Name: How I Vibe Coded a CFO Metrics Dashboard
Uploaded: 2026-08-07T18:30:11.805Z
Duration: 43 min 40 s
Description: How I Vibe Coded a CFO Metrics Dashboard

Transcript for "How I Vibe Coded a CFO Metrics Dashboard": Hi, everyone. Welcome to our webinar. So if you wanted to go ahead and put into our q and a where you're calling in from today, I am calling in from San Francisco. I'm calling in from Plymouth, Mass. Oh, Plymouth. Okay. How far is that from Boston? It's about an hour on the train drive, but it's. worth it to see the Plymouth Rock. I'm gonna have to I'm gonna have to go and see that the next time I'm there. Montreal, LA. Oh my gosh. Really strong Boston cohort here. Everybody must know Emily already. Anyway, we're gonna be starting very quickly here because we just have a ton of people, a ton of content to get through, and we wanna make sure that we have enough time for it. So to kick things off, my name's Kat. I'm the controller here at Campfire. I lead accounting in house, as well as discussing product. So, I want to just let you know that this is the first in our webinar series of how I built this. So in July and June, Campfire actually hosted three HackLabs across North America. And the feedback we received was that the show and tell portion helped people gain a lot of traction in using AI. And because there's a big difference between talking with Claude, talking with, Chad GPT about what to do, we actually saw a big lift when peers started sharing with one another. And that's why we are here today to help empower you to build and share what what we've been working on. So for the How I Built This series, it's meant to break down those barriers and build together, which is why I have Emily here, one of my favorite people and she has really great content. So, Emily, would you introduce yourself? Yeah. Absolutely. So I'm the director of finance and accounting at Cloud Zero. I have been my scope is anything with a dollar sign, so a p a r all the way to being responsible for our financial plan and closing the books, etcetera. So in my career, I've done public accounting. I've done the accounting side. I've done the FP and A side. I've done investor relations. And I love my role right now because I get to do kind of everything and then also get to speak to people like yourself and solving problems. Love it. So, Emily, I am really excited today because you promised me some AI ROI. So if you wanted to just, like, jump in, get started, I know that kind of the plan for today, folks, just a rough agenda is for Emily to share what she's been working on, explain the tools that she's used, and, we're gonna have some time for q and a. So if anything comes up along the way, put it in the chat, and we're going to address those kind of, you know, where they are appropriate to pump in. You know, if there's, like, a quick question on, like, why did you do it that way or a larger, broader, we might save until the end. So feel free to use the q and a. And if you just join, please add where you're calling in from. Cool. So, actually, at Cloud Zero, we do a similar kind of internal session where we do a Thursday build that's called good way to wrap up the week. But, essentially, we all share what we've built sort of like the series, and we say, you know, what was the problem that you were trying to solve? How did you do it? Show and tell. And then, how can this help? So for me, the problem I was trying to solve in what I was looking to build is our financial fiscal year ends January 31. So our quarter end for q two h one was July 31. So about the second week of July, and I'm sure a lot of other finance folks in this call can, relate, it started getting a lot of questions. Where are we trending? What is our margin looking like? Where is our revenue at? What are we expected to spend for AI? What are we looking at q one versus q two? What does that look like? And it's a different flavors, a lot of different stakeholders asking a lot of the same questions. And so I've been challenging myself and my team to really think about, okay, when we are getting these repeatable tasks or predictable tasks, right, how can we automate? How can we utilize the tools that we have access to to really, you know, speed up some of that analysis to give who needs to make those decisions kind of real time insights. So that was my thinking and what I started to do. So I'm gonna just jump to my finished product, and we can talk through. So I will preface that this is my the IR in me that this is completely fake data. It's not cloud zero data. But I did just wanna highlight that just in case anyone from cloud zero is on this call that they it's fake data. But yeah. So here's what the end sort of product looks like. Right? And really shows some of those metrics. And how I pulled those metrics were from the sources that I trust. We are a Campfire customer, so I wanted to get data that I knew and was correct in our ERP. So we're talking revenue. We're talking different expenses. We're talking salaries, etcetera, things that I know and can trust. But I also wanted to relate that to a lot of the cost data that really is important for things like margin, for things like ROI conversations, in AI spend. So what I was