Webinar
The Agentic Finance Playbook for Modern FP&A
Transform the way your finance team works with AI — with governed data, autonomous FP&Agents, and every number traceable to the GL.
Date & Time
On Demand
Duration
45 minutes
Details
In this webinar, you'll see how Cube turns AI from a finance liability into a finance asset — with governed data, autonomous agents, and traceability all the way back to the GL. We'll cover:
Why the AI Trust Gap Is Finance's Biggest Risk Right Now
AI can be 80% right — and in finance, that's 100% wrong. We'll unpack how to close the AI trust gap so that you get the right answers 100% of the time.
How Cube Becomes the Agentic Finance Layer
Cube consolidates your ERP, CRM, HRIS, and spreadsheets into one clean, decision-ready layer — capturing the shadow data other platforms never see and giving every tool the business context it needs.
FP&Agents in Action
See the four teams of purpose-built agents running across Data Management, Analysis, Planning, and Business Partnering — from the Data Manager reconciling your GL to the Analyst explaining variance to the Business Partner drafting your board narrative.
Trace to Truth — Every Number, Defensible
Every AI-generated insight maps back to the specific GL transaction. We'll demo the drill-through from a board slide to a single line item, so every number you put in front of leadership can be defended.
The MCP Server — Your AI Stack, Finance-Literate
Connect Claude, ChatGPT, Copilot, and Gemini to your governed financial data in under two minutes. Every answer, traceable. No code. No consultants. No IT lift.
Speakers
.png)
Video Transcription
Welcome, everybody. Thank you so much for joining today. Today's webinar, we're gonna get started in about two minutes. We're just gonna wait a couple minutes for people to join and get settled, and we will get started shortly.
For those just joining, we're gonna get started in about two minutes. One to two minutes, we're just gonna wait for people to join and get settled.
Welcome, everybody. We'll get started in about a minute here. Thanks for joining.
Perfect. Well, I think everybody has time to join, so I'm gonna go ahead and get this webinar kicked off and share my screen.
Awesome. Well, good afternoon, everybody, and welcome. Thanks so much for joining us for today's webinar. We're gonna dive into Cube's AI functionality showcase covering what's live today, what's coming, and what it means for your finance team.
Before we get started, just a couple of quick housekeeping items that I would like to go over today. First thing is the live q and a function that we're gonna be having at the end. So we'd love to make this session interactive, so we will have a chance for us to ask questions at the end. Normally, the session will run for about thirty minutes, and we will reserve the last fifteen minutes for questions. So you can submit questions at any time during the session using the toolbar at the bottom of your screen. Feel free to submit your questions at any time, and we will either answer them in the chat in real time or cover them at the end.
Another, quick thing I wanted to mention is that we will be recording this webinar. So after the recording, you will get an email from myself with the recording, a copy of the slides, as well as some other information to dig into as well.
To walk us through today's session, I'm thrilled to introduce Jim Bullis, our VP of solutions here at Cube. Jim has been with Cube for nearly five years, building and leading our pre sales implementation and support teams. He's worked side by side with customers from first demo all the way through implementation, so he knows exactly how Cube delivers value and practice. With that being said, take it away, Jim.
Absolutely. Thank you, Lauren. And we also have Moe on the line. He will be helping answer questions in the chat as they pop up.
So as Lauren said, feel free to fire those away. So let me just get my stuff settled here. Alright. What gonna walk through today is we're gonna talk about the moment we're in.
Right? AI is is here.
People are starting to use it. We're gonna talk through what finance teams are using it for today outside of just the cube context, but in general, how are they embracing it? So maybe just some tips and tricks on what other folks are doing if you're just getting into it yourself. Obviously, we're here for Cube, so we'll showcase, hey.
How does Cube's AI come into play? Where do we fit in in this process? And then finally, showcase the functionality and talk through kind of the gains that you'll receive. So we'll we'll walk through all the AI capabilities that you have with Cube and also combine it with, like, an MCP to utilize Clog or chat p chat GPT on top of it.
So where are we at today? We'll start there.
Right? Traditionally, this has always been the case. So I've been here at Cube for five years. Before that, I spent a decade in FP and A consulting, working with a lot of the legacy platforms in the space.
But, traditionally, finance teams are spending eighty percent of their time just gathering data, reconciling it, formatting it, not analyzing it. Finally, the help is here. Right? That AI inflection point is now.
Teams are starting to use that to reverse that number, right, to spend more time actually analyzing, planning out new forecast, new scenarios based on what's happening.
Managers are looking for ways to limit, you know, increases in headcount. Right? Hey. Hey. Where can we use AI to maybe not need to make that extra hire to get more of our current employees?
And, you know, finance is now the strategic engine of AI in a lot of organizations. Right? Because you sit across all these different datasets. Right? Traditionally, it's always the GL, your ERP.
But what we're seeing is, you know, we have HR systems, CRMs, operational data warehouses that you're tasked with gathering data from. So you're familiar with those. You're helping those other department heads. It's natural you're gonna fill in to stop start, you know, up leveling those broader teams as well.
So what we're hearing in the market, Gartner found that, you know, eighty three percent of professionals are already using AI tools in their workflow. What's not so great is seventy percent four percent actually report using personal or free AI tools that aren't sanctioned by their IT teams.
Great that they're up leveling, maximizing their time, being more efficient, but there's some inherent downsides if you're, you know, navigating outside of your internal processes just around security, sharing data, things like that. But in general, you know, finance teams are seeing massive increases in their speed. Right? You know, doubling or decreasing the amount of time it takes to to close the books, putting out more analysis, so on and so forth. So really fantastic.
What are some of the ways they're doing it? So if you're just starting to scratch the surface, what are some easy ways you can jump in? And I was not some expert coder, anything like that when I started exploring this. You just start prompting.
You can ask AI how to use it. It'll give you good feedback. But some of the basic ways that finance teams are starting to embrace this are just what they do every day anyway. Slide deck generation.
Right? You can pump data into cloud. It'll produce nice looking decks for you with your company's formatting and colors and all that kind of good stuff.
Dashboard generation off that as well.
Big thing finance teams have to do every month, produce the narrative for your numbers. Right? What is driving those variances in the business? What increased or decreased that versus what you expected to happen in that period. Having a layer that can jump in and go and dissect your data and produce some analysis and variance explanations is huge.
Obviously, just ad hoc research and insights. If you can pump data in and look for new insights that maybe you didn't have today, that can really uplevel you and maybe expose some areas of the business you can improve that maybe weren't in the forefront before. So just using the AI that's out there. Again, this has nothing to do with Cube.
You can do these types of things today, so I encourage you to go do that. What are some other areas that teams are interacting with AI? Using it to increase the speed which they can close and reconcile. Right?
You can upload two files, have them compare against each other using AI and find discrepancies, maybe find duplicate journal entries without having to comb through it. Right? You can use AI to do that.
Also, big time saver. We're in spreadsheets all the time. You can have AI help you write formulas. Right?
And some of it's just base like, I'm taking raw and structured data. I have to add some extra columns to be able to filter it the way I want. I used to have to come up with these really lengthy formulas or do a bunch of copy pasting. Now I can just prompt AI and say, hey.
Take this column. If it's empty, do this. If this column says this, do this. They'll build that formula for you, do lookups for you, all that kind of stuff, just saving you time.
Right? You still have to double check it. That's where all your finance know how your spreadsheet expertise comes into play because you can double check and do some QA on it to make sure that actually is producing the results you want. But what used to take you an hour might take you a minute from that perspective.
Really interesting one I heard from some customers that I wasn't doing in my day to day, but makes tons of sense, is stress testing your models. Right? Maybe you have to put together a plan for the next quarter or the next year. You know you're gonna walk into that executive stakeholder meeting or board meeting, and that person who's gonna stress test it and push back against you, do that with AI before you even get to that meeting.
Tell it to be an aggressive, you know, board member who always pushes back and see where the gaps are in your plan or something that might be off. If you're projecting thirty percent growth for the last six six months or only ten percent growth, it can spit that back to you. And maybe you adjust your plan or maybe you just come up with a narrative to explain why you built those assumptions in. So I thought that was really interesting use case that I hadn't played around with, which is, you know, creating some, AI to kinda challenge you and pressure test your plans.
So really exciting. You're gonna see some of these as we showcase Cube's functionality. But but the problem is if you don't have a governed data layer in between, AI has kind of inherent trust problem because AI can be eighty, ninety, ninety five percent right.
But in finance, that's not good enough. Right? And there's a few reasons this comes up. Number one, it it lacks full spreadsheet awareness.
It's really great. You can do really cool things in spreadsheets with AI. The problem is it doesn't always understand the full context of what you're trying to do. So we're talking with a a CFO, and you're saying, hey.
We have to be really careful because we add in a row or change a calculation. It doesn't always understand why that change was made. It might overcorrect for it or not notice it. So having that full spreadsheet awareness to understand what data is missing or or not is really important.
It's probable list probabilistic versus deterministic. Right? It's gonna protect the next word, but that next number might just seem great. It can kind of misstate budgets by hundreds of thousands or sometimes based on how big your numbers are, because it's it's trying to be or it's being probabilistic.
Big one is there's no governance or audit trail. So when we're laying this on top of our data and I wanna share it with my team, if I'm sharing it with marketing, I don't want them to see engineering's payroll data. Right? So I have to be really careful about that.
