Name: Finance Forward 2025 - The AI-Native Finance Stack (Demos)
Uploaded: 2026-01-22T04:16:06.451Z
Duration: 1 h 8 s
Description: Finance Forward 2025 - The AI-Native Finance Stack (Demos)

Transcript for "Finance Forward 2025 - The AI-Native Finance Stack (Demos)": Alright. Hi, everyone. My name is Calvin, and I lead the product management team at Ramp that's focused on our spend management product. I'm sure we have a number of customers that are in the audience today. But for those of you who aren't familiar, Ramp is a modern financial operating platform for businesses who value their time and money. We started out as a corporate card and expense management platform and over the years have evolved into a suite of finance tools that includes bill pay, treasury, procurement, and more to come. And in the six years that we've been around, we've grown to 45,000 customers and saved them over $10,000,000,000 in cash and over twenty seven million hours in time spent, and we've done this through cutting down on wasteful processes and automating a lot of manual tasks. Now this morning, we were talking about how finance teams can scale through AI and at the same time manage the balance between automation and the right level of control, oversight, and governance. And so I wanted to ground us an example that we see a lot from our customers. How do you catch out a policy spend today? Do you have processes in place? Do you rely on managers? Or do you just hope that employees will follow the policy? What we've heard generally is always that employees never know what's in policy and what's not. They don't know what the policy is. And finance teams are often left manually reviewing expenses every day or every week. And they're just sifting through all of these expenses that are mostly in policy just for the few that are actually out of policy. So to address this, we built an AI agent, that helps you enforce and manage your policy agent. And unlike managers who don't have the time, the willpower, the knowledge of the policy to actually enforce it, they don't wanna be the bad cop after all, Ramp can actually do that for you. And so these agents that we built are experts in your policy. They pay attention to every little detail. They never sleep. And they can adjudicate and administer judgments in an impartial and consistent fashion at large scale. So say you have a 100 transactions to review. In the past, you probably had to review these one by one. But with the policy agent, you can now have ramp review a 100% of those transactions, automatically approve the 85% that are straightforward, low risk, routine transactions that are obviously in policy, and then just leaving the remaining 10 to 15% for you to review. Customers who are using the policy agent today have found that they are finding 99% accuracy in terms of the actual decisions we recommend, and they're catching 15 times more out of policy spend because of the agent. But, enough chatter. Let's actually look at what the demo looks what the product looks like live. Alright. I'm gonna zoom in real quick here. So this is the Ramp dashboard. It is the one central place for finance teams, managers, employees to come look and see the state of their finances and what tasks are waiting for them. And you'll see that on this homepage, Ramp has already grouped transactions into ones that require review from me and ones that are pretty safe to approve. And before we dive into those individual transactions, I wanna first show you what's kind of going on behind the scenes here. This is a policy center. Within the policy center, we've actually created a home for your policy to live in a digital fashion. It's easily consumed by the AI agent, and it's also a place where you can continue to maintain it, share it with your employees, but also add additional guidance, for your finance teams and for the policy agent to be able to enforce behind the scenes. With that in mind, the policy agent then interprets all of that and then applies it to every transaction using the context that we have. So let's jump into a couple of the transactions here. So this is a pretty routine looking dinner expense, $276 from Lacora Bianca. And you'll see that the policy agent has actually gone ahead and said that this requires my review because the transaction amount exceeds the $50 meal allowance. And it's also looked at the receipt here and found that there is a guest count of four. So this might actually be in policy. Maybe there's an attendee missing or it's a client dinner. But I can now easily just send this back to the individual and say, were there attendees? Please add them. Pretty straightforward. Let's look at this example. Air b u n Airbnb expense, $85. You'll see that this is actually for a gift card. And the agent has recommended that you should actually reject it because according to the policy, gift cards are explicitly never allowed. And it actually shows you a direct citation in the policy of the clause that states that that's the case. So whether you are an employee or manager or an admin, you actually have the direct reference or citation to the policy document that's actually governing this decision. Also, it's noticed that the receipt here is a little bit fishy and it looks like it's AI generated. So we're definitely gonna reject this one and request repayment. Lastly, this is a fun one. It's the $36 Uber transaction. Seems pretty routine until you realize that this is actually to a club which may not necessarily qualify as work appropriate. Now this is not a use case that we programmed into the agent. It is the fundamental emergent intelligence of these models and what they're able to do. But it looks at the Uber route, it looks at the information and makes this interpretation and tells us that actually this seems a little bit suspicious. So we're also gonna request repayment here. No club visits. Alright. Because the policy agent has done such a good job, I trust that all the ones that are actually in policy, can just approve in one click. And so now you save a whole ton of time every day managing expenses. For those of you who are into dashboards though, we do have dashboards and it helps you monitor your employees' overall, adherence to policy as well as the actual performance of the agent. So you can see where the hotspots are in your organization. You can also see where your policy is falling short. Are there certain trends that tend to trip people up? Are there pieces of your policy that are actually unclear? And now Rep will actually go ahead and make recommendations about specific language to add to