How AI Automation Saves FP&A Teams 70 Hours Monthly with Gabi Steele
In this insightful episode of the Future Finance Show, hosts Paul Barnhurst and Glenn Hopper explore the evolving landscape of finance, data analytics, and automation with Gabi Steele, a seasoned data visualization expert and co-founder of Preql. The discussion navigates the challenges finance teams face in building data-driven cultures, leveraging AI tools, and creating efficiencies in data management.
Gabi Steele is a graphic designer turned data visionary and entrepreneur. She is the co-founder of Data Culture and Preql, a platform streamlining data metrics for businesses. With a background in data visualization at The Washington Post and a passion for making data accessible, Gabi brings valuable insights on transforming data into actionable business strategies.
In this episode, you will discover:
How finance teams can leverage AI tools to reduce manual data processes and improve accuracy.
The importance of storytelling and accessibility in data visualization.
Ways to foster a data-driven culture within organizations of all sizes.
Challenges and solutions for integrating AI with financial planning and analysis (FP&A).
The future of low-code/no-code tools in democratizing data analytics.
Gabi Steele offered a masterclass in how finance professionals can embrace the transformative power of data and AI. From simplifying complex data processes to fostering a culture of analytics, Gabi underscored the importance of aligning teams, tools, and technology for smarter decision-making.
Follow Gabi:
LinkedIn: https://www.linkedin.com/in/gabriellasteele/
Website: https://www.preql.com/
Join hosts Glenn and Paul as they unravel the complexities of AI in finance:
Follow Glenn:
LinkedIn: https://www.linkedin.com/in/gbhopperiii
Follow Paul:
LinkedIn: https://www.linkedin.com/in/thefpandaguy
Follow QFlow.AI:
Website - https://bit.ly/4fYK9vY
Future Finance is sponsored by QFlow.ai, the strategic finance platform solving the toughest part of planning and analysis: B2B revenue. Align sales, marketing, and finance, speed up decision-making, and lock in accountability with QFlow.ai.
Stay tuned for a deeper understanding of how AI is shaping the future of finance and what it means for businesses and individuals alike.
In Today’s Episode:
[01:48] - Introduction to the Episode
[08:23] - Financial Modeling World Cup
[19:37] - Discussion on AI Subscriptions
[25:53] - Meet Guest: Gabi Steele
[26:59] - Data's Role in Finance
[33:58] - Leveraging AI in Data Management
[39:07] Democratizing Data Access
[44:57] Building a Data-Driven Culture
[49:45] Challenges in SMB Data Integration
[59:51] Fun Segment: Get to Know Gabi
Full Show Transcript
[00:01:48] Host 1: Paul Barnhurst: Hello and welcome to another episode of Future Finance. Glenn and I are going to talk about some of the things we've seen over the last few weeks, since our last episode. Some exciting news announcements, but we're going to start with a little bit of fun here. Some of you may have seen this on my LinkedIn feed, but a guy by the name of Duke Henninger. He put a post out saying he asked, based on everything to do about it, to create a picture of him, and it came up with this very professional picture, but it had a pink shirt. He's like in the pink is from my daughter accusing AI. That's how he described it. So I'm going to show you for those who watch it on video, I'm going to show you what mine looks like as well. And we're going to have Glenn do it live here in a minute. If you listen to the audio, you might want to switch over to the video for this part might be a little hard, but we'll do our best to describe it. So you're ready for me to share what I look like, Glenn?
[00:02:35] Host 2: Glenn Hopper: Yeah. Let's do it. I mean, I'm certain they're going to capture the beard.
[00:02:38] Host 1: Paul Barnhurst: We shall see. So, Glenn, can you see the screen there?
[00:02:43] Host 2: Glenn Hopper: That's not you, Paul.
[00:02:44] Host 1: Paul Barnhurst: What? The ferret and the stuffed animals and the all girl colors. Is it me and the brown curly hair?
[00:02:50] Host 2: Glenn Hopper: You're not the only one using your ChatGPT. I would imagine.
[00:02:54] Host 1: Paul Barnhurst: My daughter probably uses it. I don't know, four hours a day. I use it half an hour so you can see who it picked up. But what gets better is let's have a little fun with a couple of the comments while we're at it. So a few people did it. Mark Proctor and his comment was, I hate the image because it has me using a mac and I hate you can't use Excel on a mac because he's an Excel guy. Not to mention it didn't worry about the fact that he has a finger in front that's detached with two fingernails and 12 fingers. Say, I don't care about that. It's the Mac, that's the problem.
[00:03:30] Host 2: Glenn Hopper: Well, is he left handed because the mouse is positioned for a left handed user? It would be interesting if it did actually pick that up.
[00:03:36] Host 1: Paul Barnhurst: That's a good question. Let's ask him right now. We won't get the answer today, but we'll just put it in the chat here real quick.
[00:03:43] Host 2: Glenn Hopper: While you're doing that, is there anything from you that appeared in that picture, or is it all your daughter's stuff?
[00:03:49] Host 1: Paul Barnhurst: Let me go back and look again. I didn't see anything. But we'll do one more time. Mouse is for a left handed person. Are you left handed? And we'll just see what he says. All right. Reply. So let's go back and look at my picture one more time. You know, not really because I do use mine daily, just about. But she has to be 95% of use. If I let her, she'd literally spend 12 hours a day on my account. And I'm not exaggerating.
[00:04:17] Host 2: Glenn Hopper: It's pretty funny looking at like, how awesome would it be if there was like one like financial chart in the background on the wall amidst the pictures of the parents.
[00:04:25] Host 1: Paul Barnhurst: And that's what I'm trying to see. And I'm just not I mean, I'm looking really close at the moment and I'm not seeing it. I would love that. Like if, you know, on the left hand side, all of a sudden there was Excel, right? Or a chart or a podcast mic or, or something that describes me. I think one of the challenges is when it comes to podcasts, I do all my transcripts with Claude, I write, I use both, I do use ChatGPT quite a bit for finance stuff, more course building and things like that. But at last, nothing. But the comments are great before we do yours. I love the couple of things. Someone said at least the hair color matches. The response was yeah, it's the right length but the wrong location. And then someone shared, you know, theirs was like a slimmer me. And then if you moved much of the hair from top to bottom of head, it might be a better likeness. Not so sure about that background. And someone else. If only they gave her a beard so you can see a theme there. I love your room. Paul. Looks cute. So then, all right, it's your turn to share. Glenn, I'm going to stop sharing. Go in line. You're going to ask it?
[00:05:34] Host 2: Glenn Hopper: Yeah. Alright, so let me present, share a screen.
[00:05:38] Host 1: Paul Barnhurst: We'll see if you really do as much eye stuff as you talk. Or is Han Solo going to come up.
[00:05:44] Host 2: Glenn Hopper: Yeah. All right. So I am now sharing my screen you see on that. Yep.
[00:05:49] Host 1: Paul Barnhurst: What he shared on LinkedIn. This is the exact prompt I used. So it was from all that you know about me. Create an image of me. And I did put it straight in ChatGPT. But Dall-E is what they use. So.
[00:06:02] Host 2: Glenn Hopper: Yep, so we do have to go back to GPT.
[00:06:06] Host 1: Paul Barnhurst: Yeah. Go to GPT. That's where I did mine. See we're learning on the fly. These are the things that you guys need to know as finance professionals, right, Glenn? I mean forget the charts and excel and all that stuff. It's about having a little bit of fun.
[00:06:22] Host 2: Glenn Hopper: So from all that you know about me.
[00:06:24] Host 1: Paul Barnhurst: Create an image of me.
[00:06:25] Host 2: Glenn Hopper: So it's interesting that dall-e that so you know Dall-E doesn't carry over because this does have a memory. I'm sort of scared to see what the memory is going to be.