There's no governance there, and there's no audit trail. Right? If I'm tracking or if I'm making changes to my budget or my plan off in a spreadsheet and then pumping it back into AI, it doesn't always know what changed or have that log. And then finally, a lot of times, I'm it's great in Excel working me individually, but if I'm collaborating with my team across five, six different spreadsheets, like, if Lauren makes an update, I'm not always is that in into my model that I'm using in my AI over here.
So these are some of the the big drawbacks right now today around where it's great, but it's it's falling short in finance. So that's where, right, Q comes in. Right? What you need to make this work is a layer that's bringing your data together that has that governance, that audit trail capability on top of it to make sure that security is in place, your data is good.
You want a layer that's understanding your business hierarchies, your your logic, your multidimensional aspects to make sure that as your data is flowing in, you're not missing data. It's rolled up. It's already got that business logic built in. It already knows how you calculate gross margin or the rule of forty, right, that's already in a system that it can read off of and work with.
And then finally, you need a live connection. So as you're interacting and updating plans, you can see the impacts of that flow through. So that's where come Q comes in. It's the first AI native, kinda workflow native FP and A platform finance layer that's out there.
And what that means is when we say workflow flow native, sure finance loves to work in spreadsheets, but you also collaborate across the org. And you have some users that maybe don't even have Excel installed on their machine, you need a way to collaborate with them. And that could be through collaboration tools like Slack or Teams. That could be through a web interface where you can send them a link.
They can go view a report, make updates to a plan there. Or for those more savvy team members, it could be an Excel or Google Sheets, but we need a spot that works across all those. We wanna keep finance in the driver's seat. We don't wanna burden the finance team with tons of code and syntax they have to learn.
We have to make it really easy for them to do new modeling, do new planning, but synchronize it and allow the team to see those impacts. And then you need trusted AI for finance, which means you know all the data's in a good spot. You know it's governed.
You know you have all the transactional detail there underlying it. So if you wanna go trace down into a detail, you have every insight there down to the transaction level or aggregate up aggregated up across those business hierarchies that we built out.
So that's in general how Cube works. Right? We've got this agentic finance layer in the middle, which is Cube. We're bringing your various datasets together, and that can be source systems like you have on the left.
Right? NetSuite, Sage Intact, SAP, any system out there. But with finance, again, you're pulling together from data from tons of different systems, your CRM, your HRIS, your operational data warehouses. Right?
You need that all in one layer so the data can talk to each other and be lined up in the same time hierarchies, the same metrics, everything like that. But when you're actually interacting with it, where you interact with it differs based on the audience or the team member you're working with. That could be an Excel for finance or PowerPoint, or it could be somebody asking a question via Slack. It happens to me all day.
Right? Or maybe your team's on Google Sheets, or maybe it's a mix of both. Right? The broader org uses goo the Google Suite, but the finance team's in Excel.
But, again, nowadays, we're taking advantage of chatty, petite, flawed. We wanna have AI help us drive these answers, or I wanna send somebody a link. Right? I want them to log in to a nice workspace that's been curated for them to come in and look at the right numbers.
I don't wanna have to send them a spreadsheet that gets stale the next day. So Cube is that financial intelligence layer that sits in the middle, has our agents working, and allowing you to interact with that data wherever you need, you know, aim to become headless here in terms of, you know, allowing you to interact with that from wherever you need.
Alright. So I mentioned agents. We've got this financial intelligence layer on top of it. We've created the, you know, agentic finance layer now that we've built in all these agents. And we've brought them together in these different teams.
So there's tons of agents behind the scenes. But based on what you're trying to do day to day, you're gonna have your data manager, your analyst, your planner agents, and your business partner agents. And what that looks like is your data managers are helping you maintain your model in the back end. It's making sure that all of your accounts are in place, that the mappings are good to go. Your data has been loaded, and everything is up to date.
Next is I need to analyze my data. That's where I can prompt a question. It'll find an insight for me. It'll help me sort out your biggest variances and dive into it from that perspective.
When it comes to FP and A, the second letter in that acronym is planning. Right? So we want a really cohesive planning agents that allow us to go ahead and expedite our plans and allow us to cut out a lot of the initial legwork it takes to to build those out. And then finally, we need to collaborate with all our business partners across the org.
We want really good AI agents that are gonna help us collaborate, build compelling dashboards really quickly. And then sitting across all this, we have the MCP as well to allow you to connect in your LLM tools to help build all this out.
Right? So that's that Cube plus Claude. Again, ChatGPT, anybody that's using that MCP, we can go ahead and interact with that. But we are building that agentic finance layer in the middle that has all of our agents working to help build out your plans, work in our workspace. But if you wanna spin up a PowerPoint deck using Cloud, you have the data in place. If a team member is looking at that data via Cloud or ChatGPT, it recognizes who they are and will restrict it to the slice of the pie that they can see. Again, so if they're in the marketing team, they can't see engineering's data as they're prompting and asking questions.
Right? So tons of cool stuff you're gonna see today, but I'm even more excited for what's gonna come in the next few months here. You can see these four teams of agents we have set up. Again, there's a lot more behind the scenes, but we're we're looking to improve it across the board, make it easier to maintain your models.
There's gonna be some connection setup agents to expedite how fast we can, you know, get new connections spun up. Probably the one I'm most excited about is the planner agent. So what you're gonna see me do today is generate statistically driven forecast looking back across trends and growth rates, things like that. But something that's really cool that I always hear from finance teams is, hey.
I wanna spin up a new forecast. I wanna hold my personnel cost steady, because we're using AI now. I don't need to have those extra headcount in. We're gonna increase revenue by three percent, and we wanna increase our marketing spend by five percent.
I'll be able to talk to a planner agent. It'll spin up a new scenario for me. I can see what it looks like that across my full three statement model without having to do any of the legwork aside to prompt and then doing some variance analysis there. So really cool stuff.
It does not end with where we sit today. We are building out tons of new agents to help you build out, you know, slide decks, so on and so forth, help with month end close processes and reconciliation, help you build more advanced formulas, in the back end of the platform. So really exciting from that perspective. And what we see our customers who have already embraced AI in with Cube is they're flipping that narrative.
Right? Eighty percent reduction in manual time. They're gaining more insights. Things that used to take hours or days are now taking minutes.
Day one, finance can come in and own the process. You don't have to spend eighty hours learning how to build reports. If you're in spreadsheets, you can generate it really quickly. If you're in using LLMs, you can generate that quickly as well.
So really exciting to showcase these for you as we go through. And what we're gonna show today in our live demo is we're gonna walk through kind of a day in the life of a finance team member and what they have to do every anyway. Like, none of this is really AI driven. If you wake up early in the month, you have to go ensure the data is all coming in, updated, mapped.
Then you're gonna be tasked with discovering your top variances. Hey. Budget was this. Actuals is this.
Why? Then you have to create the narratives around those, answer questions from team members about, hey. What was my marketing spend last month? You're driving new insights.
You end up updating all your reports, sending those out to the team. You have to go update your forecast for the new month, maybe doing a rolling forecast. Then you got a board meeting coming. Right?
You have to put your board and assets together. And then finally, just for the sake of this one, we're gonna build a new balance sheet forecast. It's something we've never done at this organization take organization I work for, but I wanna go build a new balance sheet forecast. Now, again, these are all tasks that I have to do today, but we're gonna show you how Cube's AI folds into each of these to make your life tons and tons easier.
So we're gonna go ahead and dive in. So let me end my slideshow here, and I'll pull this out of the way.
And what we're gonna do is I'm gonna start with that first step, which is it's end of the month.
If I'm just doing things in spreadsheets today, I have to make sure everything's flowing into Cube appropriately or just into my reports appropriately. With Cube, have all your data sources synced. And a big one is I have new accounts, new vendors that are flowing into the system. I need to make sure those are mapped incorrectly.
I can go do that manually if I want to. Like, a new account pops in. I can go map that in. If it's just one, maybe that's great.
But if I have a hundred vendors come in, it's gonna be a bigger deal. So we have built in our first agent that we'll see today is our mapping assistant. So in this case, if I come back here, it says it found a new account, which is one zero four zero zero investments and securities.
What Cube does based on how my map renewal is set up is it's detecting, hey. Based on the account number that I see, the description, where do I think it would map into the model? And it's saying, hey. I think cash and equivalents is a good spot.
Right? Now, obviously, you're you're in finance. You're gonna know and detect that. But maybe it's a list of vendors, and you categorize those by what type of vendor it is.
Is it a software vendor? Is it a maintenance vendor? Whatever it might be, it will go look at that and make an initial interpretation to save you that initial time of, hey. Where should this go on my hierarchy?
So then I can follow prompts. It'll go generate the proper mappings for me, and I can just click a button to get those into cubes. So trying to save you time on that front end. Like, what are the things that always, like, stink in your process?
It's identifying those new accounts that flow through. Where do I put them? So on and so forth. Cube is catching those for you, suggesting mappings for you that you can then run with.
So, again, month end, use an AI to save you time right out of the gate.
There you go. There's that new screen. Again, if it's just one, it's probably pretty easy for me to go interact with that. But the next step, month end, we got our data flowing in.
We feel good about our mappings. As I mentioned, we need to go analyze our data. And one of the first things I need to do is go find my top variances for the month. So what chain or I expected this in my budget for May.
Well, here are the actuals. Where are those top variances? We built out a smart variance agent, which will go through, and you can compare any scenarios. But most of the time, for our customers, it's actuals versus budget.
And you can pick the specific time frames. If I wanna do entire quarter, entire year, I can go ahead and do that. Let's just go do May here.