your policy so that it's clearer for your employees and for the agents to be able to adjudicate. And lastly, I showed you sort of a contrived example because you can actually go and you're still manually reviewing in the example I gave. But if you really wanna fully embrace the policy agent, you can actually set up these workflows then that automatically approve, automatically reject, and therefore you skip the whole set of steps that I just showed you. Awesome. So that's the policy agent. In closing, we wanna help set finance teams free. You have much better things to do with your time than manually review every single expense looking for needles in a haystack. We wanna help finance teams be able to spend their time on higher leverage strategic initiatives that actually move the business forward. And so that's our mission, and that's what we're here to do. So thank you all for your time. For those of you who are not on Ramp, we'd be more than happy to get you on it and get you set up. So feel free to look for me or some other folks afterwards. Thank you, guys. Eric, their CEO for the first time. So it's fun to talk about the future of finance. We now have Ryan from Abacum, and, excited to have you up here. Ryan's worked longer in FP and A than most. So you want us how long you've in FP and A? Oh, yeah. Sure. So I've been in, SaaS for about twenty years and out of that, seven years I've been in the FP and A space. So I joined Abacum because of the product. It's truly amazing. I'm really excited to share this with you guys today. And a little bit about us, we are an AI native FP and A platform for mid market companies. And it's not it's more than just FP and A too. We'll do revenue planning, headcount planning, inventory planning, everything around it. And you'll see the user interface today. It makes it really easy for non finance users to get in, be self-service as well. And so yeah. And we'll talk a little bit about the AI capabilities too. We're, based in headquartered in New York, global company. We just raised $60,000,000 through, Scale Ventures and Cathay back in June, so we're scaling really fast. And we've got offices in Barcelona, London, and probably clients across maybe 40 different countries now. So when you think about companies that are like multinational, multi entity, lots of dimensions, multiple currencies, we handle that really really well. And we work with companies anywhere from like, I think when replic came on, maybe like 50 to 70 employees, it grew extremely fast all the way up to some publicly traded companies, thousands of employees. So we've got really good pricing and packaging to start with lower mid market and then also scale up and good time to value too. A little bit about what makes me makes us different. So you'll see a lot of native AI capabilities built into the platform. Some of it will help you clean your data. Some of it will help you analyze and summarize your data. Some of it will help, detect anomalies. And then other pieces are gonna actually help you forecast and budget. And then we're releasing AgenTik AI, next month too that we've been working on for a while. So a lot to come. And tasks and workflow are pervasive throughout the platform. So again, making it easy for non finance users to come in for departmental budgeting or budget on projects and we handle unlimited dimensionality. So when you're thinking about Campfire and all the dimensions you have in an entities, we'll pull that all in. You can slice and dice the data. And then we also have a really strong ETL layer that allows you to manage the data and clean the data too. So it's very, very flexible. When we when we talk about our AI capabilities, we'll take a look at a couple today. The AI capabilities that are built and functional within the platform today are gonna help you like summarize what's behind your data and will help you maybe write your monthly report analysis, helps you forecast. When we talk about AI classification, ramp was just up here. Sometimes the ramp transactions might put the vendor in the memo field. Well, our AI classifier will automatically take it out and put it into the vendor dimension for you. Right? So there's some really cool agents that maybe aren't as sexy as some of the reporting parts of it, but will help keep your data clean. And then anomaly detection and AI and then also the agent agentic AI, which is coming are the core capabilities that we're working on and focus really really hard on on developing today. Alright. Let's get into demo. I'm also watching my my timer here so we can spend most of the time in the product. Alright. So we've just logged into Abacum. You'll see I'm in full admin mode. So on the left side, we've got everything. If I'm like the head of a department, I'm gonna have a space that's set up just for me, makes it super easy for me to come in and put my budget in or also collaborate with people too. And spaces are more than just your, like, static reporting. You're gonna be able to use it for your board reports. It's cool. There's a client called Cortex and Evan, the head of finance, he runs all of his board reports just right out of Abacum invites the board in to a space whenever he closes the books and everything refreshes. People can comment on it, but then you also have workflows and tasks and stuff too. So again, space is more than just like a dashboard. It's also where you're gonna be able to collaborate with the team. Before I show you a couple AI things, think about like finance being self-service and being able to make quick adjustments. Like if I wanna just update this table, you can see like very easy to drill through the dimensions. And I might not even like this dimension structure. I might wanna see like, I wanna see entities at top and then drill down. So I can just edit the table and come in. You'll see like sales and marketing. There's that row we were just looking at. And I'm gonna edit this. Right? And so I'll say I want subsidiary at top. And after that, I'll say let's bring in department. You know what? I also wanna pull in suppliers to this view. So the system reacts extremely fast. You can pivot dimensions around. Trying to do this in a legacy system, you'd have to go write code. So the flexibility to quickly manage your dimensions and be able to do this like independently across tables as well as within your modeling is a very unique thing. And kinda looks trivial, but it's nontrivial too once you start using the platform. Okay. Now I'm in this board report. I'm getting ready to kinda create some analysis. Some of the first capabilities I can sit just