[00:06:33] Host 1: Paul Barnhurst: Yeah. The memory we now know it doesn't carry over to Dall-E. That's why I was wondering, will it carry over? My daughter gives it. Asked ChatGPT so much stuff. I have to dump the memory pretty often. That tells you how often she uses it. And I'm not a paid account, so not the 300 or whatever we'll talk about in a few minutes. Okay. Not bad.
[00:06:55] Host 2: Glenn Hopper: Yeah, yeah.
[00:06:57] Host 1: Paul Barnhurst: It's a little bit of gray in your hair.
[00:06:59] Host 2: Glenn Hopper: Much stronger chin than I have.
[00:07:01] Host 1: Paul Barnhurst: Yeah. And you're a little more clean shaven. You got a little bit more of that little bit of that shadow going in yourself. A little bit of a beard.
[00:07:08] Host 2: Glenn Hopper: Quite the coif to do, though.
[00:07:10] Host 1: Paul Barnhurst: I like how it has a lot of AI and graph. It makes it clear that you're kind of an AI guy. But notice once again, the computer looks like a mac.
[00:07:18] Host 2: Glenn Hopper: Yep. Well, I'm a mac guy.
[00:07:20] Host 1: Paul Barnhurst: You are a mac guy. Okay. So got that right. I was wondering if it's more just kind of. It does that deliberately.
[00:07:26] Host 2: Glenn Hopper: I haven't put any pictures of, you know, myself or my office in there. I'm not in my office today. But, you know, those bookshelves back there are kind of similar to the ones that are in my a little bit.
[00:07:35] Host 1: Paul Barnhurst: Of a brick in your office. It's darker, but this kind of has the brick older feel a little bit like your office?
[00:07:40] Host 2: Glenn Hopper: Yeah. I don't know how it would have that. This.
[00:07:43] Host 1: Paul Barnhurst: Maybe it would be the random, but.
[00:07:44] Host 2: Glenn Hopper: Yeah. Or it could be really scary that we don't realize how much these AI things know about us.
[00:07:50] Host 1: Paul Barnhurst: Well, It's like perfectly like if you can see my picture, right? There's like no bend in it. It's like this perfect model hand. Although the veins. Well, there's some massive veins.
[00:08:00] Host 2: Glenn Hopper: Yeah, I don't know. I don't think I have that vascularity. I don't.
[00:08:03] Host 1: Paul Barnhurst: Blood flowing so quickly getting ready to have a heart attack. Blood pressure's a little high.
[00:08:06] Host 2: Glenn Hopper: Oh, you know what? I was thinking of Stranger Things season. Whatever it was for when Billy was possessed by the thing. And you could see all the veins coming out. That's what I look like. The guy, the bad kid from Stranger Things.
[00:08:20] Host 1: Paul Barnhurst: All right? Well, that was a fun thought experiment. One other fun thought experiment I saw this week. And then we'll get into the 12 days of shipments. All right, so I went to Vegas. I was there all week for the Financial Modeling World Cup televised on ESPN. They had the College World Championships. We had a group of students that came from Madagascar to compete. It's really cool. I mean, just people all over the world. Three time champion in Excel Esports, other guys that have won it, one that had won twice, another one that had won it once. And there was what they call this last chance qualification round. So there were a certain number of people that already made it through, and then they selected another 40 something people for the first night of the event in the arena to narrow it down. And over time they narrowed it down. So they got to one man. And so the case we had was like Axel the elf, the Christmas elf, and there was this list of people and you had to help find out who was naughty and nice. Which states had the most toys, which state only had one toy because it had been recalled, and all the other states had recalled it, but they had, and it got harder and harder toward the end. It got into you got to figure out how many trips Santa has to take because he has to refill his bag.
[00:09:36] Host 1: Paul Barnhurst: It's not big enough to take all the toys. So you start from super easy all the way through. So seven levels, five bonus questions. You know, max score of like 150 points or something like that. And David Fortin, who created the actual case for the finals, did a presentation where he showed I took the case after it came out, and I went with copilot and tried to see what I could solve. So he had the case and he started asking copilot based on my data. First one was like, tell me the top states for toys, or find me what toys so-and-so gets based on the list. You know how many kids are naughty, and he solved the first three levels with copilot all the way through and showed it, you know, and to give reference, you know, I almost never practiced. But for example, I only solved the first three levels. I didn't finish level four in the 30 minutes. I finished four bonus questions, which he didn't do any of those. So frankly, it was a little over 100 points off me, so it would have been toward the bottom of competitors, but it was still crazy to see that. Hey, you can just start asking copilot. They can start solving these things.
[00:10:43] Host 2: Glenn Hopper: I mean, this goes right to the segment we did for months on, you know, will I take your job? And it's I mean, that's kind of a live example of well, this got pretty good for, you know, think about you where, where you are and then the people big competitors there. But then think about vast majority of Excel users who wouldn't come anywhere near that. And if you have copilot already coming to that level, I mean, that's pretty fast and pretty interesting.
[00:11:14] Host 1: Paul Barnhurst: Yeah, I would say, you know, at least 50% of the people out there can do what they can do, if not more. And in many cases, I'd say more like one guy even showed answering it Python. And he wrote the Python for something, right? Because the new copilot, that's one of the emphasis is Python. I said when they came out with Python, they came out. That came out with it when they did, because it's a perfect match for copilot. And except it will make things a lot easier and make almost anyone into a basic coder, not an advanced coder by any. We've had that discussion, but so that was fun to see. The other that was fun is the final case you'll get. Were you a video game player ever?
[00:11:52] Host 2: Glenn Hopper: You know, when I was a video game player in about 1980, I was the master of the Atari 2600. Since then, the games got way, way too complex. Once we went from the single joystick controller with one button, one button, I was out.
[00:12:07] Host 1: Paul Barnhurst: It's yeah, how old I was in 1980.
[00:12:10] Host 2: Glenn Hopper: One, Four okay. All right.
[00:12:12] Host 1: Paul Barnhurst: So you know, Our first video game ever was did I remember from RadioShack we had pong. I remember we had one big paddle and one little paddle, and you could play on the screen and do pong. That's the first video game I remember. Then we had Magnavox, Odyssey 64, and we bought the voice thing that went on top of it. It was this big, huge keyboard and you could play games. All right. So one other thing from the competition is to say the final game for all your fun is they made him solve a case of World of Warcraft. The other one that was fascinating is ISO, ISO top. So he rotated the grid in Excel and it had all its own renumberings. So it was a diamond at a 45 degree angle, and you had to find the answers out of the grid. But Excel is only going to understand the regular grid. So you had to convert everything. So they're trying to look at this thing thinking, I got to solve cases with this diamond grid that's at a 45 degree angle where C25 might be equivalent to B6 in the grid, right. So you kind of have to match it all out. And one guy literally went like this to the cop. Like as the competition got started when he first saw like tilted his head to go, hey, will that work to see this normal?
[00:13:26] Host 2: Glenn Hopper: Wow. So what was the mood there? I mean, when with all these, you know, world class modelers seeing how AI is catching up? Was there fear or was there sort of a pride in, you know, sort of like before open or before AI, you know, became the best chess player in the world or the best go player in the world. It's like. And now it's coming for Excel.
[00:13:48] Host 1: Paul Barnhurst: You know, I don't think there was any fear of that. I think these guys at this point just looked at it. Okay. It's coming. You know, we need to learn it. But you know, amazingly enough they were all there. Almost all of them wanted to have fun and compete, but to teach others. I've never seen a community like this where, yes, they're competitive, but to a tee, every single one is friends first. Really interesting group. Was super impressed with that. I didn't get the sense from any of them of real concern. I mean, I asked a few, I caught the guy who won the whole thing and you know, I will build the models for us. He's like, yes, but you'll still need human judgment. Yeah. You know, there's still going to always be some level of human judgment to interpret, at least in the near term, when it gets to the point where that's not needed. Well, good by humans, I guess. Right then it's where we start being really nice to our robot overlords.