able to do was actually connect two MCP servers to those and understand, what was really, you know, kind of what our metrics were looking like from those two data sources. Just gonna pause in case there's any questions, Kat, or anything so far. No. I think that this looks great. So are all of these metrics coming in from your ERP through an MCP? Because you you mentioned that this isn't clod, and, it looks kinda like clod to me. So, how much of this is did you have to, like, pump in through maybe, like, a CSV upload, or is it all coming in through MCP? So MCP, I think of and this is if there's any technical people might scream at their computers, but I think of an MCP as almost like a an app on your iPhone. So when you go to the App Store and you see all the different apps you can download on your phone and use, that's my kind of probably mediocre way of explaining that. But so when I say MCP, what I'm doing is I'm going to Claude, and I'm saying, hey. Connect to Campfire. That's our ERP. Hey. Connect to Cloud Zero. Cloud Zero has an MCP, but all there's a lot of other MCPs you can connect to. We also use Ramp. You can connect there. And so we essentially connect all of and if you have an IT team, talk to them about enabling different MCPs. It's, I think, as easy as them switching a, you know, a switch for that if there is an MCP available that you'll just see it in cloud and say, like, oh, I could connect to that. So what I ended up doing was connecting those, authenticating those, and then you can, in in cloud, simply say, hey, cloud. Pull revenue from the Campfire MCP. And then for COGS, Cloud Zero, we specialize in ingesting data from cloud and AI sources. So I said, hey, Claude. I already have all the spend in the Cloud Zero platform. Connect these two. So pull this data from here and pull that data from here and connect. So on these metrics, revenue from Campfire, cloud COGS from Cloud Zero, gross margin. You actually can tell, Claude to calculate it if you wanna do the formula, but I said gross margin, and it knew what I meant. So sometimes Claude is smart and can do that. Sometimes it's not, and you have to explain margin is revenue minus COGS. And then the percentage, same thing. And then AI spend, we also show in Cloud Zero. So I was able to say, please pull AI spend in this box from Cloud Zero. Alright. Very cool. And and if you're just joining us, maybe you came in a little bit late. We have a q and a open. We have the incomparable Emily Allen here in from Cloud Zero. I'm on the Campfire side, so we're starting to get a couple of questions. So, I think this one's pretty keen. Heather, how did you carve out AI spend? Is it a separate TL account in Campfire? So I currently have software, all going to one GL, and this is where, you know, Cloud Zero is pretty helpful and that it does allow me to see what my AI spend is separate from just putting it all in one GL account in Campfire. But Campfire also has tags. If your ERP has sort of tags or by vendor, you can though those are things you can actually ask the MCP or clog. You can say, hey. Yes. Pull the total software from Campfire. But in that total software, tell me if you know the vendors. Tell me give me a breakdown of anthropic, OpenAI, and then teach Claude that is AI. So when I say to you in future dashboards, pull AI, you're teaching Claude that that means go to Campfire, take the software spend, add up all of these vendors. Downside is if you have a new sort of vendor, you may have to just tell Claude, but I think it's like thinking of it like a intern or a first year associate really where, you know, you kinda have to walk them through how you think and how you would do that manually, and then you can sort of save that as a skill a skill, say to Claude, I would like to save that as a skill, and then it builds sort of a repeatable motion that you can you just say refresh. Just say use this again. So that's one way I would if I didn't have a Cloud Zero to pull my AI spend, and I had just had my ERP or other sources. Right? Like, Ramp has an MCP server. Brex does as well if you use other other platforms do too. So you can marry all of those data. Yeah. Yeah. I think that that's that also leads into kind of, Vivian's question. If the like, what would the logic be if you're not using your ERP or GL as a source of truth? Maybe you're running continuous syncs in the background from something like Ramp. Yes. Exactly. Or, you know, sometimes with us finance people, we still have our sheets, our spreadsheets, and I I still do. I still reconcile offline. I think it's really important to always have sort of a source of truth that you can go back to. Luckily for me, my ERP is my source of truth. But for example, let's say your ERP, you was QuickBooks or another sort of they actually have an MCP server too. But let's say that, you know, for whatever reason, you couldn't kind of access that and you had a sheet, you can upload your sheet to Claude. You can actually plug in Google Sheets to Claude and say, please access this Google Sheet and take this information. So you can actually also tell Claude this. You know? I think there's sometimes where I do say to Claude, how do I use you? This is what I have. This is what I'm