as is the summary. I can hit quick summary on any table within the platform and it'll automatically analyze. This is the expense chart. I can also have it drill through to specific dimensions and analyze those too. I can post this, edit it, comment at people, as well. And then save this, right, for my month end and and publish that to the board. So it makes it really easy to start to drive the summaries, and put them right into the system. Also, as I'm, you know, looking at this data, you know, another thing that's, important to know is like when you think about self-service, just opening transaction level detail, I can drill right through to the journal entries. So whether it's me, like, running finance or if it's, the head of marketing coming in and you want them to be self-service, they can now just be self-service, see what's driving them over budget, comment at you, that comment shows up in Slack and email, keeps it on the cell, you can resolve it. Right? So now just creating a easier way to collaborate across the organization and create ownership for expense and revenue as well. Alright. I'm gonna go into the next next AI capability that we're gonna take a look at, which is in modeling. This is my demo account. So I've got a lot of different models in here. Typically, you'll see things grouped a little cleaner. So first of all, really easy to just build like formulas within the system. It just kinda like it's kinda like Excel, but it also like starts to tell you like what to choose next. So very easy to to do that or I could say, hey, let's just drive AI forecast. This is using machine learning and also learning and also AI to like look at seasonality, look at your historical data, and then start to present a forecast. So you can you can get a top down forecast like extremely fast within the system across your entire model. Also, I like to use this to kind of pressure test some of the assumptions that I put in too to see like, alright, what does this look like compared to last year? Another way that I can also write formulas is just by generating a, like simple text. So if I click into one of my let's go down to this one. And I can say formula and then I can start to just prompt it with human language. Say, give me the last three months average and I'll write the formula and place it for me. So couple examples of how you can start leveraging AI within the modeling. Now the last thing I'm gonna touch on because we got thirty seconds is if we go into transactional level detail, you'll see like we're pulling all the data in. You can start to tag like, hey, I want customer to be a dimension. Right? So now when I bring in my data from Campfire, data from Salesforce, right, or HubSpot, customer from hub from HubSpot or Salesforce equals customer from Campfire. Right? And so now you can start to standardize your dimensions. And then also you can start to transform those too within the ETL layer. One of the transformations can be done by formulas or as I mentioned, we have this AI classifier, which you can create an agent now. This is probably the most popular used piece of AI within the platform that'll correct the the vendors from the memo field to the vendor dimension or also catch misspellings in vendors for example. Alright. That's it. Thank you everybody. Amazing. Well, is planning season, so it's not too late. Come on up. We have Ian from Anrock. Okay, amazing. Hi everyone, my name's Ian. I work at NROC. For those of you who don't know us, what we do, and Ryan mentioned being sexy just before in terms of features, we do something extremely sexy which is called sales tax compliance. Now the sexy part is we automate that. And part of reason we do that is there's one thing that tends to hold true, the more I chat with finance leaders, which is the faster you grow, the messier your compliance gets, which I think is something everybody can relate to. But to be a bit more specific, let's say that you're at 5,000,000 in revenue. You might be filing sales tax maybe twice in a year. If you're at 50,000,000, you're filing every single week somewhere in the world. And this is where NROC comes in. And this is why some of our customers that you can see here leverage us because we can automate that. We can make sure that sales tax compliance is not handled manually anymore. And as those leaders of those companies are expected not just to report the business, but to automate that business, to predict it. This is where we come in to make sure that we automate that sales tax layer so that you can focus on actual growth instead of tax filings. How do we do that? And why are we replacing some legacy tools? So all of those companies are rip and replaces from, and I won't name any companies, but part of the reason why they decided to go to NROC is we're built for the modern AI era, meaning usage based pricing, micro transactions, AI transactions, which was not what they were built for. In the background, because full disclosure, I'm now gonna show you an MCP agent today. In the background, we use a lot of AI for a couple of things. For tax research, we have LLMs that automate that research. We have also, and maybe I should have put their picture on that slide, an in house tax team. We have our own tax tables, we divide them to make sure that we are as accurate as possible. And we also use AI directly in the product as well. For example, with our exemption certificate portal where we extract all the data and we make sure that we fill out the form properly and we can raise any flags if necessary. So this is what we've built. I would not say just AI, but automation that is reliable, trustworthy, and audit ready. And companies tend to love the fact that compliance just happens, but you don't have to take my word for it. This is what some of our customers have to say about how we fare in terms of ease of use, in terms of accuracy, again, because we're built for SaaS companies, for AI companies. And also, I'm biased here. I lead integration partnerships because of the strength of our native integrations with top partners among some of the panelists here today. But without further ado, happy to show you how it works under the hood. So let's say I'm a CFO. I started a company recently, and I just received one of those notice from the state of Washington saying that I'm exposed and I owe a decent amount of money. So I will go maybe first to exposure modeling, and I will see whether or not that amount is as accurate as they thought it would be. And you can see here that is 24 k. So I'm reassured I can go back to my dashboard to show you how it works. You