[00:14:38] Host 2: Glenn Hopper: You have some other content coming out of that, too. You'll save that?
[00:14:44] Host 1: Paul Barnhurst: I'll have two Podcasts episodes I'm going to do on Financial Modelers Corner. Like I said, I interviewed David Fortini, who created the case, Michael Jarman, who run the whole competition, Andrew Gregorovich, who's the founder of the entire organization, and some others that I interviewed. Plus, I'll be doing some episodes with people I met there. So it'll be quite a bit on my Financial Modelers Corner podcast in the future. One other fun thing. So one of the guys do sumo, are you familiar with the sumo app? No. Have you heard of that? So sumo will create music for you. You come up.
[00:15:18] Host 2: Glenn Hopper: With the lyrics. Yeah, yeah. Sorry. These names get. I know exactly what you're.
[00:15:21] Host 1: Paul Barnhurst: I thought you did, but I get it. So Giles Male is his name. He is an Excel MVP and he created a song for the competition. He wrote all the lyrics and ran it through sumo and had it create. It did a few different versions and one of them was really catchy. It came out really good, to the point where Microsoft liked it enough that they were playing it in the arena half the night, you know, and up on the screen and used it for the event. So it tells you, hey, should we be worried a little bit? Right? Usually that would always be a musician doing all that. And yes, he created the lyrics, but then he just used an online app to do the rest of it. It's a far cry from someone sitting with a mixer and instruments and all that stuff to do the background of how it's traditionally done.
[00:16:05] Host 2: Glenn Hopper: Very cool.
[00:16:06] Host 1: Paul Barnhurst: You brought this up. I've been in a cave the last week in Vegas. I didn't know what was going on in the world. I'm finally coming out today. But you were mentioning something called schmooze. I wouldn't call it something else, but I'll call it shipments. I think it's a play on Christmas. Tell us about shipments.
[00:16:22] Host 2: Glenn Hopper: I'm not going to say that.
[00:16:23] Host 1: Paul Barnhurst: You're afraid. You're going to say something else.
[00:16:25] Host 2: Glenn Hopper: And OpenAI is terrible at naming things. However, what this is, is 12 days of releases and pretty cool announcements from OpenAI. They led with a pretty big one on day one, which was. So we've all been playing around with one preview for, you know, since what, September now? And, you know, honestly, it's kind of hard to find use cases for most of what we do, especially because it wasn't multimodal. You couldn't do attachments with it, you know, no images. And it was just text. They have now released one full and they showed some benchmarks of the full version of O one versus preview and kind of across the board O one was better. It's better at deciding how long to think about something and when to spend more resources versus less. And so basically, you're getting more compute in about half the time when you use O one. Now. Now, interestingly, O one is not as good as for O or as good as Claude at coding, but you have to know how to use these models. And what I think is going to happen ultimately is the chatbot will decide which model to use. You won't have to do this. Toggle back and forth.
[00:17:37] Host 1: Paul Barnhurst: Yeah, I'll be part of the workflow type thing of you'll feed in what you need. It may try all of them, and there may even be a model that reviews it and says, here's the best answer at some point, but it will send it where it thinks makes sense, or to two models or whatever.
[00:17:51] Host 2: Glenn Hopper: My typical go to sort of parlor trick with ChatGPT is to analyze financial statements and get ChatGPT data analysts to give us feedback and, you know, do an assessment of a company, but essentially to give us its its opinion on the company, the financial health of a company and help, you know, say, what are some strategies that we could use to improve performance? But historically, that takes a lot of back and forth and sort of chain of thought prompting. But I'm wondering what I'd really like to see is if I upload my three years of financial statements and then in a single prompt, tell it something like, you know, you're the CEO of this company. We need to come up with our three year plan or 3 to 5 year plan, whatever. What we're going to do with the company based on these financials, how would you improve our, you know, financial health or, you know, whatever, something like that. That's what I want to see, because a lot of the work that you and I do, it's, you know, if you wanted to come up with some kind of strategy based on research or what, you know, you could do it, but not being able to attach your own data or to search the web or use any kind of tools, it was sort of limited what it could do. So it's going to be very cool to see what, oh one fall does. And I think this weekend I'm going to geek out on it some.
[00:19:06] Host 1: Paul Barnhurst: At least you're showing your geek side. I mean, I spent a week in Vegas looking at Excel spreadsheets instead of gambling and drinking. I don't gamble, drink a lot, did both. But just the fact that, you know. Right. We went to Vegas for a spreadsheet shows how big of a geek we are. So you're showing your weekend geekiness?
[00:19:25] Host 2: Glenn Hopper: I don't know, I think our audience relates to us, though. It's fine. They wouldn't. They're confessing that they're geeks by the time they commit to hit play on this podcast. So yeah.
[00:19:34] Host 1: Paul Barnhurst: And if they get this far, we know for sure they're.
[00:19:37] Host 2: Glenn Hopper: I guess the only other thing I'll say about the release yesterday was I haven't pulled the trigger on this and I pay. I have subscriptions for if you name an AI service from Gemini to Midjourney to. All right.
[00:19:49] Host 1: Paul Barnhurst: Let's label how many you have a subscription for. So first we can ask our audience, if you're listening, take a guess. You know what the right number is and let us know if you win. So I'm going to let you guess first. How many do you think AI subscription. So kind of let you know that truly dedicated AI tools. How many subscriptions do you think I pay for it?
[00:20:08] Host 2: Glenn Hopper: I bet it's just two. I bet it's Claude and GPT.
[00:20:15] Host 1: Paul Barnhurst: Nope. Three. Claude, GPT, copilot.
[00:20:18] Host 2: Glenn Hopper: Yep. Okay.
[00:20:20] Host 1: Paul Barnhurst: And I do use the AI and it's tool called Clickup. But it's not, I just use it in the tool. Right. And other tools. But they're not dedicated AI's. How many do you have. And you can count something like sumo or things that are a dedicated application Midjourney. So you don't have to just be the traditional LLM model. I'm going to guess you are ChatGPT for sure. I'm going to try to guess you have Claude, you have copilot, you have Midjourney, and you have to have 1 or 2 others. I'm going six is my guess. I don't know what the other two are, but I'm guessing six.
[00:20:51] Host 2: Glenn Hopper: So let's see. Gamma is a presentation creator. It's gamma.app I think is their URL.
[00:20:59] Host 1: Paul Barnhurst: And you have a paid version.
[00:21:00] Host 2: Glenn Hopper: Yep. Midjourney for images. Did you already say that one?
[00:21:04] Host 1: Paul Barnhurst: I did say midjourney.
[00:21:04] Host 2: Glenn Hopper: Let's see. Gemini. I pay for Gemini as much as I dog Gemini because I want to see copilot. So when Gemini.
[00:21:13] Host 1: Paul Barnhurst: You're doing it for comparison purposes.
[00:21:14] Host 2: Glenn Hopper: Yes. Because what I want to see with Gemini is so right now I can have it summarize emails. It does pretty good at that. So it's integrated. And because I use Google Docs and all that. So sure, sure I want to see how Google is handling the sort.
[00:21:28] Host 1: Paul Barnhurst: Of platform and the different applications, how they're really integrating it into workflow because that's the area we need is finance. Is that workflow integration?
[00:21:38] Host 2: Glenn Hopper: Exactly. So I don't use, that's probably the one I use the least. And honestly, I understand that anthropic is doing cool things, but Claude has really been letting me down there. They're bogged down lately because they've, you know, the computer use and they've done some cool stuff. So it's, you know, they're abbreviating their responses. It's down. It hasn't been doing. The one of my favorite things to do would be to upload a 10-K and financial statements and I haven't used a lot lately. Build me a dashboard. And it was building these cool react dashboards. It's not doing that anymore. It'll break down. It doesn't have. And I've got you know that's on the paid version. So Claude's letting me down a little bit. I don't pay for, grok, you know, Xai and all that. I just, you know that the whole Twitter mess. I'm just, I'll skip that.