I can get really detailed specific department. I can set minimum, so only show me anything over five thousand dollar variance or a specific percentage. But we have built agents that are gonna sweep through every single intersection of budget versus actuals in our model across all of our business hierarchies and kinda serve those variances up to us on a silver platter, saving us hours of time. So here you can see, hey.
I got my top variances six hundred twenty two k. It does it by both dollars and percentage. It gives me a lower AI right up at the bottom. But what I love here is it stack ranks and sort these for you at every level.
So I can see right out of the gate, and usually is the case, my revenue forecast is off. Our budget actually, in this case, it was under. So we expected to do seven million. We did eleven.
But it doesn't stop there. I'm gonna go on and go into details. This is where the beauty continues here as, hey. I wanna go look at the specific GL accounts that broke out, see, like, was it my implementation revenue, my platform revenue?
Okay. I can see my platform revenue is my top variance, four million versus five hundred thousand. I have other hierarchies in here. I have markets and products or what business units, whatever it is that's specific to your business.
But I can click in the markets here. Now it's breaking it down and sorting those by markets. So I can see my top variance was the east market, and I'll go look at look at by product. So within a few seconds, if I wasn't yapping here, I got to my top variance sorted.
I can, you know, navigate, dig up and down to go figure that out. So tons of time saving using our SmartVariants AI agent to go sweep through your your actuals, your budget, find those top variances so you can start explaining those. So really, really cool from that perspective.
Great. Now I just need to start driving both narrative and analysis. And I mentioned before, we're workflow native here at Cube, which is we meet your stakeholder where they wanna be. So when it comes to asking questions and using AI from, like, prompting it on our side, we have the workspace here. I can ask questions in a lot of different interface. So I can ask it through the MCP and, like, Claude.
But if your team likes to work in Slack or Teams, you can ask it there. I'll show that in a second or in our cube sidebar in our our spreadsheet add in. So I'll start here, and let's just go start by asking starting in May twenty six, Looking back, I'm gonna go look I wanna see what expense has grown the most over the last three months compared to the prior three months. So looking back, what expense has grown the most over the last three months versus the prior three months.
Now what's cool about this is I'm not these are not, like, quarters or anything like this. I'm just prompting it with natural language because that came up.
While I'm doing that, let's go down to Slack and show what it looks like to interact with it there. So I've got my Cube AI assistant here. I can come in and ask a question. And maybe I'm just curious. What I was looking at the revenue before. What were my five biggest revenue transactions in April twenty five or twenty six?
Now Cube is gonna go prompt it, dig that up. And for me, maybe I'm the head of sales, I don't need well, they probably know the biggest deals or whatever, but but I don't need a license to go back into NetSuite or SAP. I can just prompt it here, and it'll spin that back up. Same thing.
Maybe you're not a Slack shop. You're a team shop. And every month, I get asked by the head of marketing, what did I spend last month? Well, I can just say, what did marketing spend last month?
Right?
Maybe I just pay post this in the Teams or Slack, and I copy paste it. Or if I want to, I'm I trust that person to be savvy.
It's gonna go give me that number really quick. Right? My marketing was eight hundred thirty three k. It broke it out between paid ads, promotional, and conferences.
If I wanna prompt it and go deeper, I can. We'll go back up to our web interface here and our workspace. And comparing the last three months so this is kind of a weird request, but this is the type of stuff that comes up in finance. March through May versus December through February, what grew the most.
Right? So it's telling me, hey. Contractors and consultants, it gives me a little write up. Paid ads is the second biggest increase.
It gives me, you know, some details there. But what's cool is I can click on this, and Qubel generate the report it used to drive that analysis. Because, again, anytime you're dealing with numbers, you wanna make sure you double check it. So you can come gut check it, and you can see here it's saying, hey.
Take February twenty sixth, the trailing three month total, and generate that up.
Now if I wanna come in here and take a peek and look at my contractor cost, right here, I can drill down and keeps gonna spin up all the transactional detail that makes up that number. So if I go and collapse this here, right, I can see every transaction. I can see the accounts that it hit, the contractor and consultant account. I can see the memo field, the transaction numbers. And, again, the head of marketing or the head of sales, they don't need a license to my g GL system because they're interacting with Cube here. They can just go ahead and ask the question, get the details, even the transaction level from that perspective.
So really slick being able to use AI to drive that deeper level of analysis. And then I can't remember if I looked at the results from Slack. So here are my five bigger biggest revenue transactions. Right?
And it pulled back this little chart here. I can actually expand it because there looks like there's six hundred and twenty four lines of transactions for this month. It gives me the top customers. Implementation revenue dominates the top three.
Those are my biggest, you know, revenue drivers. So those are completely different GL accounts. But because I just said revenue transactions versus platform transactions, it's giving me everything underneath it. So allows me to prompt further if I want to from that perspective to kinda dive in.
So really, really cool being able to service your team wherever they see fit from that perspective. Alright.
**** ** nose real quick.
As we're going through, feel free to drop questions in the the chat, the q and a.
Where do you see fit? We'll we'll either get those answered by Moe, or we'll we'll answer them as as we get going at the end of the session here. So I have come in. I've gone ahead and updated my data.
It's all flowing through here at the end of the month. I used the mapping assistant to make sure everything's accounted for.
I found my top variances using AI with Cube. I've been asking Cube questions via our AI analyst in Teams, Slack, in our workspace. Now I need to go update my reports.
I can also go ask Cube questions in our sidebar. So I'm gonna come into Excel. Just maybe I'm spinning up a new report from scratch here. I can click new.
Or, actually, before I do it, let's go ask it a question, and I'm gonna say, you know, what time shortcuts are available with Cube? So I can ask the same data questions I was asking before, but what's great is I can query our knowledge center. So if you're asking, like, hey. I have Cube.
I wanna build a new budget. How do I get started? Things like that. You can ask questions as well here in your sidebar.
If you're finance and you're always in Excel and you don't wanna go back to Slack or Teams or the web, you can just ask those questions here, and it'll spit those back for you. Because maybe I wanna build a new report, and I wanna see, like, a trailing three, six, twelve months. Right? I can go add that in now.
So let's go build a report real quick. I'm gonna go pull together a p and l, and I'll show you how easy this is to navigate. We're gonna go pull in, you know, my totals for twenty twenty four and twenty twenty five just to make it a simple report here, and we'll go ahead and add everything else. Right?
So I'm in my spreadsheet. Again, workflow native.
You're a finance person who's just doing things day to day. Right? This is you know, I could prompt you to build build this out for me. This but this is, like, a big view or report of data.
I can throw my formatting, my colors on here. But as I mentioned, I wanna go add in some time shortcuts that I just prompted our knowledge center about. And, you know, let's go see May twenty six trailing three months, and I'll just do a couple different ones. Right?
I wanna see trailing six months. So we've built in all these really cool time shortcuts. This one is just t x periods. You can put in literally anything you want.
Like, if somebody's like, give me the last twenty one months. Obviously, that's not one you'd usually have prebuilt out. Cube has that functionality, and then you can go up to, like, two hundred if you wanna go look back. Cube's gonna go ahead and and just interpret what we're trying to pull in and go calculate those values for me.
So now I have I mean, kind of a goofy report here, but I have my twenty twenty four total, my twenty twenty five total, and a bunch of trailing totals in the middle. But the key here is I'm not doing any more manual work. I'm just prompting Cube. Hey.
I would like this in my report. Please give it to me. And Cube drops those numbers in there. So there's no big prehistoric formula.
Like, that looks like a v lookup. Have to go run. I don't have to dump a bunch of data and do any math. I use Cube to figure out, hey.
What's the best way to interact with this? And now I can go ahead and query my data using time shortcuts. We have year to dates, quarter to dates, lifetime to dates. You can look forward.
If I'm doing a rolling forecast and I wanna see my total of the next twelve periods, really, really slick, you can go ahead and build that in.
Alright.
Not the fanciest, most formatted report, but I could clean this up, push it up to the web so that my team can see it there. But I'm I also like to share reports via dashboard so my team can log in every every week, every month, and go ahead and, you know, interact with their data. So I have a nice executive dashboard I've already built up. We're gonna use AI to help us build a new dashboard widget because I don't have time to, like, go sort through. Our dashboards are really intuitive, but it's always easier to get, like, a good starting point. So I have a nice blank spot here for a new widget I wanna go add in, and we're gonna go build with AI.
So let's go build a line chart maybe for the last eighteen months of revenue data. And I know I have data by market. So line chart for last eighteen months of revenue data showing different lines for my different markets.
Right. Cool. Simple prompt there. I can go ahead and generate that widget just to showcase tons of different, like, charts and widgets you can have here.
I can go ahead and filter so I can pick and choose, hey. For each visualization, do I want this to update when I filter into it based on the different time frames, things like that? Keeps going ahead and generate those values for me, showcase it. But I used AI to build a lot of these widgets.
Again, it's a drag and drop experience. It's really fun and easy to be able to get in here and create different visualizations. But I can go ahead and, you know, have Qiub do Qiub do that for me. So you can see it's it's coming through.
It's generating, what it's how you it's interpreting your your prompt, kinda giving you a thread as you go through. And this looks pretty good. Right? It asked for eighteen months of historical data line charts.
I can go ahead and and throw that into my dashboard. And as quick as that, I've got a new widget ready to go. So, again, using AI across the board to help you really, like, maximize your time and be really efficient, but also maximize your impact with your different stakeholders. So, we'll cancel this.
Maybe I'll change that later.
We've updated all of our reports. Those are the data's flowing in. We found some insights, some narrative that we wanna drive things out with. Next, I wanna go update my plan.