can get compliant in real time just answering yes or no questions, make it extremely easy even for someone like me who doesn't know a single thing initially about sales tax to be registered in that state, and we'll take care of of that for you all. Well, let's take a step back and see how that works under the hood. So you first connect your data sources. You can integrate your billing system or ERP, you can also integrate your HR IS. And the reason we do that is we natively take into account only economic nexus, but also physical nexus as part of our product out of the box. And that's what populates that dashboard that you've seen. Then typically between thirty minutes to a couple of hours, you will see, depending on how many transactions you have, you will see all those transactions flow in. You will see them broken down by jurisdiction, by date, with the right tax amount. And because all those integrations are native, that sales tax amount will be added to the invoice for your billing system, for your ERP. And that's what powers that dashboard. Couple things I wanna mention here, and you probably have noticed. This is a global solution, so we don't only have a map of US and Canada. We have other geographies where you might be exposed as well, and you can take action in all of those. Another thing I want to highlight which might not be obvious, this is real time information, meaning that we don't sell nexus studies. We give you access to a tool so that you can see live what your exposure looks like day to day. And then the colo code should be pretty straightforward. Red is when you're exposed and you can see why. Here is because of physical nexus. Yellow is when you're approaching exposure. Here because of economic nexus, you're almost at a 100 k and again, you can start getting compliant. And then green is where you're remedying sales tax. And so another thing I wanted to mention as well, because I mentioned it briefly, you can also manage all of your exemption certificates directly in app because this is never necessarily super fun. So Amrock keeps track of all of that directly within the platform. And you can also request directly an exemption certificate from a customer through the platform as well. Finally, and I will end here, you can find all of your reports and break them down by types of information here because as much as we love talking about new tools, I do know that a lot of CFOs still like Excel if they are not using Avocum yet. So we wanna make sure that even though we're the one source of truth, all of that information and data can be used by you all. This is what Amrok does. This is not sexy. That automation underneath is though, I believe. And thank you for your time, and happy to chat after this. Thank you for helping us all think about sales tax a little less. Right? Rebecca Tavs, come on up. Alright. I'm on the clock. So I'm, I'm from New York. I'm gonna talk fast. Bear with me. Okay. There we go. Alright. We are the AI native revenue platform for modern finance teams. Really excited to chat with everyone today. I'm going to move fast to the slides so we can get to the fun stuff. But we are purpose built for finance and accounting teams, a full contract to cash platform powered by AI agents. Our main mission is to help folks save a lot of time and collect a lot faster. Recently raised our series b led by Lightspeed and General Catalyst to help us scale a ton and really excited to get to know some of you in the room today. Hope I didn't need that. We're building the foundation for intelligent finance. So what that means is we hydrate our platform with a variety of sources of commercial data. Number one are your contracts and agreements, MSAs, purchase orders, order forms, renewal docs. And we extract all of the unstructured data that lives in those contracts so that you have a single source of truth to understand everything that happened in CRM, everything that happened in a contract, in emails, and side letters, and we pull all of that together. The second form of data is actually usage data for everyone who is starting to explore or has already started to explore variable models, usage commitments, anything like that. We ingest that usage data from a variety of sources, whether it's your back end, from Stripe, from BI tools, even time tracking data in professional services. We pull all of that in so that we can calculate invoices, and I'll show you what that looks like today. And lastly, if you have a PLG or self serve model, we want to bring all of that together with your sales led B2B contracts so that you can have a real foundation for intelligent finance, which is where our agents start to become really powerful. So how we work, we pull your contracts ideally from a CRM or from a PandaDoc or DocuSign, something like that. We ingest those agreements so that we have all of your PDFs in one place. We extract all the key terms. Think pricing, dates, discount, net terms, renewal language. So we bring all of that and expose it in tabs. We automate invoicing with usage data pulled in so that we can calculate and with partnership with our great friends over at Anrock and Co. And then we help actually reconcile payments, which I'll focus on today, so that when payment gets made, we are auto matching all of those transactions against your open invoices. And I'll show you a little bit about how we do Cash App and Tabs. Lastly, we handle all of your revenue recognition, so those revenue schedules get extracted from your contracts. You always have hands on the wheel. You can make adjustments. But our AI is designed to, out of the box, drive full end to end contract to cash, including all of the revenue schedules. You can see your journal entries and click of a button, push it to one of your ledgers. And we support the ability to close the loop between CRM and ERP. Everything's APIs exposed, so you can push data to tabs. You can take your tabs data elsewhere, really with the vision of becoming an integral part of the CFO tech stack. Now without further ado, I will hop into the platform. All right. So you get an overview page, AR automation, where we are driving insights at a click about the most important things to the business. So we're trying to drive quick insights around collections, billing, revenue, and summary metrics around your customers, GAAP and non GAAP revenue, and all of your collections analysis. As soon as a deal goes closed one, think Salesforce. All of the data from that CRM is going to get triggered into tabs that we can map the customer between CRM, our system, and your ERP. I'll open up an example