[00:22:29] Host 1: Paul Barnhurst: Yeah. So one other on that. That's why we're going to get you. So they announced their pro. What's going to run a month.
[00:22:34] Host 2: Glenn Hopper: So $200 a month. So think about.
[00:22:38] Host 1: Paul Barnhurst: That get into my wallet.
[00:22:39] Host 2: Glenn Hopper: Yeah that's one. Feels a little painful.
[00:22:42] Host 1: Paul Barnhurst: But what are we getting for the $200? What's the selling point that says I should pay 200 instead of 20?
[00:22:48] Host 2: Glenn Hopper: It is a ten fold increase. But the idea is. And we'll see, because I think what's going to happen is with this one fold, you're going to be even on the pay. So I'm on the OpenAI team account, which is, you know, so, everyone who works with me can use this account and we're, you know, have the protected non training. So that's 30 bucks a user per month.
[00:23:08] Host 1: Paul Barnhurst: Well, let's think that the average let's go consultant for a minute. Right. The average consulting rate is probably low end $100 for some basic stuff up to you know big huge firms. Let's just say 500. Yes. There's people that charge more, but I think that's a fair range. You say 1 to 6. So let's just say you're in the middle. You're charging 300. Righteous Ballpark. You need an hour and a half of efficiency.
[00:23:34] Host 2: Glenn Hopper: Yeah. I mean, and that's I know I get that. I would imagine I get more than that.
[00:23:38] Host 1: Paul Barnhurst: And so from an economic standpoint, ignore the emotional of that's 200 bucks economically you're better off paying it. Yeah. It sounds like and I think anyone.
[00:23:47] Host 2: Glenn Hopper: Anyone who's a power user, I'm not going to you know, I could argue and again, it would be a write off. I just don't. I'm not going to pay for it until I start hitting limits. Now I'll be interested.
[00:23:56] Host 1: Paul Barnhurst: Of course, until you actually feel like you have to have it. Why pay for it? Yeah, again, you're behaving rationally. Even finance people. There's plenty of irrational. There's a reason. There's behavioral economics.
[00:24:07] Host 2: Glenn Hopper: Thinking fast, thinking slow. We already talked about that with zero one and you longer we did.
[00:24:13] Host 1: Paul Barnhurst: It'll be fascinating to watch. Right. What people are willing to pay. How they think about it. Where the limit is as we go on and on with AI. But we've got almost 30 minutes. Maybe we'll call it here a moment, unless you have any other fun AI thing.
[00:24:27] Host 2: Glenn Hopper: I think so I just want to see how much better at ChatGPT your daughter is going to get when you upgrade to that pro plan.
[00:24:35] Host 1: Paul Barnhurst: I'm going to have to with how much she uses it, which reminded me of one other thing. It's a lot like if anyone's used Python in Excel, they give you a certain amount of pro computational because it's all off server. Everything for Python is calculated in the cloud and sent back to your machine. The code. So you know it's all on the server and you only get so much. And then it goes to ordinary speed and it runs slower. Or you can pay to keep it higher and it'll be interesting. Yeah. That's what we're seeing with a lot of these tokens. Right. That's kind of the AI approach is, hey, there's so much bandwidth. You get free so many tokens time, whatever you want to call it prompts which are proxies. And so it imagined that we did some changes in the economics and some limit. You're going to want the pro plan for her. I just let her go to a lesser one. I'm like a flop. The worst is when I want to use that version. I need it for something that she's used. All of mine. Like you're blocked out till 5 p.m. today.
[00:25:38] Host 2: Glenn Hopper: You're a good dad. You're a good dad.
[00:25:40] Host 1: Paul Barnhurst: He has fun. He's using my phone right now. Speaking of phones. So I will probably be a girl forever, according to ChatGPT. Now what? I needed the girl with the beard. We'll see if we can get that for the next one. Welcome to another episode of Future Finance. I have here with me my co-host Glenn Hopper, and we have Gabi Steele here with us today. That's going to be our guest. Gabi is a distinguished data visualization expert and entrepreneur. She has co-founded Data Culture and Preql. Data culture is a company dedicated to empowering organizations through data driven decision making. Preql is an enterprise AI powered metrics platform that recently raised 7 million as a seed round. Gabi's expertise lies in transforming complex data into accessible visual narratives, enabling businesses to harness the full potential of their information. Her innovative approach has earned her recognition in the data visualization community, and she continues to lead initiatives that bridge the gap between data science and practical applications. So, Gabi, welcome to the show.
[00:26:46] Guest: Gabi Steele: Thank you so much, Karl. I'm so excited to be here.
[00:26:49] Host 1: Paul Barnhurst: Yeah, we're really excited to have you here and get to talk a little bit about what you're doing. And, you know, the finance world. And so I want to start with this question. I noticed you earned your undergrad in graphic design, but you pretty quickly gravitated more toward the data analytics, the engineering, the visualization kind of what attracted you to that area versus graphic design?
[00:27:12] Guest: Gabi Steele: Yeah. So I share this story often, though I did do my undergrad a while ago. I look younger than I am, but basically I entered the space. I studied painting and drawing initially, and then came to this kind of creative world and was doing a lot of designs for websites and for different fashion retailers, and realized that there was just a really clear wall or ceiling or limit when it came to what you could do without some sort of engineering component. You were designing things you couldn't build. So that drew me closer to doing some front end work. And then data is really appealing for anybody who has curiosity around how businesses function, how things really work. And it just started to seep into everything I touched. So I was at Parsons School of Design, where I did a master's degree in data visualization, in a moment where a lot of folks were starting to think about how do you make data more accessible to broad audiences? What is this? Is it its own discipline? So some of that was timing, and a lot of it was curiosity. And my current co-founder, Leah Weiss, and I, who's also the co-founder of Data Culture, and we've been working together for seven plus years. We both came to the engineering space post, kind of like initial graduate or undergraduate degrees, and I think that helped us be really empathetic to audiences that didn't quite know how to write code, how to write SQL, but wanted answers. So that's been an ethos from the beginning. And then I was fortunate that I got to work at the Washington Post during the 2016 election, where I did a lot of data visualization and storytelling work, and here we are.
[00:28:52] Host 1: Paul Barnhurst: Yeah, I think what I saw you telling a little bit of that story was something when you were at The Washington Post. I saw something, I think, on YouTube or somewhere as kind of trying to do my homework, and you did a little bit of that transition. So every so often I do, occasionally.
[00:29:08] Guest: Gabi Steele: I'm allowed to ask questions. But what did you both study in undergrad? Was that an option? Like just thinking about how you got here, like what brought us all to this room.
[00:29:16] Host 1: Paul Barnhurst: So I did business management with an entrepreneurship specialization and a minor in political science.
[00:29:23] Guest: Gabi Steele: Okay. And you're an FP&A guy now?
[00:29:26] Host 2: Glenn Hopper: Yeah. So I was on the seven year undergrad plan, and from that was it from Tommy Boy where it said, yeah, a lot of people go to school for seven years. They're called doctor. I didn't, I switched majors four times. I went from mechanical engineering, hated it, and switched to philosophy. And then someone told me I couldn't get a job with philosophy. So I went English Lit, which still couldn't get a job. So then I went journalism. Then I became a journalist. And then I got married and realized, I wasn't going to make enough money as a journalist. So I went and got my MBA, and then I later got a master's in finance and a grad certificate in analytics. Terrible at visualizations, though. Colorblind. I'm no Edward Tufte, but I bet you do some badass visualizations with that graphic design for undergrad. And then with the master's focus, that's, Yeah. I mean, and that's such an important thing to be able to express your ideas visually because I can table and chart and use the basic stuff. But there's a real art that I think a lot of people don't realize for data visualization.