looking to do. How would you advise me to give you this information? Where can you go? And then it will go into a, well, I could do this. I could do that. I can upload. And so really interacting with it, I think, for your specific use case is what makes this technology so cool. Yeah. And I think that, you know, when we were talking earlier this week, you kind of mentioned the, connection between Cloud and Campfire, which kind of lends into Chris's question about, do you find connecting cloud to Campfire helpful from analytics and a closed perspective? Does that help you do analytics on AI spend? Yeah. Absolutely. So, I love MCP servers and connecting them. I think there's so many questions throughout close or even in the review process. Right? So for example, with Campfire, yesterday, I was reconciling our cache, and I couldn't find, you know, a small minuscule amount. Right? In the past, I would have downloaded the GL detail. I would have analyzed it. I would have I actually just said to Claude, I said, pull in Campfire, pull this GL account. My variance is off by $2,000. Please find it. And within five minutes or so, it said your 2,000 is made up of $1,000 here, $1,000 there. That, in ten minutes, probably would have been an hour exercise previous to AI. So those are some use cases. For me, I also love being able to bring in different metrics of different source data before I've closed. So this dashboard was because I was in the middle of July trying to figure out what the future would look like at q two close. And for me, I liked that I could bring in revenue that was already booked and already scheduled in Campfire, but then I could connect to other sources that were a little more fluid and special specialty and sort of the cost side. So Cloud Zero, for example, we're able to see, what COGS by customer is. So I was able to marry, okay, I haven't booked or closed COGS in Campfire, but I was able to bring in revenue that's already in Campfire. I was able to use another source and bring in that data where that is and able to marry it to actually I mean, this is fake data. Everything's up to the right, but you can actually see some transactions and what's going on here. And being able to drill into before the quarter closes, before you're even close. Okay. What is this AI spend look like? What what customers are really sort of our our big differentiators, which from a margin perspective, you know, everyone in the c suite wants to know who's bringing you down, who's lifting you up, how do we boost our metrics. So, Hold on, Emily. Hold on. I thought that the only metric for successful AI adoption. is how many tokens you're. using. Yeah. I guess. Maybe. Yeah. Well, Oh, god. this is this is a a nerdy thing that we can geek out about in terms of being able to get some of these metrics real time. And what's also cool about, about Claude was I was able to say, right, we were in the middle of July. So I said, okay. Pull revenue from Campfire. You got that. With Cloud Zero, pull this data from Cloud Zero. Then based on the run rate, project the end of July for me. In fact, project h two if we have everything the same staying the same. And that's able to kinda utilize what AI is great for. A lot of the kind of the modeling, the sensitivity analysis, the levers that you wanna pull. So that was a really kind of exciting thing for me was to be able to answer some of those questions that, you know, I think everyone on this call, if you're in finance, has experienced your CFO, your VP coming to you and saying, I need to prep x, y, and z person for what this is gonna look like at the end of the fiscal quarter or fiscal year. And I know everyone, every operator's response to that is, but it's not done yet. This is not final. I haven't closed it. Please caveat that left, right, and center. But this makes me feel a lot more comfortable where I'm able to grab the data that I need. Yeah. I I think that, you know, as we're thinking about this, you know, you have an enterprise cloud account where you can share out special dashboards with different people on that account. So it's, like, all internal, Mhmm. all secure, all safe data. I am curious, like, if we're thinking about, should I use an MCP, should I use a data lake, where what was your thought process in making that decision? For me, I will give, like, immense props to our security and IT teams in that they have established very early on. They recognize everyone wants to adopt this technology, but they but that people, myself included, want them to be rule followers, especially this crowd, and do it in a safe way. And so I would encourage, you know, and, hopefully, this inspires some people on this call to go to the IT team to and say, hey. I wanna use AI to the deepest of my ability, and that may require me touching other and calling on other data sources. How do you best suggest I do that if it's not already set up? And to be honest, seeing on the other side, you know, yes, it was work for our IT team and our security team to ensure that we all had that, comfort, but I would say they did it. And it was pretty now everyone in our company uses Claude and is able to feel comfortable that what they're using it for is safe, what's connected. I