And, usually, what's a huge pain for Teams is just having a starting point. Maybe it's I wanna start a new budget. I'm starting new forecast, or I'm doing a rolling forecast. We just flipped over the month. I need to figure out some data for that last time period. Cube's got smart forecasting capabilities where we're gonna use your all your growth rates, your trends, your seasonality, and you can have Cube spin up new plan data for you using AI, which is great because I can come in here, and I built one before. I called it Cube AI demo new forecast.
It already has something. I guess I'll just create one. I will call it Cube forecast.
You can pick what you wanna forecast. Right? In this case, I'll just do everything. Most of the time, I might just target, you know, income statement accounts or maybe I'm just doing revenue in this case.
You can figure out what I wanna populate. So in this case, maybe I wanna generate, like, a rolling twelve month. I wanna do that new month data. We'll do June through May.
Or why not? Because he was doing it all for me. I don't have to do the manual legwork. Let's just go through the end of next year.
Right? Super painful to do that today. Most folks, if they're doing forecasting at all, are usually doing it through the end of the current year. Now you have Cube.
Just tell it, hey. I wanna go up next eighteen, nineteen months or farther. You can have a living, breathing forecast that you can go ahead and update. Right?
You're maybe not asking your stakeholders to go update that, but if you're just working with the finance team and you wanna see what things look like in q four of next year, have Cube spin that up for you.
Finally, here, you're gonna pick what you want to inform your model. Ninety nine percent of the time is actuals, but how much history do I wanna use? Maybe twenty one and twenty two were kinda weird years. Twenty twenty three is when it kinda leveled off and got kinda normal for us. Maybe there's some anomalies back then.
Now I can go ahead and pick the custom date range that I wanna use to inform my forecasted values. So highly encourage every kind of Cube customer to go ahead and use this functionality. You can use it as a starting point for your plan. Right?
Hey. I want something I want my team to react to when they start the budget. You can use it as a gut check. Hey.
We already ran our budget. What does AI think we would do? And I can stack those two next to each other, run some variance reports, and see if AI has things were way off in some spots. So really, really cool being able to use AI to help build new passes of your plans.
So highly encourage that. Alright. So we're in the middle of the month. We've ran our reports.
We're updating our narrative.
We got dashboards. All of our internal stakeholders, they can go interact with the data. I've created a new plan. Now I got that board meeting come up coming up.
I need to prep all my assets for that. And as much as I'd love them to log in to Cube and look at our dashboards, we all know that these board members aren't gonna, like, log in. They don't wanna license to a new system, and maybe even your CEO doesn't wanna do it. Great.
Let's go use our MCP and Claude to build some really compelling assets for the team. Right? So what I'm gonna do is I was working with Claude before, and I did some some prompting here. I was building out my deck, but, you know, here, I have was building out some, like, interactive PowerPoints.
So I I because Claude is hooked up to queue, I can run some really simple prompts, which or, of course, I do have this spun up so you can see what the actual output looks like. Let me come in here and take a look.
Second here.
Notice this is the cloud problem, not a key problem. Let's pop this open.
Alright. So I asked Claude, hey. Take all my financial data. Maybe I uploaded last month's or last quarter's board deck, and I said, hey.
Just take this and go build this out. I did this all with simple prompts saying, hey. Take this. Give me a k q one KPI summary.
Here's the metrics I wanna report on. Here's the widgets. Again, if I already had a deck that I've used in the past, I can use that. I can use my company's formatting.
But I built this deck with just a simple a few simple prompts, and then I went through and made some adjustments.
Right? I said let me get rid of this preview. Hey. I'm missing kind of a summary TLDR slide, so throw that in.
Like, analyze the data. I'm if I'm a diligent finance team member, I'm gonna go make sure that it matches what I actually think. But if it can produce the various analysis, some of the stuff like that that I need to spit out, I can do that. Or the first pass I did at it, it generated a bunch of charts and widgets, but it was showing unformatted numbers.
It was just, like, ten million with no commas. I was just like, hey. This doesn't look right. Go adding some commas.
Be more efficient with the space. If it's in it makes more sense to use thousands or millions. It's incredible how fast you can put together board decks now utilizing cloud on top of of cube data. The other thing we we have customers doing is they put together, like, really cool HTML interactive dashboards.
Again, for those trusted team, like, board members that aren't gonna log in to a system, you can have it prompt and create a, you know, HTML, like, interactive experience with your data. Again, I have never built a website in my life.
I literally have all the MCP and cube and Claude sitting on top of it. And I just said, hey. Go build me, like, an interactive dashboard on this website. And I said I wanna be able to, for the first tab, do it by q one, and I can go interact and click in the different months.
I want to be able to filter by by different products, markets, really basic prompting. And I got this really cool output where they can come in and start to play around with their data. And I'm just playing with it, and I said, by default, it gave me these kind of selectors. I was like, let's spin up another one, and I'll use pick list instead of literally, I said is spin up a new one for last year, same metrics, same widgets, just instead of at the top using those selectors, make it a pick list.
And within a minute or two, I had this really cool thing that I could share with a team member or board member or something like that so they can go in and and play around with the data in a better fashion than just a board deck that I might be sending out. So depending on who you're collaborating with and how much power you wanna give them, you can get data in their hands really quickly in a really compelling, really professional looking format. Again, feed in your company's colors, basic formats, use decks you've used in the past so it can match it, but really, really cool and compelling from that perspective.
Alright.
Last item I'm gonna call out here is my CFO prompted me to go start forecasting my balance sheet. And I worked with some customers that were really diligent about balance sheet planning. Most don't. They're like, we don't really plan the balance sheet. Like, it's daunting to even think about where to start to go start planning out, like, doing a balance sheet budget or forecast.
What I did here is I had Cube go ahead and pull together all my data. So I just said, give me my income statement and balance sheet for the last couple years. So I I hit fetch with Cube. It pulled it into the spreadsheet. You can see here, I've got twenty four, twenty five. And then out in twenty six, I've got my income statement, which I'm regularly forecasting.
But down below, I don't have any, you know, balance sheet data. I just came up and said, hey, Claude. I have it in Excel here. I wanna forecast my balance sheets. Please use my existing forecast in this file for June to December from my income statement. Again, Cube centralizing that data. Make sure you have the right data there, the right security on it, the right hierarchies.
Generate a balance sheet values based on formulas. And then in column AL, which is this one, I want you to give me, like, a little description of what you did for that because this would probably gonna be a great starting point for me. It's gonna be give me ninety percent of the way there. I'm gonna change the formulas and calculations.
So I ran that this morning. It came up with the numbers. First thing I looked at is does this assets and liabilities and equity tie? Yay.
It does. But, again, just a simple extra prompt. I said, hey. What are you doing in these cells?
And, like, it added these formulas based on my forecasted income statement, and it's giving me an explanation. Right? I'm using the trailing three month average ratio of accounts receivable, the revenue.
That's gonna generate those numbers. So really, really cool stuff you can do, but super important to have that govern agentic finance layer in place to make this so you're very confident that the data is right. Because I can do all this cool stuff, but if it's the wrong balance sheet numbers I brought in, because I'm just doing this on the fly with spreadsheets, That's a huge risk and huge liability. I'm gonna spend tons of time, like, backtracking and seeing where I screwed up. So really cool. Excited to talk to to more of you guys about some of these things, but we can kinda open it up. I know we're right at about time, see if there's any questions, Lauren, Moe, that we didn't get answered that we wanna answer right now.
Yes. Absolutely. Well, thank you so much, Jim. That was a great great walk through. We do have some great questions coming in that haven't been answered yet. So one that we can talk about now is any thoughts on dynamic dimensionality, meaning using drill down or all available transactional data and not the anchoring dimensions defined in cube and typical hierarchies to pull and filter data, I e dashboards or cube fetches, pulling slash filtering data based on invoice number listed in the GL?
It's a great question.
Because we have that drill down capability, all that transactional data is underlying it, and you can get to it really quick. So where you can ask questions at the invoice level and summarize it, as you saw before, I said, hey. Give me my top five transactions.
Same thing with invoice data so you can have it there. So you're gonna have those, like, key business hierarchies that you're always anchoring off of for your different reports. But in those instances where you have those more custom one off reports where you're digging into detail fields like invoice number, Cube can get down there to that.
And worst case, if you just wanna do a drill down and dump it all out, now you have it in a spreadsheet. You can use cloud or something on top of that to go generate some reports off those transactional details without ever having to log back into your GL system or any system. It could be operational Salesforce, operational data warehouse, whatever it might be.
Awesome. Great. So our next question, any more info you're able to share on the engine driving these agents? Also, thoughts on customization for these Cube agents? Meaning, do you expect an agent design or feature, or will this continue to live as a can decision tree or workflow?
Great question. Maybe above my my, pay grade in terms of the folks who could answer that for you. I will say a lot of these agents, we are serving multiple purposes. Like, we have analyst agents that are doing a lot of the work. When you go prompt it to go build a new plan for you, it's going and using that analyst agent to go interpret the data and things like that. So I think having it open allows you to kind of, you know, work in that ecosystem as best you can, but that's something we can follow-up on and get you more details because, I'm not the one actually building all the cool stuff. I just get to show it off.
Awesome. I think we have time for, one or two more questions. So another one we got. Jim, is there any guidelines on the prompts on what things need to always be the prompts or how to make sure your dimensions are laid out properly?