here just to show you what this looks like. So we give you metadata around your customer relationship, their term, ARR, cash collections. And then we are giving you an easy place to understand all of the SKUs that that customer has contracted with you. And that might be over multiple order forms. It might be an original order form plus an amendment, all of the fun stuff in B2B. And so within that, you'll see basics, platform fees, easy subscriptions. We support that. We support seats. We support usage tiers, burn down of committed consumption, rollovers, true ups, all the fun stuff. And we give you an easy place to understand. Each of your customers may have a completely different commercial relationship with you. And their rate card is defined in their agreement. And so when your auditor ultimately asks you to understand how that agreement changed over time, we want to give them access to the originals. And so that's what we're extracting based off of. You can quickly, at a glance, see all of the invoices that have been created for you. We tee these up in future states. You can see all your draft invoices. You can approve them if you want before they go out. But the idea is that we are helping close the loop and actually create those invoices for you. They get synced one to one with your ERP. We're pulling out key terms from the contracts, things like cancellation clauses that get buried in the sands of time, as well as a lot of the renewal data. So if you have auto renews, if you have notice periods, if you have uplift clauses, all of that is getting extracted. And of course, most importantly, we store the document down so you always have it. Beyond that, we surface ARR metrics. So we are populating across all of your customers things like an ARR waterfall that give you granular breakdowns of type of revenue, whether that's I've got platform and usage, I've got two different SaaS views and I want to model them differently. And then we give you all of the insight around expansion, contraction, churn at the customer level. We're also pulling in any PLG data you might have going through, for example, Stripe. So if you do have a self serve side of your business, we we can pull that in for you to give you revenue intelligence across the whole business. There's also the GAAP revenue side of the house. Controllers, find me later. I'll get into it all day. But on this piece, we are creating all of your deferred revenue schedules. You'll see unbilled revenue at the customer level, at the category level. You can slice and dice it how you like. And then all of this will flow into a closed management workflow that we've got where at the end of the month, we guide you through a checklist workflow so that you can ensure billing is done, your transactions are matched, you can review everything including flux analysis, and then once you're ready to go, one click, push revenue to your GL. So we're trying to make this really simple and close the loop. Great regard for all the amazing ERP products like Campfire, and we're really excited to be working with them. Last thing, if I have a minute. Do I have a minute? Okay. I'm good. I'm talking fast. I'm going to show you what our payments agent looks like. So we're really, really excited about what we've got going on here. So what happens every day if you are a Tabs user is we have designed a Slack agentic experience that comes to meet you where you do your work. So in the same way that you might have a junior employee who would otherwise be doing cash application for you, going into your JPMorgan account or some other bank account and individually going through line by lines, looking at those ACH transactions and matching them up against open invoices, we are bringing that whole workflow into Slack, where we believe we can make this so seamless that you can do it on the go, you can do it without a thought, and not need to necessarily even log into the platform to experience the magic of tabs. So you'll get a digest daily with three different types of transactions. Some batch is matched, where we have incredibly high conviction that the ambient agent is doing its job correctly and we have a high confidence score, and those will get automatically matched for you. You can apply all payments. You've got unmatched transactions. These are things like money movement, interest payments, something from Carta, where you can confirm that they're unmatched and you can set your own rules and logic so it'll ambiently keep learning from changes that you're making in the system so that as you make adjustments, the system will actually learn from that. And lastly, there might be a few things where we want you to resolve it. You can click Resolve. You can always check it out in the platform. But this is designed to be completely integrated into the tabs experience, And we're rolling out Teams too soon for the big dogs. So we're really, really excited about bringing AI agents both in the platform, in query, and in chatbots, and in extraction, but also bringing it into the places where you would otherwise interface with someone on your team or someone at a firm that you work with so that you can have that same collaborative experience with an agent, day in, day out. Alright. Now I'm actually up on time. Thank you all. Amazing. And now we've got Hank from Campfire, our founding solution consultant. Come on up. I used to give all the demos at Campfire for the first two years. And, we hired Hank, and, we won't say where he worked at in the past. It's a bad word now at Campfire. But, he got here and he's like, show me how to demo Campfire. I'm like, Hank, you're a CPA. You've been demoing ERPs for five years now. You're gonna bring your way to demo to Campfire. So he took us to another level, and, we're thrilled to have him demoing today. Alright. Awesome. Well, first of all, you know, for a lot of the folks here that are existing customers of Campfire, first of all, thank you. And you've probably seen a lot of the product demonstrations from either someone from myself or someone on my team. So for today, we actually won't, you know, go from the start to finish what Campfire looks like, but what we'll do is we'll actually introduce some of the new areas that we've built recently into Campfire, including the large accounting model that we've announced with our series b. So incredibly excited to show you what we have today and what we're building in the future. Of course, we have Zach Doyle as well, one of our actual existing customers for Campfire. He's gonna walk through, actual use cases of M Bert that they use at their