[00:30:31] Guest: Gabi Steele: Yeah, I mean, a few things. One of them is we talk a lot about accessibility and making sure that your visualizations and even just web design and for our company stuff it. You can be colorblind and you can still interpret information. And I also talk a lot and I've around this idea that data visualization is a combination of disciplines. And it's become really present again in my life, returning to finance, storytelling and financial. It sounds like we've all kind of returned to this space one way or another. And one of the newer learnings I ended up in data for a while. Data engineering teams trying to support larger organizations building data engineering consulting projects and things like that. And I've realized, especially recently in the types of users we have with the Preql platform, finance teams are data teams and data is you know and all of this. So there's the design component for sure. That's like could be a specialized discipline and something that I spend a lot of time in. But when you think about storytelling with data, that's everybody's job. And it's especially important when you're telling the story of a business. So what drew me to this space, I think, is that curiosity that ties us together here as well.
[00:31:41] Host 2: Glenn Hopper: The conversation that we have all the time. And I think they're getting there. But it's taken a few years. It was amazing to me that sales and marketing really leaned into data and business analytics ahead of when finance did, but I think it's because, you know, e-commerce and all the all the data that was out there on the web, they had more data than we did early on where we just had GL in finance. So that's not big data when you've got, you know, when you have you're doing a three year analysis and you have three marches. Well, that's three data points. That's not big data. You can't use the same rules to it. It wasn't until we started finding correlations in other areas and other data points from the business that you could roll into finance to your forecasts and your metrics and what you're tracking that then you started to be able to use some of these big data tools. So it's great with what you're doing with Preql now and what you've done before. And in coming into finance and finance coming up to it. And so Paul and I are both kind of on a mission. And some people, they're still kind of an old school mentality around there of, of, you know, among even the people that, you know, we're not business analysts, we're FP&A. And it's to me, they're the skill sets have to combine. Having a data science component to FP&A makes you, in my mind, that much better.
[00:33:00] Guest: Gabi Steele: Yeah, everyone's actually a data person at the end of the day because everybody's got to care about the numbers. And maybe a little bit of a segue into this AI piece. I guess I'm the first person to use those terms today. Obviously, when I was in undergrad and graduate school. I wasn't really something you were going to get into and study. Even if finance was and the applications of these automations that are available today, I think people say, you know, even if you studied machine learning in AI in 2020 or 20 21 or 2020, what's changed so dramatically with the technology would not necessarily be applicable. So being a general data person and someone that cares about numbers and information, I started also in the consulting side of our work with a lot of e-comm and omnichannel and retail businesses. But the more data that companies generate, the more this becomes an issue. And there's these newer solutions. So yeah, excited to talk more about that.
[00:33:55] Host 2: Glenn Hopper: Total softball for my next question. With Preql using AI to automate data cleaning and governance. Sounds really cool. And I know that this is where everybody's trying to get, but I'd love to hear more about how this technology helps solve some of the big challenges that finance teams are facing, making sure their data is accurate and consistent. And that's, you know, I mean, I think everybody talks about generative AI right now. Obviously, that's what's catching everyone's attention. Which generative AI, you know, is nondeterministic. You're not always going to get the same answer. Whereas obviously with traditional machine learning it is deterministic and you can get the same answer. But I think a lot of people conflate the two. So I'd like to hear how, you are using AI right now and how that's helping finance teams deal with things where their data is accurate and consistent around, you know, their financial reporting and forecasting.
[00:34:46] Guest: Gabi Steele: Yeah. So I can share more about the current applications of the product to answer immediately. FP&A teams spend a lot of time manually consolidating information. That's what I've learned. And our story of building Preql was always to help business teams better understand in the moment. They need to make a decision. What are they? What information do they already have available, and how do you get that information to them in whatever format they need it to be. It could still be consumed in Excel, but let's make sure that the numbers they're seeing in Excel are real time, on demand, things like that. So I and some of the data cleaning work, initially it was automated. And then obviously technology makes everything that much more possible. And we can distinguish between the generative side versus what's more kind of automation and closer to what's available today. What we saw with FP&A teams was a few problems. You're mentioning one of them accuracy, and that's tied to the manual processes. So if you spend 70 hours a month or two weeks of every month just figuring out, okay, let's pull in, let's download all of the Excel CSVs from the data sources that we care about. Let's crunch the numbers together and try to understand where those discrepancies, where we join different source systems. And we need to create a new metric on top of that.
[00:36:06] Guest: Gabi Steele: And then we deliver those numbers. Yes, you're wasting 70 hours a month. What CFOs tend to care about isn't necessarily the resources wasted, but actually the accuracy that it lends itself to. The next question. You've just manually done all this work, Joe, or whoever it is on the team, that's amazing at crunching those numbers. I learned this term recently. There could be this fat finger problem that just the level of reliability and accuracy at that point, due to lack of automation and all of the manual work then becomes a challenge. And then consequently, you're spending 75% of your time doing that, that work. And the remainder of that 25% is answering questions because FP&A teams, it's not just their responsibility to hand over a report. They also need to be able to justify every single line within that report. So you're essentially hiring entire functions to do something that could have otherwise been completely automated. And then that team and that brainpower doesn't necessarily get replaced. They can then be strategic and think about the future and forecasting and, you know, making real predictions. They become sort of a generative AI function, so to speak. so that's where Preql's seen a lot of value. And to be clear on how the product was structured. We built this to solve a data engineering problem. There's been a lot of innovation that's come into the space since Amazon Redshift 2012, making data cloud warehouses fast and cheap, right? So super easy to copy data now into a cloud storage warehouse.
[00:37:39] Guest: Gabi Steele: There's a ton of providers that do that. Once the data is there, what do you do next? Do you give it to the finance team and this raw messy format? Do you, you know, start downloading it out. Do you put it in a Tableau chart? People did that for a while and they still do. You take a Tableau report and you pointed out all the messy raw production data. It's not giving you metrics or answers. So the only way to solve this challenge is really with data engineering manual kind of laborers. And so we set out to solve that problem and do so in a no code way, in a way that didn't require knowledge of SQL, knowledge of structuring data because they're such different skill sets. And so somehow that brought us full circle, right to FP&A teams, because they essentially are the ones that need to deliver the reports. We try to help people with these three just to end this thought by answering your question. We want to help teams reduce manual processes, increase accuracy, make sure that they have accurate data, and that that process scales with them, doesn't necessarily require adding on because these problems just grow over time.
[00:38:42] Host 2: Glenn Hopper: I was going to say I love where you guys are right now, because for years I would tell people, if you really want to get ahead, you need to learn Python and SQL. But what we're seeing with generative AI and with drag and drop and low code, no code tools is if you can do your job and you can say, I'm a domain expert in finance, I don't want to have to learn to be a machine learning engineer or or a coder. Then you can just really focus there. And it's so we talked about democratization of data for years, but with tools like this, it's democratization of data science, where you make it more available to.
[00:39:16] Guest: Gabi Steele: Yeah, democratization of analytics and making it so. And my belief has always been like Preql sets out to solve this challenge, which is domain experts know best how they want their data structured and what they want to do with that. Data and data engineers know best how they want to structure data and the coding side of things. So what do you do, you close that gap? Because it's just this age old problem where they go back and forth, and that's a whole new set of manual back and forth with finance teams. They're not really getting that many data resources anyway a lot of the time. So again, that's been like we've been trying to cut to the pain as a startup. Like we want to go as close to the painful part of the process as possible, but one step removed. Larger organizations are more kind of tech forward companies. You find you have a data team and a business function.
[00:40:02] Guest: Gabi Steele: The business function should be thinking about the business. So what I'm trying to do with AI is not replace business functions in any way. It's let business domain experts and finance experts even focus on how to drive revenue for their business with existing data, and eliminating all the parts of their work that an intern could do or, you know, is simply just a manual process that takes up too many resources.