Yeah. Do you and this kinda goes to, I think, Jeremy's question as well. You do not always have to prompt every single dimension. Same as in, like, a a sidebar report if you're building on a spreadsheet, you're gonna pick, like, what are my core dimensions I'm thinking about for this prompt, and it has defaults for the rest.
Right? Like, mine defaults to revenue. If I don't have accounts in there, which is what I'm using most of the time, it defaults to total departments, all entities. So you don't have to prompt and say, I want this for all entities.
It's only when you're getting specific of I'm doing this prompt for the UK entity, but because Q will default to assuming you're looking at the total of everything else or whatever your default member is.
Awesome. I think we're out of time for today.
So I think we can end there. Jim, thank you again, and thank you everybody who joined us for our webinar today. Just a quick reminder, you'll be getting a copy of the presentation as well as a recording sent to your email with some other resources in your inbox later today. So keep an eye out for that, and looking forward to seeing you guys at our next webinar.
Awesome. Thanks, everybody.
For those just joining, we're gonna get started in about two minutes. One to two minutes, we're just gonna wait for people to join and get settled.
Welcome, everybody. We'll get started in about a minute here. Thanks for joining.
Perfect. Well, I think everybody has time to join, so I'm gonna go ahead and get this webinar kicked off and share my screen.
Awesome. Well, good afternoon, everybody, and welcome. Thanks so much for joining us for today's webinar. We're gonna dive into Cube's AI functionality showcase covering what's live today, what's coming, and what it means for your finance team.
Before we get started, just a couple of quick housekeeping items that I would like to go over today. First thing is the live q and a function that we're gonna be having at the end. So we'd love to make this session interactive, so we will have a chance for us to ask questions at the end. Normally, the session will run for about thirty minutes, and we will reserve the last fifteen minutes for questions. So you can submit questions at any time during the session using the toolbar at the bottom of your screen. Feel free to submit your questions at any time, and we will either answer them in the chat in real time or cover them at the end.
Another, quick thing I wanted to mention is that we will be recording this webinar. So after the recording, you will get an email from myself with the recording, a copy of the slides, as well as some other information to dig into as well.
To walk us through today's session, I'm thrilled to introduce Jim Bullis, our VP of solutions here at Cube. Jim has been with Cube for nearly five years, building and leading our pre sales implementation and support teams. He's worked side by side with customers from first demo all the way through implementation, so he knows exactly how Cube delivers value and practice. With that being said, take it away, Jim.
Absolutely. Thank you, Lauren. And we also have Moe on the line. He will be helping answer questions in the chat as they pop up.
So as Lauren said, feel free to fire those away. So let me just get my stuff settled here. Alright. What gonna walk through today is we're gonna talk about the moment we're in.
Right? AI is is here.
People are starting to use it. We're gonna talk through what finance teams are using it for today outside of just the cube context, but in general, how are they embracing it? So maybe just some tips and tricks on what other folks are doing if you're just getting into it yourself. Obviously, we're here for Cube, so we'll showcase, hey.
How does Cube's AI come into play? Where do we fit in in this process? And then finally, showcase the functionality and talk through kind of the gains that you'll receive. So we'll we'll walk through all the AI capabilities that you have with Cube and also combine it with, like, an MCP to utilize Clog or chat p chat GPT on top of it.
So where are we at today? We'll start there.
Right? Traditionally, this has always been the case. So I've been here at Cube for five years. Before that, I spent a decade in FP and A consulting, working with a lot of the legacy platforms in the space.
But, traditionally, finance teams are spending eighty percent of their time just gathering data, reconciling it, formatting it, not analyzing it. Finally, the help is here. Right? That AI inflection point is now.
Teams are starting to use that to reverse that number, right, to spend more time actually analyzing, planning out new forecast, new scenarios based on what's happening.
Managers are looking for ways to limit, you know, increases in headcount. Right? Hey. Hey. Where can we use AI to maybe not need to make that extra hire to get more of our current employees?
And, you know, finance is now the strategic engine of AI in a lot of organizations. Right? Because you sit across all these different datasets. Right? Traditionally, it's always the GL, your ERP.
But what we're seeing is, you know, we have HR systems, CRMs, operational data warehouses that you're tasked with gathering data from. So you're familiar with those. You're helping those other department heads. It's natural you're gonna fill in to stop start, you know, up leveling those broader teams as well.
So what we're hearing in the market, Gartner found that, you know, eighty three percent of professionals are already using AI tools in their workflow. What's not so great is seventy percent four percent actually report using personal or free AI tools that aren't sanctioned by their IT teams.
Great that they're up leveling, maximizing their time, being more efficient, but there's some inherent downsides if you're, you know, navigating outside of your internal processes just around security, sharing data, things like that. But in general, you know, finance teams are seeing massive increases in their speed. Right? You know, doubling or decreasing the amount of time it takes to to close the books, putting out more analysis, so on and so forth. So really fantastic.
What are some of the ways they're doing it? So if you're just starting to scratch the surface, what are some easy ways you can jump in? And I was not some expert coder, anything like that when I started exploring this. You just start prompting.
You can ask AI how to use it. It'll give you good feedback. But some of the basic ways that finance teams are starting to embrace this are just what they do every day anyway. Slide deck generation.
Right? You can pump data into cloud. It'll produce nice looking decks for you with your company's formatting and colors and all that kind of good stuff.
Dashboard generation off that as well.
Big thing finance teams have to do every month, produce the narrative for your numbers. Right? What is driving those variances in the business? What increased or decreased that versus what you expected to happen in that period. Having a layer that can jump in and go and dissect your data and produce some analysis and variance explanations is huge.
Obviously, just ad hoc research and insights. If you can pump data in and look for new insights that maybe you didn't have today, that can really uplevel you and maybe expose some areas of the business you can improve that maybe weren't in the forefront before. So just using the AI that's out there. Again, this has nothing to do with Cube.
You can do these types of things today, so I encourage you to go do that. What are some other areas that teams are interacting with AI? Using it to increase the speed which they can close and reconcile. Right?
You can upload two files, have them compare against each other using AI and find discrepancies, maybe find duplicate journal entries without having to comb through it. Right? You can use AI to do that.
Also, big time saver. We're in spreadsheets all the time. You can have AI help you write formulas. Right?
And some of it's just base like, I'm taking raw and structured data. I have to add some extra columns to be able to filter it the way I want. I used to have to come up with these really lengthy formulas or do a bunch of copy pasting. Now I can just prompt AI and say, hey.
Take this column. If it's empty, do this. If this column says this, do this. They'll build that formula for you, do lookups for you, all that kind of stuff, just saving you time.
Right? You still have to double check it. That's where all your finance know how your spreadsheet expertise comes into play because you can double check and do some QA on it to make sure that actually is producing the results you want. But what used to take you an hour might take you a minute from that perspective.
Really interesting one I heard from some customers that I wasn't doing in my day to day, but makes tons of sense, is stress testing your models. Right? Maybe you have to put together a plan for the next quarter or the next year. You know you're gonna walk into that executive stakeholder meeting or board meeting, and that person who's gonna stress test it and push back against you, do that with AI before you even get to that meeting.
Tell it to be an aggressive, you know, board member who always pushes back and see where the gaps are in your plan or something that might be off. If you're projecting thirty percent growth for the last six six months or only ten percent growth, it can spit that back to you. And maybe you adjust your plan or maybe you just come up with a narrative to explain why you built those assumptions in. So I thought that was really interesting use case that I hadn't played around with, which is, you know, creating some, AI to kinda challenge you and pressure test your plans.
So really exciting. You're gonna see some of these as we showcase Cube's functionality. But but the problem is if you don't have a governed data layer in between, AI has kind of inherent trust problem because AI can be eighty, ninety, ninety five percent right.
But in finance, that's not good enough. Right? And there's a few reasons this comes up. Number one, it it lacks full spreadsheet awareness.
It's really great. You can do really cool things in spreadsheets with AI. The problem is it doesn't always understand the full context of what you're trying to do. So we're talking with a a CFO, and you're saying, hey.
We have to be really careful because we add in a row or change a calculation. It doesn't always understand why that change was made. It might overcorrect for it or not notice it. So having that full spreadsheet awareness to understand what data is missing or or not is really important.
It's probable list probabilistic versus deterministic. Right? It's gonna protect the next word, but that next number might just seem great. It can kind of misstate budgets by hundreds of thousands or sometimes based on how big your numbers are, because it's it's trying to be or it's being probabilistic.
Big one is there's no governance or audit trail. So when we're laying this on top of our data and I wanna share it with my team, if I'm sharing it with marketing, I don't want them to see engineering's payroll data. Right? So I have to be really careful about that.
There's no governance there, and there's no audit trail. Right? If I'm tracking or if I'm making changes to my budget or my plan off in a spreadsheet and then pumping it back into AI, it doesn't always know what changed or have that log. And then finally, a lot of times, I'm it's great in Excel working me individually, but if I'm collaborating with my team across five, six different spreadsheets, like, if Lauren makes an update, I'm not always is that in into my model that I'm using in my AI over here.
So these are some of the the big drawbacks right now today around where it's great, but it's it's falling short in finance. So that's where, right, Q comes in. Right? What you need to make this work is a layer that's bringing your data together that has that governance, that audit trail capability on top of it to make sure that security is in place, your data is good.
You want a layer that's understanding your business hierarchies, your your logic, your multidimensional aspects to make sure that as your data is flowing in, you're not missing data. It's rolled up. It's already got that business logic built in. It already knows how you calculate gross margin or the rule of forty, right, that's already in a system that it can read off of and work with.