company. One of the most rewarding areas and rewarding aspects of being here at Campfire is actually talking to all of our customers that have been on the platform for six, twelve, eighteen months or longer, and seeing actually some of the requests that they're making within Ember and some of the use cases that I think John and I have never even thought of within Ember was even possible. So crowdsourcing some of those solutions, we'll see some of those today as well. But this is a fun page of Campfire for the product demonstration. I think a lot of us have seen this before where we have a couple KPIs and metrics that come out of the box, couple reminders about, like, overdue receivables and overdue payables as well. We also build these really great revenue dashboards when it comes to specific industry or category or vertical specific KPIs that show you, like, what's your ARR, contracted ARR, churn rates, retention rates, everything like that. For our SaaS customers here, we have, of course, the recurring revenue waterfall that'll show you, like, new business versus expansion, contraction, and churn. We're even building on aspects and products in the feature that let you even customize what are the formulas that'll go into something like, you know, new business versus expansion or contraction or churn. One of the most unique aspects, and John kinda hinted at this from my last ERP experience that's very unique about Campfire, is that we have a fully built out close checklist within the product. So when it comes to organizing your close, and many of you will be using this very soon, for a close in October is that you'll be able to have, of course, actions in the background. So hitting run here will generate, you know, your journal entries for amortization of all your prepaids across all of your prepaid schedules. It So builds the journal entries for you, generates the approvals for all of those. But something we're actually building with Ember two point o is the ability to have agentic actions within the closed checklist themselves. So actually linking Ember prompts and Ember tasks within the closed checklist to say, hey. Let's do a missing accrual search. And we don't have time to show it today, but it will do in the background, it will look through all of your historical vendor bills, all the ones that maybe come in a few days late, and then look at all the vendors that have come in already and let you know if there's any potential missing accruals. In the future, within, the closed checklist, Ember will actually generate all of these, searches for you and, in the background, give you just a list of every potential missing accrual, every potential missing prepaid. If it scans the vendor bill and it sees a word like January through June, it'll give you a warning that there's a potential missing prepaid amortization, and then it can actually generate the journal entries for you as well. So the generation of, like, what an estimated accrual journal entry would be for, like, a legal accrual too. So some of the agentic actions we're building within the closed checklist. And then, of course, in the background, it'll still be telling you, you know, who this task was assigned to, if it's completed by AI or not, what's the review process, who was it completed by, the link to the journal entry attachments, custom notes, due dates, everything like that within the task. So agentic actions are coming within Ember two point o and built directly into the closed checklist to help you automate your closed process and close a lot more confidently. One of the new areas that I think a lot of us heard was that we are actually building the large accounting model within Campfire. What that's gonna be is it's gonna be a trained, foundational model that's actually trained to just your instance within Campfire. The reason we bring in so much historical data for all of our customers that are live on Campfire today is that we will train off your specific chart of accounts, we'll train off of your specific historical transactions and actually build your own localized model within Campfire. And one of the actual use cases for that is that we'll have AI categorizations. So every transaction that comes in, especially from your bank, we'll parse the descriptions for those and train off your local chart of accounts. So if it sees the word like job post here, it'll make a recommendation to hit to the recruiting GL account or like law firms or airlines or office supplies or SoCal Edison. It's gonna tell you to put it utilities expense as well. And we can coach, the AI tool as well. So if it doesn't need to go to this GL account, it'll provide a list of those recommendation GL accounts. But if we reassign it to a different GL account, we're actually learning off of the back end, off of every single assignment and every single context correction within the tool itself, and it'll actually make better recommendations in the future. So the goal is that, of course, as, like, all your bank transactions are coming in, you're just getting approve, approve, approve, approve, approve to post the entries or to make adjustments and recommendations to the AI tool as well. We even have an AI layer that's reading between, like, all of your open invoices. So if you have hundreds and hundreds of open invoices within your your GL system and you have a bunch of bank payments as well and you need those to match together, we will actually even do suggested matches where we will find a list of, like, all the payments that have been received, and then we'll actually connect it between here's all the the bank payments we've received, and then here's all the open invoices we have. And we have an approval process for that as well and a a learning process every time approve or dismiss is hit. And, of course, you know, dollar value is one of the aspects we do it for, but we're also parsing the actual description and all of your historical matches as well. So between the descriptions of you receiving the payment and then all of the text within the invoice as well. So we'll actually match and make recommendations for those matches too. We have customers that are using this today that have, again, hundreds of invoices or thousands of invoices that just kinda sit outstanding and thousands of payments either that total to the invoice or or the matching dollar value for the invoice. And they're using this tool to actually make a match automatically between the two, saving a lot of their AR teams a lot of time when it comes to the period close process and closing out a lot of those open invoices. So