was talking about how the IT team can turn on different connections or your admins of Claude. I know that if it's in the App Store, it's safe. It's approved. And if it's not in the App Store, you also can build a custom connection, which, again, sparks a nice conversation with your IT team. Hey. I noticed that this new application that we have where I do x, y, and z doesn't have an app in the App Store. Can we work on that together to build that so I can do, you know, a, b, c? I think, like, fostering those relationships is something that's unique and pretty cool and kind of necessary now in this day and age because you're right. There is a security aspect, and you wanna make sure that if you are pulling data from different sources, that it's secure data. It's not malicious. It's not putting your organization at risk. For me, where I take comfort and what I think about is what is the source of this data? So ERP, safest in the game. It's my ERP. Cloud zero, SOC certified, safest in the game. Other ramps, safe. Braps, safe. Well, you know, that's kinda how I think about it is connecting them. If that's safe, then and they're coming from safe places, then it's safe to do this in Claude. Yeah. I think that that's a really good one. So speaking of doing this in Claude, I think that there's two kind of pieces before we get into kind of that builder motion. I I really love the AI portion. Yes. People are thinking of, kind of that r and d tax credit, here in The US and, like, capitalizing software spend. I I just wanna see I I want everyone to see how you're looking at that and, and and just, like, explain, like, are we doing this all in co work? How are the numbers the same, month over month? Like, you know, that's kinda trust but verify. Yep. Yep. So I think a big part of prompting, no matter if you're doing this in code or co work, is doing it in steps. So for me, I had q one. Right? That's historical. That's closed. I had the q one metrics. Having it do q one first. So when I prompted this, I said, I want AI spend by department. I want AI spend by feature, and I want it from February to, April. I already had the answer kind of in my, you know, my reporting, and it worked. It matched. Then I was able to say, okay. Now that I know that this is good, that you understand what I'm looking for, let's move on to q two, and let's project out July. Let's add those together to give me an h one view. So it's really important when you're prompting to just take it slowly and by steps and trust but verify as you're going through those steps. I think, Kat, we talked about how sometimes I've seen on these sort of self help webinars or whatever that it almost looks like it's like a paragraph that you'll put in the cloud and you'll get this output. And that's actually not the case. It it's actually just a lot of short, quick conversations that allowed me to get to this place where I felt like, okay. This is out of spot where I can share this with my CFO or the board, etcetera. With AI specifically, I will say to your question, so Cloud Zero is does have the ability to ingest AI spend and then actually tie what the business objective is to each AI task. So, essentially, I'm able to see our anthropic spend. Let's call it a 100 k a month. Right? I'm able to see by department, by model. Yep. Kinda basic. But then I'm able to actually see, okay, in engineering, what amount of spend of that is actually tied to feature development versus testing? And then, okay, on sales, how much of that is pipeline generation? What's my and that's where you start getting into the ROI conversations and being able to marry, okay, if I have salary data, payroll entries, all encompassing in Campfire, I have revenue, Then I have the by feature, and the department in Cloud Zero. I'm able to then start kind of doing some cool analysis on, okay. That's great. Everyone's adopting AI at my company, but what's the return on that? What are we as a business getting on? And, you know, is that actually worth the incremental investment, which I do think is a blind spot right now across a lot of, teams that I've been talking to in terms of, you know, everyone's booking that anthropic, the OpenAI, the cursor, all on the p and l under OPEX and, you know, how were we dividing it? How are we determining what qualifies for that r and d tax credit? What determines you know, you used to have engineers coding and you were able to capitalize their time? What happens when it's actually anthropic coding? How do you prove what you can capitalize there? So this is this. is a, I think, unique feature of this dashboard that I was able to sort of surface to my leadership. Yeah. I think that it's going to like, if if I had this when I was going through my 25 tax credit, I think it would have been a much quicker, session. And I I I just see a lot of traction on this. So, you know, before we get in as we're getting into the building portion, what surprised. you most about how far you got without, writing the code yourself? I mean, the fact that this I was proud and happy to share this with very kinda high level people on my company. I think I came from the FP and A days and even the the accounting close days where month end, quarter end would come. I