And then finally, you need a live connection. So as you're interacting and updating plans, you can see the impacts of that flow through. So that's where come Q comes in. It's the first AI native, kinda workflow native FP and A platform finance layer that's out there.
And what that means is when we say workflow flow native, sure finance loves to work in spreadsheets, but you also collaborate across the org. And you have some users that maybe don't even have Excel installed on their machine, you need a way to collaborate with them. And that could be through collaboration tools like Slack or Teams. That could be through a web interface where you can send them a link.
They can go view a report, make updates to a plan there. Or for those more savvy team members, it could be an Excel or Google Sheets, but we need a spot that works across all those. We wanna keep finance in the driver's seat. We don't wanna burden the finance team with tons of code and syntax they have to learn.
We have to make it really easy for them to do new modeling, do new planning, but synchronize it and allow the team to see those impacts. And then you need trusted AI for finance, which means you know all the data's in a good spot. You know it's governed.
You know you have all the transactional detail there underlying it. So if you wanna go trace down into a detail, you have every insight there down to the transaction level or aggregate up aggregated up across those business hierarchies that we built out.
So that's in general how Cube works. Right? We've got this agentic finance layer in the middle, which is Cube. We're bringing your various datasets together, and that can be source systems like you have on the left.
Right? NetSuite, Sage Intact, SAP, any system out there. But with finance, again, you're pulling together from data from tons of different systems, your CRM, your HRIS, your operational data warehouses. Right?
You need that all in one layer so the data can talk to each other and be lined up in the same time hierarchies, the same metrics, everything like that. But when you're actually interacting with it, where you interact with it differs based on the audience or the team member you're working with. That could be an Excel for finance or PowerPoint, or it could be somebody asking a question via Slack. It happens to me all day.
Right? Or maybe your team's on Google Sheets, or maybe it's a mix of both. Right? The broader org uses goo the Google Suite, but the finance team's in Excel.
But, again, nowadays, we're taking advantage of chatty, petite, flawed. We wanna have AI help us drive these answers, or I wanna send somebody a link. Right? I want them to log in to a nice workspace that's been curated for them to come in and look at the right numbers.
I don't wanna have to send them a spreadsheet that gets stale the next day. So Cube is that financial intelligence layer that sits in the middle, has our agents working, and allowing you to interact with that data wherever you need, you know, aim to become headless here in terms of, you know, allowing you to interact with that from wherever you need.
Alright. So I mentioned agents. We've got this financial intelligence layer on top of it. We've created the, you know, agentic finance layer now that we've built in all these agents. And we've brought them together in these different teams.
So there's tons of agents behind the scenes. But based on what you're trying to do day to day, you're gonna have your data manager, your analyst, your planner agents, and your business partner agents. And what that looks like is your data managers are helping you maintain your model in the back end. It's making sure that all of your accounts are in place, that the mappings are good to go. Your data has been loaded, and everything is up to date.
Next is I need to analyze my data. That's where I can prompt a question. It'll find an insight for me. It'll help me sort out your biggest variances and dive into it from that perspective.
When it comes to FP and A, the second letter in that acronym is planning. Right? So we want a really cohesive planning agents that allow us to go ahead and expedite our plans and allow us to cut out a lot of the initial legwork it takes to to build those out. And then finally, we need to collaborate with all our business partners across the org.
We want really good AI agents that are gonna help us collaborate, build compelling dashboards really quickly. And then sitting across all this, we have the MCP as well to allow you to connect in your LLM tools to help build all this out.
Right? So that's that Cube plus Claude. Again, ChatGPT, anybody that's using that MCP, we can go ahead and interact with that. But we are building that agentic finance layer in the middle that has all of our agents working to help build out your plans, work in our workspace. But if you wanna spin up a PowerPoint deck using Cloud, you have the data in place. If a team member is looking at that data via Cloud or ChatGPT, it recognizes who they are and will restrict it to the slice of the pie that they can see. Again, so if they're in the marketing team, they can't see engineering's data as they're prompting and asking questions.
Right? So tons of cool stuff you're gonna see today, but I'm even more excited for what's gonna come in the next few months here. You can see these four teams of agents we have set up. Again, there's a lot more behind the scenes, but we're we're looking to improve it across the board, make it easier to maintain your models.
There's gonna be some connection setup agents to expedite how fast we can, you know, get new connections spun up. Probably the one I'm most excited about is the planner agent. So what you're gonna see me do today is generate statistically driven forecast looking back across trends and growth rates, things like that. But something that's really cool that I always hear from finance teams is, hey.
I wanna spin up a new forecast. I wanna hold my personnel cost steady, because we're using AI now. I don't need to have those extra headcount in. We're gonna increase revenue by three percent, and we wanna increase our marketing spend by five percent.
I'll be able to talk to a planner agent. It'll spin up a new scenario for me. I can see what it looks like that across my full three statement model without having to do any of the legwork aside to prompt and then doing some variance analysis there. So really cool stuff.
It does not end with where we sit today. We are building out tons of new agents to help you build out, you know, slide decks, so on and so forth, help with month end close processes and reconciliation, help you build more advanced formulas, in the back end of the platform. So really exciting from that perspective. And what we see our customers who have already embraced AI in with Cube is they're flipping that narrative.
Right? Eighty percent reduction in manual time. They're gaining more insights. Things that used to take hours or days are now taking minutes.
Day one, finance can come in and own the process. You don't have to spend eighty hours learning how to build reports. If you're in spreadsheets, you can generate it really quickly. If you're in using LLMs, you can generate that quickly as well.
So really exciting to showcase these for you as we go through. And what we're gonna show today in our live demo is we're gonna walk through kind of a day in the life of a finance team member and what they have to do every anyway. Like, none of this is really AI driven. If you wake up early in the month, you have to go ensure the data is all coming in, updated, mapped.
Then you're gonna be tasked with discovering your top variances. Hey. Budget was this. Actuals is this.
Why? Then you have to create the narratives around those, answer questions from team members about, hey. What was my marketing spend last month? You're driving new insights.
You end up updating all your reports, sending those out to the team. You have to go update your forecast for the new month, maybe doing a rolling forecast. Then you got a board meeting coming. Right?
You have to put your board and assets together. And then finally, just for the sake of this one, we're gonna build a new balance sheet forecast. It's something we've never done at this organization take organization I work for, but I wanna go build a new balance sheet forecast. Now, again, these are all tasks that I have to do today, but we're gonna show you how Cube's AI folds into each of these to make your life tons and tons easier.
So we're gonna go ahead and dive in. So let me end my slideshow here, and I'll pull this out of the way.
And what we're gonna do is I'm gonna start with that first step, which is it's end of the month.
If I'm just doing things in spreadsheets today, I have to make sure everything's flowing into Cube appropriately or just into my reports appropriately. With Cube, have all your data sources synced. And a big one is I have new accounts, new vendors that are flowing into the system. I need to make sure those are mapped incorrectly.
I can go do that manually if I want to. Like, a new account pops in. I can go map that in. If it's just one, maybe that's great.
But if I have a hundred vendors come in, it's gonna be a bigger deal. So we have built in our first agent that we'll see today is our mapping assistant. So in this case, if I come back here, it says it found a new account, which is one zero four zero zero investments and securities.
What Cube does based on how my map renewal is set up is it's detecting, hey. Based on the account number that I see, the description, where do I think it would map into the model? And it's saying, hey. I think cash and equivalents is a good spot.
Right? Now, obviously, you're you're in finance. You're gonna know and detect that. But maybe it's a list of vendors, and you categorize those by what type of vendor it is.
Is it a software vendor? Is it a maintenance vendor? Whatever it might be, it will go look at that and make an initial interpretation to save you that initial time of, hey. Where should this go on my hierarchy?
So then I can follow prompts. It'll go generate the proper mappings for me, and I can just click a button to get those into cubes. So trying to save you time on that front end. Like, what are the things that always, like, stink in your process?
It's identifying those new accounts that flow through. Where do I put them? So on and so forth. Cube is catching those for you, suggesting mappings for you that you can then run with.
So, again, month end, use an AI to save you time right out of the gate.
There you go. There's that new screen. Again, if it's just one, it's probably pretty easy for me to go interact with that. But the next step, month end, we got our data flowing in.
We feel good about our mappings. As I mentioned, we need to go analyze our data. And one of the first things I need to do is go find my top variances for the month. So what chain or I expected this in my budget for May.
Well, here are the actuals. Where are those top variances? We built out a smart variance agent, which will go through, and you can compare any scenarios. But most of the time, for our customers, it's actuals versus budget.
And you can pick the specific time frames. If I wanna do entire quarter, entire year, I can go ahead and do that. Let's just go do May here.
I can get really detailed specific department. I can set minimum, so only show me anything over five thousand dollar variance or a specific percentage. But we have built agents that are gonna sweep through every single intersection of budget versus actuals in our model across all of our business hierarchies and kinda serve those variances up to us on a silver platter, saving us hours of time. So here you can see, hey.
I got my top variances six hundred twenty two k. It does it by both dollars and percentage. It gives me a lower AI right up at the bottom. But what I love here is it stack ranks and sort these for you at every level.
So I can see right out of the gate, and usually is the case, my revenue forecast is off. Our budget actually, in this case, it was under. So we expected to do seven million. We did eleven.
But it doesn't stop there. I'm gonna go on and go into details. This is where the beauty continues here as, hey. I wanna go look at the specific GL accounts that broke out, see, like, was it my implementation revenue, my platform revenue?
Okay. I can see my platform revenue is my top variance, four million versus five hundred thousand. I have other hierarchies in here. I have markets and products or what business units, whatever it is that's specific to your business.