here's two of the use cases we have for the large accounting model that we're training off your historical data, and, of course, a couple more are being announced in the very near future. We are really focused built on building a fantastic, foundational model that's trained off your specific historical data to automate a lot of the, day to day processes and match a lot of the transactions for you. So with that, those are some of the sneak peeks into a large accounting model. And then we also have Zach Doyle as well, who will give a quick presentation on some of the use cases that he's seen specific for Ember. So Zach's one of our customers here, accounting manager at Advisor three sixty, and spends a lot of time playing around with Ember. It's always the best part of of John's week or month that actually going around and and hearing about all the kind of new use cases we have for for Ember AI. So with that, Zach, if you wanna come up. Zach's here from Boston today, so we're thrilled to have him. And we wanted just like a real AI demo from someone in seed. They're just a profile. It's, in the fintech space, wealth tech advisor three sixty, about 600 global employees. And, he was even showing us how he handles audit requests, with Amber. So I'll turn it over to Zach, because he's really been one of the folks really pushing even us internally at Campfire forward on AI. So thanks for joining us, and we're excited to see. Yeah. Thanks. I think, it doesn't look like it's coming up on the screen or should I I pulled this off. Should I just use on my current USB? I think this was our first PC of the day. Oh gosh. While we're working on it, why don't you just talk through, like, how do you how do you get comfortable using AI? I think a lot of us here is just like, do do you check its work? Do you just, like, learn to trust it over time? Like, any tips for the audience on as they're working through this, like, how do you how do you just learn to trust it? I think that's one of the things that, like, we all struggle with. Yeah. No. And I think you said it in very good way where you kinda think of it as another teammate. So even if you're, like, delegating work, you're always gonna check the work. You also have to understand the person. And AI, in this case, being Amber, is a person. So what we really have to do is basically understand where its strengths are and where its weaknesses are. So as I keep using it more and more, I start to lean off on certain areas. Like for instance, has a very good ramp integration, trust a lot of it from ramp, Doesn't hallucinate almost ever. So a lot of that, I'm just able to, you know, allow it to kinda go through, whereas some other stuff might be a little more tricky. Yeah. Yeah. Hank, why don't you just bring up your machine and make do yeah. Want me to just log in? It's working there. And no, this is not their real account. It's dummy data for him. And why don't you put that back on and do whatever you're doing before and whatever worked last time? It's on this one. Okay. So I said it's coming. Doesn't look like this is loading. Oh, there we go. Alright. Sorry about that. Yep. And now you can actually see how many prompts I've run on this demo account. It was just two, and the first one was just too long to show today. So at least you know it works. Hi, everyone. My name is Zach. I'm really excited to be here today. I lead account, accounting at Advisory three sixty. We're a 500 person wealth tech company based in Boston, provides enterprise software to broker dealers and RIAs. And eight months ago, we made the leap leaving NetSuite to join Campfire. Our NetSuite setup was actually pretty good. You know? We had really clean data. We were closing in five, six days, but we kept having the same repeat pain points each month and really just felt like, you know, as you saw with all this great software, we shouldn't be having So those we went to Campfire. And, yeah, what I'm really excited to show you today is how some of those main pain points we've been able to alleviate specifically with Ember. So the example here or the main pain point that we have is vendors that have multiple contracts under them. Right? So whether it's a reseller or the example we have here is Atlassian. They have Jira, Loom, Bitbucket, three different monthly bills that we get, but they all get coded to same dimensions, the same account department and vendor. Now previously, you know, whether it's a month influx doing your vendor or the head of product engine comes up and says, hey. I wanna say my Atlassian spend. We have to go through and basically go offline, go line by line to actually understand that. Now one of the first things I thought I was saw with Amber was how oh, on the whole line. Was how helpful it is actually using ramp data, kinda like I said before. So first prompt is just, hey. Show me how Atlassian show me my Atlassian cost by month. Please use the build descriptions to categorize these costs and display as a line graph with various lines for each build type as well as a line for total expense. And nice shows the plan over here. But, yeah, this is the first output. And I and so what we can see is we have the total expense line up top. We can see the values each month. And then we have our Jira, Bitbucket, and Loom. Now I do wanna call out I didn't prompt it to say, show me these specific ones. Right? I prompted it to say, use the build description. So just kinda like a little, like, caveat there where AI actually does kinda help in other ways. But yeah. So first thing that we can see of this, you know, is that it looks like Loom doesn't have any data from March or April. I know that we've been using Loom for years. That doesn't make any sense. I'm not gonna yell at the AI for hallucinating, but I'm say, hey. Why is there no data for Loom before May? We've used them for years. And then it comes through with this. You're absolutely right. Investigate. Oh, Loom was billed as a different vendor for April. Makes sense. Atlassian acquired them. It was probably some sort of a, you know, cutover thing. But then it goes in next step and just says, let me recreate this chart with complete Loom history. So now we can see a much clearer chart. Right? Still have all the values, and so we can start looking. Okay. Jira, you know, 2,400 a month flat. That's nice. Looks like Bitbucket, kinda like the usage base. But Loom is an interesting one because we have a Loom, Loom license for every single new hire that we have. Right? And I know that we've been increased headcount every single month. So when I look here between June and July, it