would get asked to build these reports, and it would be kind of a day of me obsessively, you know, checking, putting in different colors, borders, making it look pretty. Then if they need it in a slide format, you know, also doing all that. And those are all great skills. I think, you know, we'll always center my my objects on a slide. But, I do think what surprised me most was how, you know, kind of professional this looked and really highlighted what I wanted to share and highlighted sort of the next level of looking at the numbers versus just saying, hey. You know, here's what happened. Love it. So I know you had something to share about the the build process. Yep. Let me switch my screen real quick. Oh, and, yeah, good call out, Lauren. So Lauren's on the Campfire team. Thank you for showing up today. She's helping lead this event. If there's something that there's a lot of, like, really sharp folks on the call, and if there's something that you've created that you'd like to share with, Campfire, kickoffs, or or Emily, kick off some, light bulb moments for us maybe. Go ahead and share it at the. events@campfire.ai email. Alright. So this is my, my Claude. So if we go into I just wanted to point out one quick thing. So if you go into your settings when I was talking about that app store. Right? So if we go to connectors. So these are all of the kind of the connections that are available to us that are installed, meaning that our IT team switched on that on, you know, light switch for us to use these. But if you go to add, browse connectors, this is where that app store I was talking about. You can see all the different services companies that actually have that sort of built out. And this is where I think you could take a screenshot of this and send this to your IT person and say, hey. Any way we can get this up and running safely? So I did just wanna point out, because that can be tricky to find. They did recently move where it was. Yeah. I always, sometimes even when you ask Claude where it lives, it gives you the old answer, which is kinda funny. Yes. And that's why you trust and verify. The skills the skills that I was already referring to that they live here as well in the customized savings still. It saves there. And so one thing I did right before I wasn't going to bore you all with watching Claude do its thing, but I really just prompted and said, great. Build fully functional AI revenue. I said, don't use our data. Thanks. And then spend by models, spend by feature, Campfire revenue, kind of all of the things you saw in that dashboard. Right? And then it runs its. little commands, and voila, this this required no prompting. But I think how I got to my dashboard was this is a great start on the right. You can. always see what it produced immediately. And so this is where kind of familiar to what I showed. It it does produce sort of the similar output, but, really, what's gonna make this what you care about is, Okay. This is all q two. What if I want h one? What if I want projected q three, q four? What does that do the full year? And those are type of things that you can just free text in this comments and then iterate. Right? So, Kat, when we were talking about that dashboard that I made, that was a lot of iterations. That was getting to a point here and then saying, but I actually wanna see the feature section at the top, for example. Oh, I actually wanna see margin by customer, and I want you to tell me which customers should I spend my time on and understand the margin of those customers. And I think to someone in the comments point, they did point out two customers, and those weren't the customers that if I had just taken Claude at its word, those weren't the customers that we should focus on. So you do have to lean on your own reporting in a way to check. And that is something I always will say is that I don't really right now see a world in finance where you can only do your work and analysis in here. I think you need something that is almost like a control, like, in science or in experiments where they have a control group and they have the experimental group. This is your experimental. You need a control to sort of quickly spot check, quickly verify that Claude hasn't gone off the rails and giving you something you can't rely on. Yeah. And real quick for anyone, if there's something on Emily's screen that you want to zoom in on, there's a little, plus and zoom in and zoom out in the bottom right hand corner of your screen. So, you should be able to zoom in, rather than squinting, if you wear glasses like me, or or if you have a question, you wanna zoom in on something. Angelica did bring in. Did you create, did you create yourself the MD files, or did Claude do it for you? Claude created it for me. I to be honest, I don't even know what MD stands for. And I'm saying that humbly of, like, I just say, Claude, I need this in a PDF to share. Mark, mark down. Mark down. I thank you. Thank you. Well, I I don't know. I've just learned, like, artifacts or, like, docs. I mean, this is the thing. Right? And this is why we're having this series is that I'm not an engineer or technical by trade. I don't code. I don't really do this often. And so for me, where I use