[00:40:26] Host 1: Paul Barnhurst: But I liked all my manual processes. It's job security. I worked at know it. They wouldn't invest anything, so we had it pretty much do everything manual. And I lived in Power Query. And yeah, we made a lot of progress, but it was not the most efficient way to do things. That's why I always call it I go, it works, but it's janky. But I would always say.
[00:40:49] Guest: Gabi Steele: Yeah, because I know new manual processes will also develop. It's not like everything is going to be replaced. I tell people a lot when they ask questions. Finance seems especially like they're just setting out on their AI journey. Our idea was, you know, Preql is the first step in this journey. Think about structuring your data because you can't do anything more with it. And I think where AI is today, it does a lot of small slivers really well, but it doesn't automate start to finish a process that a human could do for good reason. And I think there will always be exciting jobs. I spoke on a panel earlier in, I think it was last week where someone who's building an accounting software was sharing that, you know, accountants aren't the happiest people. Maybe you two would know even better. I've never done so many accounting. Like, how do you make their jobs even more interesting and actually apply that brain power? Even though some of the manual work as you're sharing Paul, can be therapeutic?
[00:41:44] Host 1: Paul Barnhurst: Ever feel like your go to market teams and finance speak different languages? This misalignment is a breeding ground for failure in pairing the predictive power of forecasts and delaying decisions that drive efficient growth. It's not for lack of trying, but getting all the data in one place doesn't mean you've gotten everyone on the same page. Meet qflow.ai, the strategic finance platform purpose built to solve the toughest part of planning and analysis B2B revenue. Qflow quickly integrates key data from your go to market stack and accounting platform, then handles all the data prep and normalization. Under the hood, it automatically assembles your go to market stats, make segmented scenario planning a breeze, and closes the planning loop, create air tight alignment, improve decision latency, and ensure accountability across the teams. Yeah, no, there are times when you just want to sit and waste. I shouldn't say waste. Spend mindless time working on a task, but not very often. There are occasionally times it's like, all right, I'll take that. Just because you get to avoid, you know, the having to think a ton and just clean up some data. But no, constantly cleaning up data is not very fun. I've done more than my fair share of it. Like I said, Power Query became my best friend for a couple of years. I would rather spend more time with the business, but I had to do both.
[00:43:26] Guest: Gabi Steele: But you'll have new best friends that are AI software's. You know, I think that that's kind of the part. It's like, how do you use whatever tool you have to be the best? And can you then how does the whole business change if everybody is like super powered with. Yeah.
[00:43:41] Host 1: Paul Barnhurst: I mean, the thing is, like you said, full processes. I mean right now, you know, the agents aren't there yet to do full processes. And will they ever be, you know, which parts will they do? Like I said, there's always going to be human interaction. If you want to call that manual, fine. Whatever you want to call it, there's always going to be a human element to the finance, to FP&A, to what we do, regardless of how much we're able to automate and change things with AI.
[00:44:06] Host 2: Glenn Hopper: And it's, you know, right now, too, it's as stuff is automated, you know, it's like more automation leads to more data, which leads to need to automate more and more analytics. I mean, they're always joined at the hip, but then that skill set, if you're if you're lowering the barriers there, between having to learn a new skill and just being able to work with new data. I mean, that is a very strong value proposition. So, I'm thinking about organizations trying to build a data driven culture. And I work mostly with SMBs. And we just know where they want to get. A lot of times they don't have enough data, but they don't have the in-house resources. Typically under the old paradigm, they don't have the in-house resources to build things from scratch. So it's very hard to move, you know, kind of move up that data maturity scale. So when you're working with businesses, trying to keep that balance between keeping things accessible for stakeholders who may not be super technical, like how do you strike that balance between a CFO who wants their team to be more data savvy, but they have the limited resources? I mean, how do you work in that space?
[00:45:16] Guest: Gabi Steele: That's a really, really good question. And I love working with old school businesses, I call them, or legacy businesses, real businesses, because those are usually the folks where the opportunity is the greatest. And the way that you can prioritize, like if you've got a limited number of internal resources and your list of desires or new tools that you want to bring on or new solutions is super long, the best way is to find a couple of champions within that group that have an appetite for trialing some form of automation, and then they can kind of disseminate it around the organization, because I don't really believe, like I think what we're going to see with AI is these SMBs that would otherwise not have the chance to compete with counterparts that had just larger functions. Because finance has been so manual, it's been so headcount forward, even though you all are the ones that actually, you know, are trying to reduce the headcount and manage the headcount, which just makes it a little bit easier when our product, a lot of the time, is making it so that you don't need to hire a new function. You sit with that group and you prioritize. Where are we leaving money on the table? Where do we already have data that could have been applied to change course? Forget like forecasting ten years in advance. Forget, you know, anything so generative and complex that we didn't use last year that we can apply today. And Preql is very versatile. Like the idea is centralization is number one.
[00:46:39] Guest: Gabi Steele: And what we're trying to do is create a single source of truth. Being data driven is something companies have aspired to do since before I got into the space more than ten, 15 years, and we've never really reached that moment. And we talked a bit about data visualization. But what BI tools have done really well is branded themselves and even some of the data softwares as by this solution. And you will be data driven and then you find an SMB spending sometimes $1 million a year on a BI solution. And they're like, but the numbers aren't quite right, and nobody's trusting these numbers. And why aren't we? We have charts, but we have a chart. And what I've learned designing a lot of charts is it doesn't matter how beautiful or how complex a visualization is, if the numbers that are feeding into it aren't trusted. So that's the approach with Preql, and why I set out to build a product that is focused on making the numbers match versus making the fanciest kind of visuals. And I think some of the smaller organizations where there's limited resources, it's even easier to prioritize. Like one mistake that cost us X million. I did a lot of that work back at WeWork as well. How are we leaving money on the table, which unfortunately couldn't save the whole company. But there was a lot of money being thrown around that you couldn't because every dollar counts at that point especially.
[00:48:00] Host 1: Paul Barnhurst: Not if you have community adjusted EBITDA. Wasn't that what they called it?
[00:48:07] Guest: Gabi Steele: So first. That was exactly what they called. Oh, wow.
[00:48:11] Host 1: Paul Barnhurst: Oh I know that's what they called it. That's what I said. It's like, well, all this that we do 80% of our business, that's not real money. That's focused on the parts that make us look really good.
[00:48:20] Guest: Gabi Steele: And you know, it's impressive. It's impressive. And it was still a great product I think, like companies can win or lose based on these like financial reporting. A lot of like how well they're able to structure existing data or manufacture existing. We can yeah move away from exactly what went on there. But the idea being like, data can be really powerful. Very few organizations use it. Well, there's like the top Faang companies which like, you know, Preql wasn't built only to serve Google or it's like not the direction they're able to kind of build internal processes, though. They've been in our pipeline before, like things like that. But it's more about how do we serve companies where they've never thought about automation, but now they're forced to. And AI has put it in their face. And what can we do? What's the lowest hanging fruit? And then the quick wins will lead to the larger, more meaningful solution. Kind of development.
[00:49:17] Host 1: Paul Barnhurst: Makes a lot of sense. I can tell you. Think you're about to say something, Glenn?
[00:49:22] Host 2: Glenn Hopper: Yeah. What do you think? No, if we go back and we look at the word count, I'm probably going to have more words than anybody on the show, which I don't think that's what you're supposed to do as a co-host.
[00:49:33] Guest: Gabi Steele: I can give you a run for, like, we can compete. But if there's a specific story also that you're thinking about, I'm happy to address it as well, because I think they tend to have the same challenges a lot of the time. Data seems so unique and as like having done consulting for many years and seeing the metrics that people care about, especially within similar Verticals. Your problems are not as isolated as you would think. Everybody is struggling to be data driven. Everybody wants to measure generally the same thing. So I think that we will see, a lot of exciting changes happening and more off the shelf things that a more junior employee could leverage, or a smaller team that the CFO will appreciate.