But I can click in the markets here. Now it's breaking it down and sorting those by markets. So I can see my top variance was the east market, and I'll go look at look at by product. So within a few seconds, if I wasn't yapping here, I got to my top variance sorted.
I can, you know, navigate, dig up and down to go figure that out. So tons of time saving using our SmartVariants AI agent to go sweep through your your actuals, your budget, find those top variances so you can start explaining those. So really, really cool from that perspective.
Great. Now I just need to start driving both narrative and analysis. And I mentioned before, we're workflow native here at Cube, which is we meet your stakeholder where they wanna be. So when it comes to asking questions and using AI from, like, prompting it on our side, we have the workspace here. I can ask questions in a lot of different interface. So I can ask it through the MCP and, like, Claude.
But if your team likes to work in Slack or Teams, you can ask it there. I'll show that in a second or in our cube sidebar in our our spreadsheet add in. So I'll start here, and let's just go start by asking starting in May twenty six, Looking back, I'm gonna go look I wanna see what expense has grown the most over the last three months compared to the prior three months. So looking back, what expense has grown the most over the last three months versus the prior three months.
Now what's cool about this is I'm not these are not, like, quarters or anything like this. I'm just prompting it with natural language because that came up.
While I'm doing that, let's go down to Slack and show what it looks like to interact with it there. So I've got my Cube AI assistant here. I can come in and ask a question. And maybe I'm just curious. What I was looking at the revenue before. What were my five biggest revenue transactions in April twenty five or twenty six?
Now Cube is gonna go prompt it, dig that up. And for me, maybe I'm the head of sales, I don't need well, they probably know the biggest deals or whatever, but but I don't need a license to go back into NetSuite or SAP. I can just prompt it here, and it'll spin that back up. Same thing.
Maybe you're not a Slack shop. You're a team shop. And every month, I get asked by the head of marketing, what did I spend last month? Well, I can just say, what did marketing spend last month?
Right?
Maybe I just pay post this in the Teams or Slack, and I copy paste it. Or if I want to, I'm I trust that person to be savvy.
It's gonna go give me that number really quick. Right? My marketing was eight hundred thirty three k. It broke it out between paid ads, promotional, and conferences.
If I wanna prompt it and go deeper, I can. We'll go back up to our web interface here and our workspace. And comparing the last three months so this is kind of a weird request, but this is the type of stuff that comes up in finance. March through May versus December through February, what grew the most.
Right? So it's telling me, hey. Contractors and consultants, it gives me a little write up. Paid ads is the second biggest increase.
It gives me, you know, some details there. But what's cool is I can click on this, and Qubel generate the report it used to drive that analysis. Because, again, anytime you're dealing with numbers, you wanna make sure you double check it. So you can come gut check it, and you can see here it's saying, hey.
Take February twenty sixth, the trailing three month total, and generate that up.
Now if I wanna come in here and take a peek and look at my contractor cost, right here, I can drill down and keeps gonna spin up all the transactional detail that makes up that number. So if I go and collapse this here, right, I can see every transaction. I can see the accounts that it hit, the contractor and consultant account. I can see the memo field, the transaction numbers. And, again, the head of marketing or the head of sales, they don't need a license to my g GL system because they're interacting with Cube here. They can just go ahead and ask the question, get the details, even the transaction level from that perspective.
So really slick being able to use AI to drive that deeper level of analysis. And then I can't remember if I looked at the results from Slack. So here are my five bigger biggest revenue transactions. Right?
And it pulled back this little chart here. I can actually expand it because there looks like there's six hundred and twenty four lines of transactions for this month. It gives me the top customers. Implementation revenue dominates the top three.
Those are my biggest, you know, revenue drivers. So those are completely different GL accounts. But because I just said revenue transactions versus platform transactions, it's giving me everything underneath it. So allows me to prompt further if I want to from that perspective to kinda dive in.
So really, really cool being able to service your team wherever they see fit from that perspective. Alright.
**** ** nose real quick.
As we're going through, feel free to drop questions in the the chat, the q and a.
Where do you see fit? We'll we'll either get those answered by Moe, or we'll we'll answer them as as we get going at the end of the session here. So I have come in. I've gone ahead and updated my data.
It's all flowing through here at the end of the month. I used the mapping assistant to make sure everything's accounted for.
I found my top variances using AI with Cube. I've been asking Cube questions via our AI analyst in Teams, Slack, in our workspace. Now I need to go update my reports.
I can also go ask Cube questions in our sidebar. So I'm gonna come into Excel. Just maybe I'm spinning up a new report from scratch here. I can click new.
Or, actually, before I do it, let's go ask it a question, and I'm gonna say, you know, what time shortcuts are available with Cube? So I can ask the same data questions I was asking before, but what's great is I can query our knowledge center. So if you're asking, like, hey. I have Cube.
I wanna build a new budget. How do I get started? Things like that. You can ask questions as well here in your sidebar.
If you're finance and you're always in Excel and you don't wanna go back to Slack or Teams or the web, you can just ask those questions here, and it'll spit those back for you. Because maybe I wanna build a new report, and I wanna see, like, a trailing three, six, twelve months. Right? I can go add that in now.
So let's go build a report real quick. I'm gonna go pull together a p and l, and I'll show you how easy this is to navigate. We're gonna go pull in, you know, my totals for twenty twenty four and twenty twenty five just to make it a simple report here, and we'll go ahead and add everything else. Right?
So I'm in my spreadsheet. Again, workflow native.
You're a finance person who's just doing things day to day. Right? This is you know, I could prompt you to build build this out for me. This but this is, like, a big view or report of data.
I can throw my formatting, my colors on here. But as I mentioned, I wanna go add in some time shortcuts that I just prompted our knowledge center about. And, you know, let's go see May twenty six trailing three months, and I'll just do a couple different ones. Right?
I wanna see trailing six months. So we've built in all these really cool time shortcuts. This one is just t x periods. You can put in literally anything you want.
Like, if somebody's like, give me the last twenty one months. Obviously, that's not one you'd usually have prebuilt out. Cube has that functionality, and then you can go up to, like, two hundred if you wanna go look back. Cube's gonna go ahead and and just interpret what we're trying to pull in and go calculate those values for me.
So now I have I mean, kind of a goofy report here, but I have my twenty twenty four total, my twenty twenty five total, and a bunch of trailing totals in the middle. But the key here is I'm not doing any more manual work. I'm just prompting Cube. Hey.
I would like this in my report. Please give it to me. And Cube drops those numbers in there. So there's no big prehistoric formula.
Like, that looks like a v lookup. Have to go run. I don't have to dump a bunch of data and do any math. I use Cube to figure out, hey.
What's the best way to interact with this? And now I can go ahead and query my data using time shortcuts. We have year to dates, quarter to dates, lifetime to dates. You can look forward.
If I'm doing a rolling forecast and I wanna see my total of the next twelve periods, really, really slick, you can go ahead and build that in.
Alright.
Not the fanciest, most formatted report, but I could clean this up, push it up to the web so that my team can see it there. But I'm I also like to share reports via dashboard so my team can log in every every week, every month, and go ahead and, you know, interact with their data. So I have a nice executive dashboard I've already built up. We're gonna use AI to help us build a new dashboard widget because I don't have time to, like, go sort through. Our dashboards are really intuitive, but it's always easier to get, like, a good starting point. So I have a nice blank spot here for a new widget I wanna go add in, and we're gonna go build with AI.
So let's go build a line chart maybe for the last eighteen months of revenue data. And I know I have data by market. So line chart for last eighteen months of revenue data showing different lines for my different markets.
Right. Cool. Simple prompt there. I can go ahead and generate that widget just to showcase tons of different, like, charts and widgets you can have here.
I can go ahead and filter so I can pick and choose, hey. For each visualization, do I want this to update when I filter into it based on the different time frames, things like that? Keeps going ahead and generate those values for me, showcase it. But I used AI to build a lot of these widgets.
Again, it's a drag and drop experience. It's really fun and easy to be able to get in here and create different visualizations. But I can go ahead and, you know, have Qiub do Qiub do that for me. So you can see it's it's coming through.
It's generating, what it's how you it's interpreting your your prompt, kinda giving you a thread as you go through. And this looks pretty good. Right? It asked for eighteen months of historical data line charts.
I can go ahead and and throw that into my dashboard. And as quick as that, I've got a new widget ready to go. So, again, using AI across the board to help you really, like, maximize your time and be really efficient, but also maximize your impact with your different stakeholders. So, we'll cancel this.
Maybe I'll change that later.
We've updated all of our reports. Those are the data's flowing in. We found some insights, some narrative that we wanna drive things out with. Next, I wanna go update my plan.
And, usually, what's a huge pain for Teams is just having a starting point. Maybe it's I wanna start a new budget. I'm starting new forecast, or I'm doing a rolling forecast. We just flipped over the month. I need to figure out some data for that last time period. Cube's got smart forecasting capabilities where we're gonna use your all your growth rates, your trends, your seasonality, and you can have Cube spin up new plan data for you using AI, which is great because I can come in here, and I built one before. I called it Cube AI demo new forecast.
It already has something. I guess I'll just create one. I will call it Cube forecast.
You can pick what you wanna forecast. Right? In this case, I'll just do everything. Most of the time, I might just target, you know, income statement accounts or maybe I'm just doing revenue in this case.
You can figure out what I wanna populate. So in this case, maybe I wanna generate, like, a rolling twelve month. I wanna do that new month data. We'll do June through May.