this is I like to say, okay. Show me where in my folder I can go back to this, and Claude will point that out to you. So it's, like, kind of almost being humble with yourself and just being like, hey. I need this in a PowerPoint format. Give it to me in that so that I can just download and plop it right in. And there's been sometimes where I can't copy and paste it or it doesn't look good, and I tell Claude that. I say this looks awful. Can you fix this? And, Yeah. obviously, maybe Claude's not the right tool. I mean, OpenAI is great and for visuals and same with Gemini, of course. And that's where I think as you get more familiar with what each of those tools specialize in, you can kinda switch and maybe take the download from Claude and upload it to chat, GPT in their newest version and make it more PowerPoint friendly. Yeah. I like to use the like, there's an engineering, thought that if you have an adversarial model on, like, creating the best AI engines, then you're gonna produce the this best results. I personally kind of use that adversarial model in my own prompting and creating agents, creating the best skills. So I'll I'll create my own prompt or skill or agent, and then I'll I'll give it to a different AI to critique it, make those changes. And, you know, I think we could go on a pretty, like, deep dive on, like, how best to, like, get your get the best out of your, like, token usage and, you know, do I do I create with, like, the most powerful AI model and then, like, dummy it down for the runs, kinda. And. that I mean, specifically using some of the cloud zero data, I've been able to hone in on right? Like, some of the things that we are using AI for are not effective. I think we were talking yesterday about I was able to see that some spend on reaching out to prospective customers actually would have been better used if that person had just sat down and wrote the email. And and so there is going to be this whole new wave of is that spend actually efficient? Is that driving hours saved? Is that is that driving, you know, return on your pipeline, etcetera? So that's a that's a whole new way that's about to come for this industry, I think. Yeah. Okay. Love to hear from the audience. You know, what what questions are still burning in in your mind? Is this kind of, like, creating some ideas for you about how you might be able to use it? Another thing I can share as well just while we're waiting for that to trickle in. Yeah. So, And and how many iterations did you go through? Sorry. no. That fits perfectly into what I'm about to share. So let me just make this bigger. So I actually can send this out, but, really, it was a lot of steps. So I reconciled first. I said, hey. Pull the different metrics. I stopped. I tied that out. Fix certain data. Right? Claude will ask you questions after kind of a a poll, if you will. So for me, I had, you know, one source of data said x customer. Another source of data had a, you know, a separate avenue for how it named the the customer. So I had to fix all of those. I didn't fix all of those. I just told Claude to. Then I said, okay. Great. I know it matches. Everything looks good. Let's widen. Let's pull in software. Let's pull in AI by by model, by feature. And then I said, okay. Now that we have all this data, I need you to actually put those metrics on the dashboard. I need you to calculate. Then keep sharpening. Right? I think I and at one point just said, this looks ugly. Move this down. I don't like this visual. Let's see. Yeah. I mean, it's really iterating. Right? Insisting on a meaningful metric. I think I talked earlier about how absolute dollars versus margin percentage. It was just pulling straight margin percentage, and I said, okay. Well, if I spent all my time on these two customers, then nothing will change in our overall margin. I need actually important ones. And then how, is that how you're also I see Vivian came in asking about the AI ROI. Is that kind of how you're also thinking about the AI ROI in your internal discussions and how. you're measuring it? Yeah. Absolutely. I mean, because we're we are able to see what people are doing on the AI side. Our internal discussions are really kinda up leveled in terms of being able to say, Last night, I see that this feature was being developed, and it cost us 10 k, let's say. And then that plan's ongoing live in next month. We will need to have made x amount of revenue within four weeks in order to say that that was appropriate to spend on from an AI perspective, as well as that engineer's time reviewing it. So I would argue that our conversations are okay. That's great that they're spending. They're getting that feature out, but then what capacity does that open up to to then work on another feature and then another feature? And so that next level of capacity planning, that next level of are we investing in the right things, especially in this AI environment where it can get expensive very quick, especially with some of those newer models. It can be thousands of dollars. We used to say, or your slogan at CloudZero used to be, every engineering decision is a buying decision. I would say with AI, every decision is a buying decision because everyone in your department