[00:50:15] Host 1: Paul Barnhurst: I definitely do think we are seeing more of what you mentioned, you know, with Preql and with others. I talk to companies all the time of trying to standardize because yes, there's a ton of difference. I think the area where it's most difficult is every company plans revenue and has some unique nuances. But in general, the data, the metrics we're looking at, there's a ton of similarities. There's a lot that can be done to standardize it. Just because 5% is unique doesn't mean you can't standardize the other 95%.
[00:50:44] Guest: Gabi Steele: Yeah, I don't know if you've seen, I'm really curious. I don't mean to turn the questions back on you both, but like as folks that are now entering, you know, you came before I blew everything up. Like, I think what we've seen and the reason I gravitate so much to this space, because I do also gravitate to these companies where there's just no infrastructure whatsoever when it comes to data analytics. How are you seeing this FP&A shift? Where are people adopting automation first, and where do you feel like your clients or people you talk to are the furthest behind?
[00:51:17] Host 2: Glenn Hopper: Paul, I can probably speak to this one because I do. Yeah. So SMB space. So let's say businesses are the most interesting. Well, actually I've got a client right now. They're 25 million in revenue. They're global. They have 20. They're in 26 different countries. They have their chart of accounts is a nightmare. So, all their revenue, they have a different revenue account for each client. So they have retainer clients that go over time. So they had when I saw them when we first started working with them. Something like 900 accounts. It just some absurd amount of data that they couldn't. And they had a great sales force implementation, their sales force, they did a really good job, but it wasn't connected to anything. And they were using QuickBooks desktop at their corporate location. And all the subsidiaries were sending in their just spreadsheets. Basically, here's a spreadsheet of our GL for this month. And then somebody was having to key in all that. So this has been a long process. Like eventually maybe they get to an ERP, but before they're even to an ERP, it is about cleaning up the chart of accounts and identifying those KPIs, because they had all these different metrics and they're trying to bonus our employees, you know, on their sort of the profit that they're responsible for.
[00:52:39] Host 2: Glenn Hopper: But when the chart of accounts is that messed up, you couldn't even get to it. So they didn't know, they couldn't, you know, couldn't properly allocate to, what went to corporate and what went to the markets. And maybe somebody was working on the market, but the client was in a different market. So it's this is where businesses get if you founder led businesses, you get up, you know, you're doing significant business, but you never had the resources to to really think about data. You're just kind of going and going and then we're all there. All there is happening at first by their tax people and tax people. It's funny, small businesses always want their tax people to do for for them. And that is not that doesn't go anywhere in a tax person's brain. All they're trying to do is minimize your tax burden. And so it's a big ask. So that's pretty big I mean, this client was that they let themselves get so big without cleaning it up. It's pretty amazing. But they had, you know, kind of hockey stick growth to get there.
[00:53:36] Host 2: Glenn Hopper: So they're just, you know, they're serving clients and not thinking about their back office. But most of them really it starts with they've got to get their GL in order. And as soon as you get your GL in order and you can make sense of that data, well then you can start bringing in external data and other departments. But being able like in that company, tying Salesforce to their GL is going to be very significant because then you've got visibility going back from prospect, you know, all through the sales funnel all the way through. So it's and everybody wants it to be this cool flip a switch. And now we've got analytics and it's there's so much. And you have to go through it. And you have to look at data flows and you have to look at processes. So building it out, that's why it's hard to sell to these groups sometimes because they know what they want. And then when you walk them through and tell them what they've got to do to get there, they think, well, I don't want to commit 18 months to a data cleanup or whatever. Yeah.
[00:54:30] Host 1: Paul Barnhurst: Like you said, with any of your SMBs, it starts with dealing with the technical debt. They all want to use AI, but none of them. They have that not technical. You could call it accounting debt if you want. Sometimes it's system, sometimes it's chart of accounts, sometimes it's the fact they've been doing a cash basis and you just can't build a proper budget off that as they're scaling and so you have to start by fixing that and giving attention. They never did because many of them didn't have any financial support and they were just concerned with the business. So the SMB space, I'd say the biggest challenge is the data. And from every fractional CFO I talked to often your first six months is cleaning things up. Then you actually get to the point where you can really budget forecast, start to bring analytics. You know, if you get a little more mid market, they've usually done a lot of that. But they can still, depending on how many acquisitions they've done, can still be quite a bit of a mess. So they're trying to figure out how do I get to that next level where I can really start using the data? Typically with the SMB, you're just starting with, how do we get it to a clean place where I can budget, then start doing analytics? You know, mid market, it's okay. What more do we need to be to advance our analytics to get better and go deeper?
[00:55:38] Guest: Gabi Steele: That is such a, yeah, I really appreciate that Paul as a breakdown. And also Glenn the story, the idea that so many organizations have different definitions of revenue. It's kind of what brought us into the space to begin with. Like, I don't know if I can do it in five minutes, but if we could reduce the 18 months to 18 days, that's something that Preql's definitely been able to solve for. And we should talk more about it afterwards, because it is absolutely wild to have that many definitions of revenue, but be so revenue positive. You know, obviously they were focused on the right things and the way we're describing it here, it's sort of like the video game, like you got to get through level one, and then you can bring in more data and unlock more and more and more. And then I guess that lends itself more to the sort of mid-sized businesses which we found are very ripe for our accessibility features. So like the last piece of what we've done and everybody's building in Copilots and things like that to their products now, like AI agents, Preql was initially created as a no code semantic layer, so a way that you can have semantic English language on top of multiple data sources. So our AI and conversational features are not querying your ERP and your messy GL files or whatever else it is. It's querying the semantic layer. And that's just made the conversational AI so much better. And that's where mid-size businesses also get excited because it's like, okay, well, we have all this data now. Everybody can be it's the democratization. Yeah. Can you really reach a place of self-service? I'm not totally sure. That's a different podcast, a different day and the AI and whatever, but.
[00:57:16] Host 1: Paul Barnhurst: I think we could spend a couple hours on just that subject and not come to a clear conclusion. But yeah. What do you define as true self-service as well? It's very subjective.
[00:57:26] Guest: Gabi Steele: Exactly.
[00:57:27] Host 1: Paul Barnhurst: But I thought it was all everything's truly no code in self-service. Right. And everybody ever an implementation.
[00:57:35] Guest: Gabi Steele: Pull the numbers match.
[00:57:36] Host 1: Paul Barnhurst: All your graphs look beautiful and you get wonderful insights. Isn't that what all the vendors sell you on? I always love that you have to build the back end to that. It doesn't just flip a switch.
[00:57:48] Guest: Gabi Steele: Yeah. I mean, I think that what we're looking for as well, like as a newer kid on the block in the space is like, who should like I like working with the older school legacy brands and trying to get them further ahead because it's not like one. There are so many solutions that claim they do the same thing. So either you've been burned by one and you're like migrating off and you're saying they promised me they'd do all these things and they didn't, or there's no adoption yet. So I'm really curious to see how these brands, how these organizations start thinking, knowing that they're a little bit more skeptical, at least of you can be data driven in an instant because you've worked in enough places where you've had X, Y, and Z tool and it didn't change the bottom line. And then the last thing I'll say is my background and the work that I've done with the consulting space. But also like on education, I think it is a cultural thing as well. Like you do need people to be excited and interested. And it's an era right now where people are learning and listening to podcasts to understand because no one knows what to expect with AI. So trying to seize that opportunity as a way to bring in advocates, those like the stewards, the ambassadors that we're talking about in these SMB organizations, one person that's like, yes, I will learn and then teach others is really critical. And we look for that in our champions when we're not just trying to sell product through, we want to transform organizations and make sure that everybody is on board. So I think that yeah, and it doesn't matter how long you've been in the space to be excited about the future of AI, it's not just like, yeah, newer generation.