Or why not? Because he was doing it all for me. I don't have to do the manual legwork. Let's just go through the end of next year.
Right? Super painful to do that today. Most folks, if they're doing forecasting at all, are usually doing it through the end of the current year. Now you have Cube.
Just tell it, hey. I wanna go up next eighteen, nineteen months or farther. You can have a living, breathing forecast that you can go ahead and update. Right?
You're maybe not asking your stakeholders to go update that, but if you're just working with the finance team and you wanna see what things look like in q four of next year, have Cube spin that up for you.
Finally, here, you're gonna pick what you want to inform your model. Ninety nine percent of the time is actuals, but how much history do I wanna use? Maybe twenty one and twenty two were kinda weird years. Twenty twenty three is when it kinda leveled off and got kinda normal for us. Maybe there's some anomalies back then.
Now I can go ahead and pick the custom date range that I wanna use to inform my forecasted values. So highly encourage every kind of Cube customer to go ahead and use this functionality. You can use it as a starting point for your plan. Right?
Hey. I want something I want my team to react to when they start the budget. You can use it as a gut check. Hey.
We already ran our budget. What does AI think we would do? And I can stack those two next to each other, run some variance reports, and see if AI has things were way off in some spots. So really, really cool being able to use AI to help build new passes of your plans.
So highly encourage that. Alright. So we're in the middle of the month. We've ran our reports.
We're updating our narrative.
We got dashboards. All of our internal stakeholders, they can go interact with the data. I've created a new plan. Now I got that board meeting come up coming up.
I need to prep all my assets for that. And as much as I'd love them to log in to Cube and look at our dashboards, we all know that these board members aren't gonna, like, log in. They don't wanna license to a new system, and maybe even your CEO doesn't wanna do it. Great.
Let's go use our MCP and Claude to build some really compelling assets for the team. Right? So what I'm gonna do is I was working with Claude before, and I did some some prompting here. I was building out my deck, but, you know, here, I have was building out some, like, interactive PowerPoints.
So I I because Claude is hooked up to queue, I can run some really simple prompts, which or, of course, I do have this spun up so you can see what the actual output looks like. Let me come in here and take a look.
Second here.
Notice this is the cloud problem, not a key problem. Let's pop this open.
Alright. So I asked Claude, hey. Take all my financial data. Maybe I uploaded last month's or last quarter's board deck, and I said, hey.
Just take this and go build this out. I did this all with simple prompts saying, hey. Take this. Give me a k q one KPI summary.
Here's the metrics I wanna report on. Here's the widgets. Again, if I already had a deck that I've used in the past, I can use that. I can use my company's formatting.
But I built this deck with just a simple a few simple prompts, and then I went through and made some adjustments.
Right? I said let me get rid of this preview. Hey. I'm missing kind of a summary TLDR slide, so throw that in.
Like, analyze the data. I'm if I'm a diligent finance team member, I'm gonna go make sure that it matches what I actually think. But if it can produce the various analysis, some of the stuff like that that I need to spit out, I can do that. Or the first pass I did at it, it generated a bunch of charts and widgets, but it was showing unformatted numbers.
It was just, like, ten million with no commas. I was just like, hey. This doesn't look right. Go adding some commas.
Be more efficient with the space. If it's in it makes more sense to use thousands or millions. It's incredible how fast you can put together board decks now utilizing cloud on top of of cube data. The other thing we we have customers doing is they put together, like, really cool HTML interactive dashboards.
Again, for those trusted team, like, board members that aren't gonna log in to a system, you can have it prompt and create a, you know, HTML, like, interactive experience with your data. Again, I have never built a website in my life.
I literally have all the MCP and cube and Claude sitting on top of it. And I just said, hey. Go build me, like, an interactive dashboard on this website. And I said I wanna be able to, for the first tab, do it by q one, and I can go interact and click in the different months.
I want to be able to filter by by different products, markets, really basic prompting. And I got this really cool output where they can come in and start to play around with their data. And I'm just playing with it, and I said, by default, it gave me these kind of selectors. I was like, let's spin up another one, and I'll use pick list instead of literally, I said is spin up a new one for last year, same metrics, same widgets, just instead of at the top using those selectors, make it a pick list.
And within a minute or two, I had this really cool thing that I could share with a team member or board member or something like that so they can go in and and play around with the data in a better fashion than just a board deck that I might be sending out. So depending on who you're collaborating with and how much power you wanna give them, you can get data in their hands really quickly in a really compelling, really professional looking format. Again, feed in your company's colors, basic formats, use decks you've used in the past so it can match it, but really, really cool and compelling from that perspective.
Alright.
Last item I'm gonna call out here is my CFO prompted me to go start forecasting my balance sheet. And I worked with some customers that were really diligent about balance sheet planning. Most don't. They're like, we don't really plan the balance sheet. Like, it's daunting to even think about where to start to go start planning out, like, doing a balance sheet budget or forecast.
What I did here is I had Cube go ahead and pull together all my data. So I just said, give me my income statement and balance sheet for the last couple years. So I I hit fetch with Cube. It pulled it into the spreadsheet. You can see here, I've got twenty four, twenty five. And then out in twenty six, I've got my income statement, which I'm regularly forecasting.
But down below, I don't have any, you know, balance sheet data. I just came up and said, hey, Claude. I have it in Excel here. I wanna forecast my balance sheets. Please use my existing forecast in this file for June to December from my income statement. Again, Cube centralizing that data. Make sure you have the right data there, the right security on it, the right hierarchies.
Generate a balance sheet values based on formulas. And then in column AL, which is this one, I want you to give me, like, a little description of what you did for that because this would probably gonna be a great starting point for me. It's gonna be give me ninety percent of the way there. I'm gonna change the formulas and calculations.
So I ran that this morning. It came up with the numbers. First thing I looked at is does this assets and liabilities and equity tie? Yay.
It does. But, again, just a simple extra prompt. I said, hey. What are you doing in these cells?
And, like, it added these formulas based on my forecasted income statement, and it's giving me an explanation. Right? I'm using the trailing three month average ratio of accounts receivable, the revenue.
That's gonna generate those numbers. So really, really cool stuff you can do, but super important to have that govern agentic finance layer in place to make this so you're very confident that the data is right. Because I can do all this cool stuff, but if it's the wrong balance sheet numbers I brought in, because I'm just doing this on the fly with spreadsheets, That's a huge risk and huge liability. I'm gonna spend tons of time, like, backtracking and seeing where I screwed up. So really cool. Excited to talk to to more of you guys about some of these things, but we can kinda open it up. I know we're right at about time, see if there's any questions, Lauren, Moe, that we didn't get answered that we wanna answer right now.
Yes. Absolutely. Well, thank you so much, Jim. That was a great great walk through. We do have some great questions coming in that haven't been answered yet. So one that we can talk about now is any thoughts on dynamic dimensionality, meaning using drill down or all available transactional data and not the anchoring dimensions defined in cube and typical hierarchies to pull and filter data, I e dashboards or cube fetches, pulling slash filtering data based on invoice number listed in the GL?
It's a great question.
Because we have that drill down capability, all that transactional data is underlying it, and you can get to it really quick. So where you can ask questions at the invoice level and summarize it, as you saw before, I said, hey. Give me my top five transactions.
Same thing with invoice data so you can have it there. So you're gonna have those, like, key business hierarchies that you're always anchoring off of for your different reports. But in those instances where you have those more custom one off reports where you're digging into detail fields like invoice number, Cube can get down there to that.
And worst case, if you just wanna do a drill down and dump it all out, now you have it in a spreadsheet. You can use cloud or something on top of that to go generate some reports off those transactional details without ever having to log back into your GL system or any system. It could be operational Salesforce, operational data warehouse, whatever it might be.
Awesome. Great. So our next question, any more info you're able to share on the engine driving these agents? Also, thoughts on customization for these Cube agents? Meaning, do you expect an agent design or feature, or will this continue to live as a can decision tree or workflow?
Great question. Maybe above my my, pay grade in terms of the folks who could answer that for you. I will say a lot of these agents, we are serving multiple purposes. Like, we have analyst agents that are doing a lot of the work. When you go prompt it to go build a new plan for you, it's going and using that analyst agent to go interpret the data and things like that. So I think having it open allows you to kind of, you know, work in that ecosystem as best you can, but that's something we can follow-up on and get you more details because, I'm not the one actually building all the cool stuff. I just get to show it off.
Awesome. I think we have time for, one or two more questions. So another one we got. Jim, is there any guidelines on the prompts on what things need to always be the prompts or how to make sure your dimensions are laid out properly?
Yeah. Do you and this kinda goes to, I think, Jeremy's question as well. You do not always have to prompt every single dimension. Same as in, like, a a sidebar report if you're building on a spreadsheet, you're gonna pick, like, what are my core dimensions I'm thinking about for this prompt, and it has defaults for the rest.
Right? Like, mine defaults to revenue. If I don't have accounts in there, which is what I'm using most of the time, it defaults to total departments, all entities. So you don't have to prompt and say, I want this for all entities.
It's only when you're getting specific of I'm doing this prompt for the UK entity, but because Q will default to assuming you're looking at the total of everything else or whatever your default member is.
Awesome. I think we're out of time for today.
So I think we can end there. Jim, thank you again, and thank you everybody who joined us for our webinar today. Just a quick reminder, you'll be getting a copy of the presentation as well as a recording sent to your email with some other resources in your inbox later today. So keep an eye out for that, and looking forward to seeing you guys at our next webinar.
Awesome. Thanks, everybody.