looks like Loom was flat. So, again, I'm gonna ask, like, hey. Why was Loom flat? I would have expected to see it grow. And it comes back with this. And this is a literal analysis that I've had my AP accountant do a handful of times is going through every single vendor bill, taking down the usages, comparing the costs, and basically showing us our cost per seat. And this just did it right here. Right? Again, very low effort prompt that just kinda shows what the power of it is. So let's go back to that example of head of product nenge, ask for Atlassian costs, how to understand those. Now let's say he says, thanks. This is great. But Jira and Loom are product spend. Those are hitting my engineering department. I don't want that. Can you reclass them? Data reclasses are obviously a very kinda finicky thing in accounting because, you know, coming from audit, I've seen people who are as long as it's in the fiscal year, we're willing to, you know, go back and change the little transactions as well as other people who are saying, do not touch my actuals. Actuals are actuals and you do not get anything. But moving forward, we'll fix it. At a three sixty, we're a little bit nicer. We do close every month, so we don't like to go back and change those. So what we'll do is a year to date true up, right, where we'll move cost by vendor, GL, and we'll move them out of engineering into product. So previously, data reclasses are export transactions, meld them together, create these new entries that have to go in, upload, check, make sure it all works. Now, our life, it's just one prompt. So I just say, hey. I need you to reclass Atlassian spend for the periods March through September 2025. Can you create an entry to book the Jira and Loom spend to the product department and pull it out in engineering department? Please record the vendors and keep the same GLs. This can be one lump sum entry for the whole amount. So this is a little bit of a denser prompt, and this is because it's something I use every month And, you know, kinda like Hank said with the coming that they were gonna put number of prompts in closed checklist. That was my feature because I just copy and paste it from the notes every month. But, yeah, it's a little bit denser, but that's what you need sometimes when you want specific results. So looking at this, it looks fine. I'm not even gonna view the plan because it kinda says its fork right here. So GRF 2,400 a month at seven months. That looks good. Loom, we have the seven values. That looks correct. The one thing that it doesn't call out is anything about the vendors. So obviously, Loom hit the Loom vendors of Atlassian for the first two months of the of the year. So I wanna make sure that that actually gets it correct. So it's the first thing we're gonna look over and looks like it does. Yep. Everything's coming out engineering, going to product. We have the vendors correct, drafts those. So, yeah, now we're gonna hit create transaction. It's October, so we're gonna make that ten one. Right? But yeah. And then we can just literally post this to our GL, and and that's that. So yeah. I think it's use cases like this that kinda make it really important to kinda remember how easy it is to implement AI into your day to day and, like, really show, like, real gains that you're getting out of that. So thank you for to Campfire for making that possible. Thank you, Zach, doing that. Does anyone have a burning question for Zach on what you just saw or just saying AI AI in general before we let him let him go here? How many people are running to? We have me, two under me, and then one in India doing our cost plus sub. Four. Same amount for Campfire and Campfire and Yeah. So we did Campfire and Ramp at the same time. Were able to shave off one. One person? Yeah. I think it's definitely helped us scale more and do more. So I know that was something that they talked about previously. And I think that's how I really view it, right, is that as you try to stay lean, what these are allowing you to do is more critical work and stuff like that. Thank you for doing that. And Campfire is integrated with all the tools you saw today. Oh, we got one more. Krista? This is probably more of a question for you, John, but it's relevant to what Zach was just presenting, and Hank as well. With the LAM that you guys have, obviously, the the quality of the analysis will be based on the data that's in the system and what you continue to teach it as you use the product. What happens if you are implementing Campfire where the data is not beautiful and you have historicals that maybe you wouldn't classify that way going forward, is how does it account for previous treatment of data versus new policies that you're doing moving forward? That's the question. Yeah. This is a fun one because, since every customer has their own micro model sitting on top of our foundational model, literally, is how we can score accountants in a way because you'll actually see the quality of the accounting in the account, and that's gonna, impact the ability for the yeah. To be helpful. The, the quick answer for you, is it trains as you go, Krista. So essentially, it'll maybe start low if you have, like, very different data, but it's still gonna rely on the foundational model to help kinda infer, even if you've got, like, messy historical data or you've redone the the dimensionality in your account, that, essentially, you'll see as you kinda continue to close in Campfire, you'll see the rates and the quality accuracy go up on the on the accounting model. Yeah. We can question is, can you just start at a certain point in time from a model training perspective? And because we built it ourselves, yeah, we're in full control. Yeah. Of course. Oh, yeah. We can keep going. Go for it. Patrick. Yeah. Again, probably, John, for you. Is there any thoughts to building, like, a customer network so, like, we could engage directly with Zach on some of these questions instead of having to route through Campfire to get the connection and and kind of build momentum that way? We do have an accounting community. There's, like, probably three, four hundred folks in a Slack channel. Highly recommend you to join. It's called Beyond the Books. You can find it on our website, and that would be the best way, to connect with folks like Zach. We try and do webinars and more content and then, obviously, days like today. So go talk to Zach or your peers at lunch and dinner and trade contact info and go create your own AI working session. Amazing. Well, thank you everybody. We're gonna move on to the next session.