can I mean, I ran a skill one time overnight, and it got stuck on a loop, and it cost me about a thousand dollars? That's that's not me kind of using Claude wrong. That's not that's just it can happen with these things. And so it it really is needed to be able to track and understand what the kind of the features or the different use cases people are using for AI from a financial perspective. Yeah. And and I like I like what you shared. You've shared about, like, COB zero and how you're able to really quantify that down, because, you know, especially with r and d tax credit and any sort of, like, quantification and budget review and people review on how efficient the teams are and how well they're tooling. I think that that brings in, like, a much more concise answer. And without having people to fill out, like, a survey or, you know, kind of, like, be in a kind of big brother state where we're recording the engineering screens, it's it's giving us, like, actual data and, like, hours served because we are. all using AI. We call it we call it on the edge. So it's kind of like almost looking over and seeing from the edge without necessarily, you know, invading that space or allowing people to create a freedom. Right? But looking over the edge and just seeing, oh, that relates to that. Oh, that relates to that from sort of a distance, but it allows people like yourself, me, like, we need to be able to understand if our AI spend is growing a 100% quarter over quarter, I need to be able to know what we did and if that was worth it. Right? I can't go and hire a bunch of contractors for a 100 k, let them just go and then not tell anyone what they built or what they provided or it's it's the same concept. So we are in this unique place where that's kind of a gap right now is understanding where it's being spent, what activities are happening. And so it it has been a really cool experience, being able to merge that data with other data from other sources. Right? That shows sort of, okay, what sources are on credit cards versus our enterprise agreements? What sources are on OpenAI versus anthropic? You know, does it make sense for our entire company to be on Claude versus some on OpenAI, some on Gemini, some on Perplexity? You know, what makes sense there? There's a lot of kind of unit economics that you can start going into in this world. Yeah. Yeah. I love it. And and, you know, I think that a lot of people are having these conversations, and how you've done this is so so precise and so like, the ability to change it and get started on it. Did you wanna share the link to this? Because, I know that, you know, for me, I I always like to have a takeaway, as I get. through these sessions. For me, that it it's always helpful to say, like, okay. Now it's Friday. I might have it's day five of close. Maybe I have a moment to think about what happens next with reporting. Maybe people on the call are also going to be asked about their AI spend, and, and this can help us get started on quantifying that and building out a visual so that, you know, we're answering the questions and getting a lot as much traction as possible with the data that we have available to us already. Yeah. Absolutely. So, this is absolutely available, and we can share this sort of guidebook. Kinda tells you that step through that I was talking about. And then we can also share the PDF version of the dashboard. And then, obviously, I'm available. I'm sure Kat will love it too. I think in the beyond the books, we I know we both are members, and, I think we should keep this going. I I think finance people don't get enough kind of FaceTime with each other to kinda talk through these things. And so please add me on LinkedIn. We can we can work through anything together. I think I'm in no way an expert, and so it's all about how can we build different things together and tackle the challenges. Yeah. Emily, that was a good shout out to the Beyond the Books. It's a community that Campfire hosts, of finance professionals. And, one of the recent questions I put in was like, hey. How are you looking at AI r ROI? And it was probably one of the hottest topics I've seen in there for in a while. Yeah. Yep. Well, hopefully, we we answered or covered some of that for people. Yeah. Well, I, you know, I I or any parting words, Emily? I I've really enjoyed talking to you today, and, I've I've enjoyed the questions that came in from the group. Oh, Krista. Oh, that's a familiar name to me. I see you on LinkedIn puzzles every day. I think you too, Emily. I love it. Yeah. No. Just parting words. I think for me, I've really loved the openness and honesty of professionals kind of in the network. Just being able to share how we're doing things and not gatekeeping either. Yeah. I think this is the next wave. This is this is here for us, and so I think we should all try to help. If you you made a margin dashboard or something else, that you think I could learn from, please send it my way. I'm I'm happy to learn. Yeah. Well, thank you everybody for joining, and, Emily, I truly appreciate your time today. It's been really wonderful talking to you and looking through your dashboard. Thanks. Happy to. be here. Alright. Happy happy closing, everyone. Bye.