or the biggest bug in the cash to collections process that you're experiencing today? Rebecca, I'll start with you. Come on, man. You're giving me a layup here. I got no problems. Hardest part is that we're not yet using tabs, carving out time and remembering to send the invoice. Oh no, Fred, I got you, man. This is going to sound very on the nose, but it's true. It's revenue recognition and invoicing. We're primarily in the enterprise and our billing and invoicing process up until, you know, recent developments and software upgrades, etc.

42:33is having our outsourced service provider generate an invoice and then some poor human being on our ops team has to email that invoice in a timely fashion. As you can imagine, that's not an especially high-fidelity process. And so automating that is going to increase our collection cycle time quite a bit. The number one reason why collections are delayed is because people don't send invoices. People are highly focused on the post-invoice send collections process, but actually people just don't send them on time or don't send them at all. All right, gearing towards a close here so we can get people back to the bar.

43:08Statistically speaking, there's someone in this audience who is going to hit this rapid growth period that they didn't expect, much like all of you have been able to achieve in their riding right now. If you could pick out one piece of advice for them, maybe we'll focus on the operational side of things just to be more tactical. What would it be? Being forward thinking about a payroll provider actually is very helpful. because once you get to a certain scale where suddenly you need a PEO and you need to actually have, you know, payroll function, I think that would have been one of the best things to do is perhaps find something beyond gusto a little bit earlier for us, at least.

43:43And that would have been very helpful for a fast scale up. This is operations and strategy and, you know, in some ways all encompassing. Be AGI pilled. I got this piece of advice from a friend and that friend's been consistently right. If you bet on model intelligence continuing to compound to borrow the Dario line, country of geniuses in every data center, that shapes how you think about product. That shapes how you think about, you know, what your customers' needs will be. But that also shapes how you think about building and architecting your functions internally. You know, six months ago, we were having, you know, human beings kind of look at, you know, Sigma or Metabase or Omni dashboard data and making PowerPoint slides.

44:19But everything is architected around, you know, if we believe that all these firms we're partnering with are going to be AI native and they're going to have the benefit bit of sort of, you know, again, something that looks and resembles AGI, then we should be building our own business around that set of assumptions too. Finance function, but other functions as well. Rebecca, what about you? Hire great talent and have them in the same office, in person. That's awesome. Well, thank you everybody for joining us on this beautiful rooftop and thank you Tabs for hosting us. Appreciate it. Run the Numbers is a mostly media production.

44:48Yelling an intro by Fat Joe. Artwork by Meg D 'Alessandro. Show is executive produced by Ben Hillman. Nothing said on this podcast is intended to be business or investment advice. It's the sole opinion of me, a guy who feeds his dog way too much ice cream and has a history of net operating losses. LOL. If you like this podcast, hit subscribe and give us five stars. It will take like two seconds and our algorithm overlords love it. Drink water, call your mom and have a great day. Peace.