[00:59:25] Host 1: Paul Barnhurst: Oh, and if you want to keep customers, you want them to be using the tool. You have to have those champions. Otherwise they're just going to churn. It's a question of when, not if. If they don't have people that really believe in the product and that you're delivering value. So there's both sides that you love to see them transform their business. You also know if you want them to really use the tool and get advantage of it. You kind of have to do that, right. So it's kind of that there's both those benefits. I think what we're going to move on to. So we have a fun where we ask a couple get to know you questions.
[00:59:54] Guest: Gabi Steele: Cool.
[00:59:55] Host 1: Paul Barnhurst: How it works in this case is I actually fed your LinkedIn an article. You did the questions I had all in the chat GPT played with a little bit. I didn't like the first one because it was basically like you're supposed to get to know you questions, and it came up with all the questions we'd ask for the podcast. So I changed it a little bit and it came up with 25 questions. I'll go first and then Glenn will go. So you get two options. You can pick a number between 1 and 25, or the random number generator can pick a number for you. And that's the question. Blast.
[01:00:23] Guest: Gabi Steele: Cool. I have a number.
[01:00:25] Host 1: Paul Barnhurst: All right. What number?
[01:00:25] Guest: Gabi Steele: I want 17.
[01:00:28] Host 1: Paul Barnhurst: All right 17. And this was all by ChatGPT this time. So we've used different tools. All right. Who inspires you most in the tech or data world and why.
[01:00:40] Guest: Gabi Steele: That is such a good question. I'm going to give a personal answer, but I work every day with Leah Weiss, my co-founder, and I think that you can look out into the tech world and there's a lot of people doing exciting things. I'm going to say two things. She's personal to me in the sense that we partner so frequently in thinking about the challenges everybody faces, and then looking outwards and seeing what's new and what's already what's being done by everybody else. I think the people that are most inspiring when it comes to what we're seeing in the data space, I'll give another answer. That's also somebody I've been following on LinkedIn. You need to stay grounded with what the real challenges are and not just follow the trends. And Leah is amazing at that. And always going back to like, what is the core problem and making sure that, you know, we are never driving too far ahead when it comes to what's going on with AI. There's also a guy who I've been following recently, who's been talking a lot about kind of how FP&A teams, And this is new to us because FP&A teams we didn't always know. His name is Walter born?
[01:01:46] Host 1: Paul Barnhurst: Walter. Yeah. Walter. I know him well.
[01:01:48] Guest: Gabi Steele: I'd love to meet him. His post right now?
[01:01:51] Host 1: Paul Barnhurst: Yeah. If you want to meet, I could facilitate a tour.
[01:01:54] Guest: Gabi Steele: Now. I've dropped his name. Would love to meet Walter. I think he's posting really insightful things and empowering things. FP&A teams deserve to be empowered. And I think that that is the most exciting thing that I've discovered in this space.
[01:02:07] Host 1: Paul Barnhurst: Got it. Yeah, actually, I had a message from him this morning. We're trying to get him on the podcast, funny enough.
[01:02:11] Guest: Gabi Steele: Great. Get him. I would love to hear his perspective.
[01:02:13] Host 1: Paul Barnhurst: Definitely. I'll connect you with him. All right, Glenn, your turn to ask one.
[01:02:17] Host 2: Glenn Hopper: So Paul did a nice job with this. So normally I'll go and just pick what we bounce back and forth between the different LMS and pick these. And usually just to make fun of Gemini. I keep hearing Gemini is going to get better, but I haven't seen it yet. But, Paul's got a good list here and normally, so we'd take a different approach. He does the number of the random generator and I normally because I have the list, I just say give me a random one here. But I think so. I've got that. And I actually put it in to pick a random one. So I'm going to give you the option of we can go with what ChatGPT picks or we can go with what I pick.
[01:02:52] Guest: Gabi Steele: Cool. This is a fun podcast guys.
[01:02:54] Host 1: Paul Barnhurst: Which one you want to pick or you want his?
[01:02:57] Guest: Gabi Steele: You pick, you pick. I will do ChatGPT for today.
[01:03:01] Host 2: Glenn Hopper: Okay, so I'll tell you, I actually liked what ChatGPT gives me.
[01:03:04] Guest: Gabi Steele: But if you're in charge.
[01:03:07] Host 2: Glenn Hopper: So let's go. Let's go with the quick answer for that and I'll ask mine. So ChatGPT picked. If your life had a theme song, what would it be?
[01:03:19] Guest: Gabi Steele: Oh my God.
[01:03:21] Guest: Gabi Steele: That is really funny. There's a fight song. This is my fight song that I've been listening to a lot as a founder. And just like it came out, it was. It was a Hillary Clinton song. But I don't want to get political. But just like being able to cons like pivoting, moving forward, making sure that you can just bat away the stuff that isn't working, and always like focusing on the vision and the goal because things are changing around us all the time and you just have to go with what's happening. These are fun.
[01:03:51] Host 1: Paul Barnhurst: Got it. We got it.
[01:03:52] Host 2: Glenn Hopper: I was going to ask mine because I was curious on this one in yours.
[01:03:56] Guest: Gabi Steele: What are your guys's theme songs?
[01:03:58] Host 2: Glenn Hopper: Mine's I Can't Get Up by James Brown.
[01:04:01] Host 1: Paul Barnhurst: What's mine?
[01:04:01] Host 2: Glenn Hopper: That actually kicks in on long runs. That's my mantra that goes in my theme song.
[01:04:06] Host 1: Paul Barnhurst: I'm gonna have to think about that one.
[01:04:08] Host 2: Glenn Hopper: Cool. My question was going to be, you know, you started in graphic design. You mentioned painting. Do you still go back and as a hobby, do you still pursue art at all as an escape or anything? Do you stay involved with that?
[01:04:23] Guest: Gabi Steele: Yeah, I won't, like, swivel my camera around and show you paintings in my home today because I am home. But I don't paint as much as I'd like to. I have all the materials for it, and I, but I do. I have it in the house. I think it's kind of connected to this journey as a founder, where you need to step away and kind of return to your hobbies or whatever else it is. So yeah, I still paint, and I think that starting my career by focusing on harnessing this like creativity skill and this idea that you start with a blank canvas and then you build something really did kind of lay the groundwork for the later entrepreneurship and serial entrepreneurship. Like, this isn't the last company I'm going to found, and I've now found my way into this amazing community of people. But what's the next community? I think, Paul, you mentioned you did study entrepreneurship and you've done so many things since then. So yes, I still paint. I wish I painted more and I think everybody should be stepping aside and like thinking about doing things just beyond and especially with the screens and everything today.
[01:05:24] Host 1: Paul Barnhurst: So I got to start painting.
[01:05:26] Guest: Gabi Steele: It sounds like, yeah, we should do a painting class. We could do that's a fun offsite.
[01:05:30] Host 1: Paul Barnhurst: That could be an interesting podcast. We could, you know, video one we could do we could each be doing our zoom instructor. Thank you for joining us. It's been a real pleasure having you, Gabi. We've had a lot of fun. If we had more time, we could definitely go deeper, I'm sure, and talk probably the rest of the day. But, you know, we're getting close to the holiday and I'm sure we all want to wrap up work so we can spend time with family. So thank you so much for joining us. Really enjoyed chatting with you today.
[01:05:54] Guest: Gabi Steele: Likewise.
[01:05:55] Host 2: Glenn Hopper: Thanks, Gabi.
[01:05:56] Host 1: Paul Barnhurst: Thanks for listening to the Future Finance Show and thanks to our sponsor, qflow.ai. If you enjoyed this episode, please leave a rating and review on your podcast platform of choice and may your robot overlords be with you.